diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,28097 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.351078331609941,
+  "eval_steps": 500,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 8.776958290248525e-05,
+      "grad_norm": 0.40625,
+      "learning_rate": 3e-06,
+      "loss": 1.7689,
+      "step": 1
+    },
+    {
+      "epoch": 0.0001755391658049705,
+      "grad_norm": 0.38671875,
+      "learning_rate": 6e-06,
+      "loss": 1.7451,
+      "step": 2
+    },
+    {
+      "epoch": 0.00026330874870745575,
+      "grad_norm": 0.388671875,
+      "learning_rate": 9e-06,
+      "loss": 1.7571,
+      "step": 3
+    },
+    {
+      "epoch": 0.000351078331609941,
+      "grad_norm": 0.39453125,
+      "learning_rate": 1.2e-05,
+      "loss": 1.7319,
+      "step": 4
+    },
+    {
+      "epoch": 0.00043884791451242626,
+      "grad_norm": 0.36328125,
+      "learning_rate": 1.5e-05,
+      "loss": 1.7674,
+      "step": 5
+    },
+    {
+      "epoch": 0.0005266174974149115,
+      "grad_norm": 0.404296875,
+      "learning_rate": 1.8e-05,
+      "loss": 1.7341,
+      "step": 6
+    },
+    {
+      "epoch": 0.0006143870803173967,
+      "grad_norm": 0.357421875,
+      "learning_rate": 2.1000000000000002e-05,
+      "loss": 1.7525,
+      "step": 7
+    },
+    {
+      "epoch": 0.000702156663219882,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.4e-05,
+      "loss": 1.7367,
+      "step": 8
+    },
+    {
+      "epoch": 0.0007899262461223672,
+      "grad_norm": 0.3515625,
+      "learning_rate": 2.7e-05,
+      "loss": 1.7394,
+      "step": 9
+    },
+    {
+      "epoch": 0.0008776958290248525,
+      "grad_norm": 0.328125,
+      "learning_rate": 3e-05,
+      "loss": 1.7194,
+      "step": 10
+    },
+    {
+      "epoch": 0.0009654654119273377,
+      "grad_norm": 0.31640625,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 1.7542,
+      "step": 11
+    },
+    {
+      "epoch": 0.001053234994829823,
+      "grad_norm": 0.310546875,
+      "learning_rate": 3.6e-05,
+      "loss": 1.7498,
+      "step": 12
+    },
+    {
+      "epoch": 0.0011410045777323083,
+      "grad_norm": 0.27734375,
+      "learning_rate": 3.9e-05,
+      "loss": 1.7251,
+      "step": 13
+    },
+    {
+      "epoch": 0.0012287741606347934,
+      "grad_norm": 0.271484375,
+      "learning_rate": 4.2000000000000004e-05,
+      "loss": 1.7456,
+      "step": 14
+    },
+    {
+      "epoch": 0.0013165437435372787,
+      "grad_norm": 0.251953125,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 1.7261,
+      "step": 15
+    },
+    {
+      "epoch": 0.001404313326439764,
+      "grad_norm": 0.248046875,
+      "learning_rate": 4.8e-05,
+      "loss": 1.7426,
+      "step": 16
+    },
+    {
+      "epoch": 0.0014920829093422493,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 1.7373,
+      "step": 17
+    },
+    {
+      "epoch": 0.0015798524922447344,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 5.4e-05,
+      "loss": 1.7427,
+      "step": 18
+    },
+    {
+      "epoch": 0.0016676220751472197,
+      "grad_norm": 0.244140625,
+      "learning_rate": 5.7e-05,
+      "loss": 1.7142,
+      "step": 19
+    },
+    {
+      "epoch": 0.001755391658049705,
+      "grad_norm": 0.224609375,
+      "learning_rate": 6e-05,
+      "loss": 1.7586,
+      "step": 20
+    },
+    {
+      "epoch": 0.0018431612409521903,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 6.3e-05,
+      "loss": 1.7348,
+      "step": 21
+    },
+    {
+      "epoch": 0.0019309308238546754,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 1.7164,
+      "step": 22
+    },
+    {
+      "epoch": 0.002018700406757161,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 6.9e-05,
+      "loss": 1.7185,
+      "step": 23
+    },
+    {
+      "epoch": 0.002106469989659646,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 7.2e-05,
+      "loss": 1.7122,
+      "step": 24
+    },
+    {
+      "epoch": 0.002194239572562131,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.7534,
+      "step": 25
+    },
+    {
+      "epoch": 0.0022820091554646166,
+      "grad_norm": 0.212890625,
+      "learning_rate": 7.8e-05,
+      "loss": 1.7097,
+      "step": 26
+    },
+    {
+      "epoch": 0.0023697787383671017,
+      "grad_norm": 0.197265625,
+      "learning_rate": 8.1e-05,
+      "loss": 1.7264,
+      "step": 27
+    },
+    {
+      "epoch": 0.002457548321269587,
+      "grad_norm": 0.1953125,
+      "learning_rate": 8.400000000000001e-05,
+      "loss": 1.7221,
+      "step": 28
+    },
+    {
+      "epoch": 0.0025453179041720723,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 8.7e-05,
+      "loss": 1.7413,
+      "step": 29
+    },
+    {
+      "epoch": 0.0026330874870745574,
+      "grad_norm": 0.19140625,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 1.7129,
+      "step": 30
+    },
+    {
+      "epoch": 0.002720857069977043,
+      "grad_norm": 0.177734375,
+      "learning_rate": 9.3e-05,
+      "loss": 1.6926,
+      "step": 31
+    },
+    {
+      "epoch": 0.002808626652879528,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 9.6e-05,
+      "loss": 1.7416,
+      "step": 32
+    },
+    {
+      "epoch": 0.002896396235782013,
+      "grad_norm": 0.1796875,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 1.7328,
+      "step": 33
+    },
+    {
+      "epoch": 0.0029841658186844987,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.00010200000000000001,
+      "loss": 1.7141,
+      "step": 34
+    },
+    {
+      "epoch": 0.0030719354015869837,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00010500000000000002,
+      "loss": 1.7105,
+      "step": 35
+    },
+    {
+      "epoch": 0.003159704984489469,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.000108,
+      "loss": 1.7226,
+      "step": 36
+    },
+    {
+      "epoch": 0.0032474745673919544,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.000111,
+      "loss": 1.7251,
+      "step": 37
+    },
+    {
+      "epoch": 0.0033352441502944394,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.000114,
+      "loss": 1.7446,
+      "step": 38
+    },
+    {
+      "epoch": 0.003423013733196925,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.000117,
+      "loss": 1.741,
+      "step": 39
+    },
+    {
+      "epoch": 0.00351078331609941,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00012,
+      "loss": 1.6584,
+      "step": 40
+    },
+    {
+      "epoch": 0.003598552899001895,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.000123,
+      "loss": 1.6891,
+      "step": 41
+    },
+    {
+      "epoch": 0.0036863224819043807,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.000126,
+      "loss": 1.7028,
+      "step": 42
+    },
+    {
+      "epoch": 0.0037740920648068658,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.000129,
+      "loss": 1.6706,
+      "step": 43
+    },
+    {
+      "epoch": 0.003861861647709351,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 1.7151,
+      "step": 44
+    },
+    {
+      "epoch": 0.003949631230611836,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.000135,
+      "loss": 1.7094,
+      "step": 45
+    },
+    {
+      "epoch": 0.004037400813514322,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.000138,
+      "loss": 1.7015,
+      "step": 46
+    },
+    {
+      "epoch": 0.0041251703964168066,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.000141,
+      "loss": 1.6753,
+      "step": 47
+    },
+    {
+      "epoch": 0.004212939979319292,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.000144,
+      "loss": 1.6983,
+      "step": 48
+    },
+    {
+      "epoch": 0.004300709562221778,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.000147,
+      "loss": 1.6713,
+      "step": 49
+    },
+    {
+      "epoch": 0.004388479145124262,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.7099,
+      "step": 50
+    },
+    {
+      "epoch": 0.004476248728026748,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.000153,
+      "loss": 1.698,
+      "step": 51
+    },
+    {
+      "epoch": 0.004564018310929233,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.000156,
+      "loss": 1.6855,
+      "step": 52
+    },
+    {
+      "epoch": 0.004651787893831718,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.000159,
+      "loss": 1.7034,
+      "step": 53
+    },
+    {
+      "epoch": 0.0047395574767342035,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.000162,
+      "loss": 1.7015,
+      "step": 54
+    },
+    {
+      "epoch": 0.004827327059636689,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.000165,
+      "loss": 1.6715,
+      "step": 55
+    },
+    {
+      "epoch": 0.004915096642539174,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00016800000000000002,
+      "loss": 1.6808,
+      "step": 56
+    },
+    {
+      "epoch": 0.005002866225441659,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.000171,
+      "loss": 1.6848,
+      "step": 57
+    },
+    {
+      "epoch": 0.005090635808344145,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.000174,
+      "loss": 1.6683,
+      "step": 58
+    },
+    {
+      "epoch": 0.005178405391246629,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.000177,
+      "loss": 1.6896,
+      "step": 59
+    },
+    {
+      "epoch": 0.005266174974149115,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 1.7075,
+      "step": 60
+    },
+    {
+      "epoch": 0.0053539445570516,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.000183,
+      "loss": 1.6605,
+      "step": 61
+    },
+    {
+      "epoch": 0.005441714139954086,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.000186,
+      "loss": 1.6536,
+      "step": 62
+    },
+    {
+      "epoch": 0.005529483722856571,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.000189,
+      "loss": 1.6499,
+      "step": 63
+    },
+    {
+      "epoch": 0.005617253305759056,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.000192,
+      "loss": 1.6682,
+      "step": 64
+    },
+    {
+      "epoch": 0.005705022888661542,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00019500000000000002,
+      "loss": 1.6727,
+      "step": 65
+    },
+    {
+      "epoch": 0.005792792471564026,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 1.6554,
+      "step": 66
+    },
+    {
+      "epoch": 0.005880562054466512,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.000201,
+      "loss": 1.651,
+      "step": 67
+    },
+    {
+      "epoch": 0.005968331637368997,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00020400000000000003,
+      "loss": 1.6457,
+      "step": 68
+    },
+    {
+      "epoch": 0.006056101220271482,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.00020700000000000002,
+      "loss": 1.6536,
+      "step": 69
+    },
+    {
+      "epoch": 0.0061438708031739675,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.00021000000000000004,
+      "loss": 1.6843,
+      "step": 70
+    },
+    {
+      "epoch": 0.006231640386076453,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 1.6222,
+      "step": 71
+    },
+    {
+      "epoch": 0.006319409968978938,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.000216,
+      "loss": 1.6791,
+      "step": 72
+    },
+    {
+      "epoch": 0.006407179551881423,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 1.6917,
+      "step": 73
+    },
+    {
+      "epoch": 0.006494949134783909,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.000222,
+      "loss": 1.6895,
+      "step": 74
+    },
+    {
+      "epoch": 0.006582718717686393,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.000225,
+      "loss": 1.7024,
+      "step": 75
+    },
+    {
+      "epoch": 0.006670488300588879,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.000228,
+      "loss": 1.6649,
+      "step": 76
+    },
+    {
+      "epoch": 0.006758257883491364,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.000231,
+      "loss": 1.6537,
+      "step": 77
+    },
+    {
+      "epoch": 0.00684602746639385,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.000234,
+      "loss": 1.6345,
+      "step": 78
+    },
+    {
+      "epoch": 0.006933797049296335,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.00023700000000000001,
+      "loss": 1.6799,
+      "step": 79
+    },
+    {
+      "epoch": 0.00702156663219882,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00024,
+      "loss": 1.6535,
+      "step": 80
+    },
+    {
+      "epoch": 0.007109336215101306,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00024300000000000002,
+      "loss": 1.6421,
+      "step": 81
+    },
+    {
+      "epoch": 0.00719710579800379,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.000246,
+      "loss": 1.6371,
+      "step": 82
+    },
+    {
+      "epoch": 0.007284875380906276,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00024900000000000004,
+      "loss": 1.6963,
+      "step": 83
+    },
+    {
+      "epoch": 0.007372644963808761,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.000252,
+      "loss": 1.6568,
+      "step": 84
+    },
+    {
+      "epoch": 0.007460414546711246,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.000255,
+      "loss": 1.6686,
+      "step": 85
+    },
+    {
+      "epoch": 0.0075481841296137315,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.000258,
+      "loss": 1.6236,
+      "step": 86
+    },
+    {
+      "epoch": 0.007635953712516217,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.000261,
+      "loss": 1.6198,
+      "step": 87
+    },
+    {
+      "epoch": 0.007723723295418702,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 1.6159,
+      "step": 88
+    },
+    {
+      "epoch": 0.007811492878321187,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.000267,
+      "loss": 1.6369,
+      "step": 89
+    },
+    {
+      "epoch": 0.007899262461223673,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00027,
+      "loss": 1.5979,
+      "step": 90
+    },
+    {
+      "epoch": 0.007987032044126157,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.000273,
+      "loss": 1.6274,
+      "step": 91
+    },
+    {
+      "epoch": 0.008074801627028644,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.000276,
+      "loss": 1.6396,
+      "step": 92
+    },
+    {
+      "epoch": 0.008162571209931128,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.000279,
+      "loss": 1.6311,
+      "step": 93
+    },
+    {
+      "epoch": 0.008250340792833613,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.000282,
+      "loss": 1.6497,
+      "step": 94
+    },
+    {
+      "epoch": 0.0083381103757361,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.000285,
+      "loss": 1.6387,
+      "step": 95
+    },
+    {
+      "epoch": 0.008425879958638584,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.000288,
+      "loss": 1.6348,
+      "step": 96
+    },
+    {
+      "epoch": 0.008513649541541069,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.000291,
+      "loss": 1.6238,
+      "step": 97
+    },
+    {
+      "epoch": 0.008601419124443555,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.000294,
+      "loss": 1.6424,
+      "step": 98
+    },
+    {
+      "epoch": 0.00868918870734604,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.000297,
+      "loss": 1.6115,
+      "step": 99
+    },
+    {
+      "epoch": 0.008776958290248525,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.00030000000000000003,
+      "loss": 1.6163,
+      "step": 100
+    },
+    {
+      "epoch": 0.008864727873151011,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.00030300000000000005,
+      "loss": 1.6289,
+      "step": 101
+    },
+    {
+      "epoch": 0.008952497456053496,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.000306,
+      "loss": 1.6104,
+      "step": 102
+    },
+    {
+      "epoch": 0.00904026703895598,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.000309,
+      "loss": 1.5959,
+      "step": 103
+    },
+    {
+      "epoch": 0.009128036621858467,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.000312,
+      "loss": 1.6469,
+      "step": 104
+    },
+    {
+      "epoch": 0.009215806204760951,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.000315,
+      "loss": 1.6033,
+      "step": 105
+    },
+    {
+      "epoch": 0.009303575787663436,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.000318,
+      "loss": 1.628,
+      "step": 106
+    },
+    {
+      "epoch": 0.009391345370565922,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.000321,
+      "loss": 1.6601,
+      "step": 107
+    },
+    {
+      "epoch": 0.009479114953468407,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.000324,
+      "loss": 1.6207,
+      "step": 108
+    },
+    {
+      "epoch": 0.009566884536370892,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.000327,
+      "loss": 1.5845,
+      "step": 109
+    },
+    {
+      "epoch": 0.009654654119273378,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00033,
+      "loss": 1.6281,
+      "step": 110
+    },
+    {
+      "epoch": 0.009742423702175863,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.000333,
+      "loss": 1.6241,
+      "step": 111
+    },
+    {
+      "epoch": 0.009830193285078347,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.00033600000000000004,
+      "loss": 1.6173,
+      "step": 112
+    },
+    {
+      "epoch": 0.009917962867980834,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.000339,
+      "loss": 1.6026,
+      "step": 113
+    },
+    {
+      "epoch": 0.010005732450883318,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.000342,
+      "loss": 1.6295,
+      "step": 114
+    },
+    {
+      "epoch": 0.010093502033785803,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00034500000000000004,
+      "loss": 1.6142,
+      "step": 115
+    },
+    {
+      "epoch": 0.01018127161668829,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.000348,
+      "loss": 1.6091,
+      "step": 116
+    },
+    {
+      "epoch": 0.010269041199590774,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.000351,
+      "loss": 1.5936,
+      "step": 117
+    },
+    {
+      "epoch": 0.010356810782493259,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.000354,
+      "loss": 1.6154,
+      "step": 118
+    },
+    {
+      "epoch": 0.010444580365395745,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.000357,
+      "loss": 1.5782,
+      "step": 119
+    },
+    {
+      "epoch": 0.01053234994829823,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 1.6065,
+      "step": 120
+    },
+    {
+      "epoch": 0.010620119531200716,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.000363,
+      "loss": 1.5985,
+      "step": 121
+    },
+    {
+      "epoch": 0.0107078891141032,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.000366,
+      "loss": 1.5969,
+      "step": 122
+    },
+    {
+      "epoch": 0.010795658697005685,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.000369,
+      "loss": 1.6173,
+      "step": 123
+    },
+    {
+      "epoch": 0.010883428279908172,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.000372,
+      "loss": 1.5828,
+      "step": 124
+    },
+    {
+      "epoch": 0.010971197862810656,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.000375,
+      "loss": 1.5594,
+      "step": 125
+    },
+    {
+      "epoch": 0.011058967445713141,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.000378,
+      "loss": 1.6131,
+      "step": 126
+    },
+    {
+      "epoch": 0.011146737028615628,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.000381,
+      "loss": 1.6203,
+      "step": 127
+    },
+    {
+      "epoch": 0.011234506611518112,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.000384,
+      "loss": 1.598,
+      "step": 128
+    },
+    {
+      "epoch": 0.011322276194420597,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00038700000000000003,
+      "loss": 1.5817,
+      "step": 129
+    },
+    {
+      "epoch": 0.011410045777323083,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.00039000000000000005,
+      "loss": 1.5968,
+      "step": 130
+    },
+    {
+      "epoch": 0.011497815360225568,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.000393,
+      "loss": 1.6066,
+      "step": 131
+    },
+    {
+      "epoch": 0.011585584943128053,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 1.6208,
+      "step": 132
+    },
+    {
+      "epoch": 0.011673354526030539,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.00039900000000000005,
+      "loss": 1.5973,
+      "step": 133
+    },
+    {
+      "epoch": 0.011761124108933024,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.000402,
+      "loss": 1.588,
+      "step": 134
+    },
+    {
+      "epoch": 0.011848893691835508,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00040500000000000003,
+      "loss": 1.6202,
+      "step": 135
+    },
+    {
+      "epoch": 0.011936663274737995,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00040800000000000005,
+      "loss": 1.5611,
+      "step": 136
+    },
+    {
+      "epoch": 0.01202443285764048,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.000411,
+      "loss": 1.5582,
+      "step": 137
+    },
+    {
+      "epoch": 0.012112202440542964,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00041400000000000003,
+      "loss": 1.5798,
+      "step": 138
+    },
+    {
+      "epoch": 0.01219997202344545,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.00041700000000000005,
+      "loss": 1.5895,
+      "step": 139
+    },
+    {
+      "epoch": 0.012287741606347935,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00042000000000000007,
+      "loss": 1.605,
+      "step": 140
+    },
+    {
+      "epoch": 0.01237551118925042,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.000423,
+      "loss": 1.5972,
+      "step": 141
+    },
+    {
+      "epoch": 0.012463280772152906,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.00042599999999999995,
+      "loss": 1.5651,
+      "step": 142
+    },
+    {
+      "epoch": 0.01255105035505539,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.00042899999999999997,
+      "loss": 1.5516,
+      "step": 143
+    },
+    {
+      "epoch": 0.012638819937957875,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.000432,
+      "loss": 1.5887,
+      "step": 144
+    },
+    {
+      "epoch": 0.012726589520860362,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.000435,
+      "loss": 1.5829,
+      "step": 145
+    },
+    {
+      "epoch": 0.012814359103762846,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00043799999999999997,
+      "loss": 1.5895,
+      "step": 146
+    },
+    {
+      "epoch": 0.012902128686665331,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.000441,
+      "loss": 1.6111,
+      "step": 147
+    },
+    {
+      "epoch": 0.012989898269567817,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.000444,
+      "loss": 1.5881,
+      "step": 148
+    },
+    {
+      "epoch": 0.013077667852470302,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.00044699999999999997,
+      "loss": 1.5744,
+      "step": 149
+    },
+    {
+      "epoch": 0.013165437435372787,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00045,
+      "loss": 1.5884,
+      "step": 150
+    },
+    {
+      "epoch": 0.013253207018275273,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.000453,
+      "loss": 1.5649,
+      "step": 151
+    },
+    {
+      "epoch": 0.013340976601177758,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.000456,
+      "loss": 1.5982,
+      "step": 152
+    },
+    {
+      "epoch": 0.013428746184080244,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.000459,
+      "loss": 1.5878,
+      "step": 153
+    },
+    {
+      "epoch": 0.013516515766982729,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.000462,
+      "loss": 1.5519,
+      "step": 154
+    },
+    {
+      "epoch": 0.013604285349885213,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.000465,
+      "loss": 1.5693,
+      "step": 155
+    },
+    {
+      "epoch": 0.0136920549327877,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.000468,
+      "loss": 1.5546,
+      "step": 156
+    },
+    {
+      "epoch": 0.013779824515690185,
+      "grad_norm": 0.375,
+      "learning_rate": 0.000471,
+      "loss": 1.5526,
+      "step": 157
+    },
+    {
+      "epoch": 0.01386759409859267,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00047400000000000003,
+      "loss": 1.5593,
+      "step": 158
+    },
+    {
+      "epoch": 0.013955363681495156,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.000477,
+      "loss": 1.5608,
+      "step": 159
+    },
+    {
+      "epoch": 0.01404313326439764,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00048,
+      "loss": 1.5527,
+      "step": 160
+    },
+    {
+      "epoch": 0.014130902847300125,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00048300000000000003,
+      "loss": 1.5818,
+      "step": 161
+    },
+    {
+      "epoch": 0.014218672430202611,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00048600000000000005,
+      "loss": 1.5687,
+      "step": 162
+    },
+    {
+      "epoch": 0.014306442013105096,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.0004890000000000001,
+      "loss": 1.5944,
+      "step": 163
+    },
+    {
+      "epoch": 0.01439421159600758,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.000492,
+      "loss": 1.6024,
+      "step": 164
+    },
+    {
+      "epoch": 0.014481981178910067,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.000495,
+      "loss": 1.561,
+      "step": 165
+    },
+    {
+      "epoch": 0.014569750761812552,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0004980000000000001,
+      "loss": 1.576,
+      "step": 166
+    },
+    {
+      "epoch": 0.014657520344715036,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.000501,
+      "loss": 1.536,
+      "step": 167
+    },
+    {
+      "epoch": 0.014745289927617523,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.000504,
+      "loss": 1.5691,
+      "step": 168
+    },
+    {
+      "epoch": 0.014833059510520007,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.0005070000000000001,
+      "loss": 1.5502,
+      "step": 169
+    },
+    {
+      "epoch": 0.014920829093422492,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00051,
+      "loss": 1.5524,
+      "step": 170
+    },
+    {
+      "epoch": 0.015008598676324978,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.000513,
+      "loss": 1.582,
+      "step": 171
+    },
+    {
+      "epoch": 0.015096368259227463,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.000516,
+      "loss": 1.5702,
+      "step": 172
+    },
+    {
+      "epoch": 0.015184137842129948,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0005189999999999999,
+      "loss": 1.578,
+      "step": 173
+    },
+    {
+      "epoch": 0.015271907425032434,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.000522,
+      "loss": 1.534,
+      "step": 174
+    },
+    {
+      "epoch": 0.015359677007934919,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.000525,
+      "loss": 1.5624,
+      "step": 175
+    },
+    {
+      "epoch": 0.015447446590837403,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.0005279999999999999,
+      "loss": 1.5523,
+      "step": 176
+    },
+    {
+      "epoch": 0.01553521617373989,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.000531,
+      "loss": 1.5955,
+      "step": 177
+    },
+    {
+      "epoch": 0.015622985756642374,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.000534,
+      "loss": 1.5625,
+      "step": 178
+    },
+    {
+      "epoch": 0.01571075533954486,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.000537,
+      "loss": 1.553,
+      "step": 179
+    },
+    {
+      "epoch": 0.015798524922447345,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00054,
+      "loss": 1.5868,
+      "step": 180
+    },
+    {
+      "epoch": 0.01588629450534983,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.000543,
+      "loss": 1.5586,
+      "step": 181
+    },
+    {
+      "epoch": 0.015974064088252315,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.000546,
+      "loss": 1.5364,
+      "step": 182
+    },
+    {
+      "epoch": 0.0160618336711548,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.000549,
+      "loss": 1.5587,
+      "step": 183
+    },
+    {
+      "epoch": 0.016149603254057288,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.000552,
+      "loss": 1.5493,
+      "step": 184
+    },
+    {
+      "epoch": 0.016237372836959772,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.000555,
+      "loss": 1.5635,
+      "step": 185
+    },
+    {
+      "epoch": 0.016325142419862257,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.000558,
+      "loss": 1.5035,
+      "step": 186
+    },
+    {
+      "epoch": 0.01641291200276474,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.000561,
+      "loss": 1.5546,
+      "step": 187
+    },
+    {
+      "epoch": 0.016500681585667226,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.000564,
+      "loss": 1.5222,
+      "step": 188
+    },
+    {
+      "epoch": 0.01658845116856971,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.000567,
+      "loss": 1.5477,
+      "step": 189
+    },
+    {
+      "epoch": 0.0166762207514722,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00057,
+      "loss": 1.5579,
+      "step": 190
+    },
+    {
+      "epoch": 0.016763990334374684,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.000573,
+      "loss": 1.5225,
+      "step": 191
+    },
+    {
+      "epoch": 0.01685175991727717,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.000576,
+      "loss": 1.5301,
+      "step": 192
+    },
+    {
+      "epoch": 0.016939529500179653,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.000579,
+      "loss": 1.5545,
+      "step": 193
+    },
+    {
+      "epoch": 0.017027299083082138,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.000582,
+      "loss": 1.5481,
+      "step": 194
+    },
+    {
+      "epoch": 0.017115068665984622,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.000585,
+      "loss": 1.5556,
+      "step": 195
+    },
+    {
+      "epoch": 0.01720283824888711,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.000588,
+      "loss": 1.5607,
+      "step": 196
+    },
+    {
+      "epoch": 0.017290607831789595,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.000591,
+      "loss": 1.5166,
+      "step": 197
+    },
+    {
+      "epoch": 0.01737837741469208,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.000594,
+      "loss": 1.5709,
+      "step": 198
+    },
+    {
+      "epoch": 0.017466146997594564,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0005970000000000001,
+      "loss": 1.5871,
+      "step": 199
+    },
+    {
+      "epoch": 0.01755391658049705,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0006000000000000001,
+      "loss": 1.5793,
+      "step": 200
+    },
+    {
+      "epoch": 0.017641686163399534,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.000603,
+      "loss": 1.5614,
+      "step": 201
+    },
+    {
+      "epoch": 0.017729455746302022,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0006060000000000001,
+      "loss": 1.566,
+      "step": 202
+    },
+    {
+      "epoch": 0.017817225329204506,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0006090000000000001,
+      "loss": 1.5251,
+      "step": 203
+    },
+    {
+      "epoch": 0.01790499491210699,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.000612,
+      "loss": 1.5759,
+      "step": 204
+    },
+    {
+      "epoch": 0.017992764495009476,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.000615,
+      "loss": 1.568,
+      "step": 205
+    },
+    {
+      "epoch": 0.01808053407791196,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.000618,
+      "loss": 1.5376,
+      "step": 206
+    },
+    {
+      "epoch": 0.01816830366081445,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.000621,
+      "loss": 1.529,
+      "step": 207
+    },
+    {
+      "epoch": 0.018256073243716933,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.000624,
+      "loss": 1.5411,
+      "step": 208
+    },
+    {
+      "epoch": 0.018343842826619418,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.000627,
+      "loss": 1.5323,
+      "step": 209
+    },
+    {
+      "epoch": 0.018431612409521902,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00063,
+      "loss": 1.5671,
+      "step": 210
+    },
+    {
+      "epoch": 0.018519381992424387,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.000633,
+      "loss": 1.528,
+      "step": 211
+    },
+    {
+      "epoch": 0.018607151575326872,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.000636,
+      "loss": 1.5338,
+      "step": 212
+    },
+    {
+      "epoch": 0.01869492115822936,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.000639,
+      "loss": 1.4899,
+      "step": 213
+    },
+    {
+      "epoch": 0.018782690741131845,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.000642,
+      "loss": 1.5698,
+      "step": 214
+    },
+    {
+      "epoch": 0.01887046032403433,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.000645,
+      "loss": 1.5581,
+      "step": 215
+    },
+    {
+      "epoch": 0.018958229906936814,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.000648,
+      "loss": 1.5541,
+      "step": 216
+    },
+    {
+      "epoch": 0.0190459994898393,
+      "grad_norm": 0.25,
+      "learning_rate": 0.000651,
+      "loss": 1.5394,
+      "step": 217
+    },
+    {
+      "epoch": 0.019133769072741783,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.000654,
+      "loss": 1.5317,
+      "step": 218
+    },
+    {
+      "epoch": 0.01922153865564427,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.000657,
+      "loss": 1.4966,
+      "step": 219
+    },
+    {
+      "epoch": 0.019309308238546756,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00066,
+      "loss": 1.54,
+      "step": 220
+    },
+    {
+      "epoch": 0.01939707782144924,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0006630000000000001,
+      "loss": 1.5467,
+      "step": 221
+    },
+    {
+      "epoch": 0.019484847404351725,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.000666,
+      "loss": 1.5047,
+      "step": 222
+    },
+    {
+      "epoch": 0.01957261698725421,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.000669,
+      "loss": 1.5274,
+      "step": 223
+    },
+    {
+      "epoch": 0.019660386570156695,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0006720000000000001,
+      "loss": 1.5205,
+      "step": 224
+    },
+    {
+      "epoch": 0.019748156153059183,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.000675,
+      "loss": 1.528,
+      "step": 225
+    },
+    {
+      "epoch": 0.019835925735961667,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.000678,
+      "loss": 1.5186,
+      "step": 226
+    },
+    {
+      "epoch": 0.019923695318864152,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0006810000000000001,
+      "loss": 1.5292,
+      "step": 227
+    },
+    {
+      "epoch": 0.020011464901766637,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.000684,
+      "loss": 1.5265,
+      "step": 228
+    },
+    {
+      "epoch": 0.02009923448466912,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.000687,
+      "loss": 1.4965,
+      "step": 229
+    },
+    {
+      "epoch": 0.020187004067571606,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0006900000000000001,
+      "loss": 1.4951,
+      "step": 230
+    },
+    {
+      "epoch": 0.020274773650474094,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.000693,
+      "loss": 1.5387,
+      "step": 231
+    },
+    {
+      "epoch": 0.02036254323337658,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.000696,
+      "loss": 1.5147,
+      "step": 232
+    },
+    {
+      "epoch": 0.020450312816279063,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0006990000000000001,
+      "loss": 1.4739,
+      "step": 233
+    },
+    {
+      "epoch": 0.020538082399181548,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.000702,
+      "loss": 1.5632,
+      "step": 234
+    },
+    {
+      "epoch": 0.020625851982084033,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.000705,
+      "loss": 1.5118,
+      "step": 235
+    },
+    {
+      "epoch": 0.020713621564986517,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.000708,
+      "loss": 1.5593,
+      "step": 236
+    },
+    {
+      "epoch": 0.020801391147889006,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0007109999999999999,
+      "loss": 1.5134,
+      "step": 237
+    },
+    {
+      "epoch": 0.02088916073079149,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.000714,
+      "loss": 1.4698,
+      "step": 238
+    },
+    {
+      "epoch": 0.020976930313693975,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.000717,
+      "loss": 1.5551,
+      "step": 239
+    },
+    {
+      "epoch": 0.02106469989659646,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0007199999999999999,
+      "loss": 1.5045,
+      "step": 240
+    },
+    {
+      "epoch": 0.021152469479498944,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.000723,
+      "loss": 1.5291,
+      "step": 241
+    },
+    {
+      "epoch": 0.021240239062401432,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.000726,
+      "loss": 1.5537,
+      "step": 242
+    },
+    {
+      "epoch": 0.021328008645303917,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.000729,
+      "loss": 1.5157,
+      "step": 243
+    },
+    {
+      "epoch": 0.0214157782282064,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.000732,
+      "loss": 1.4763,
+      "step": 244
+    },
+    {
+      "epoch": 0.021503547811108886,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.000735,
+      "loss": 1.5265,
+      "step": 245
+    },
+    {
+      "epoch": 0.02159131739401137,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.000738,
+      "loss": 1.5019,
+      "step": 246
+    },
+    {
+      "epoch": 0.021679086976913856,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.000741,
+      "loss": 1.5275,
+      "step": 247
+    },
+    {
+      "epoch": 0.021766856559816344,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.000744,
+      "loss": 1.4965,
+      "step": 248
+    },
+    {
+      "epoch": 0.02185462614271883,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.000747,
+      "loss": 1.5335,
+      "step": 249
+    },
+    {
+      "epoch": 0.021942395725621313,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00075,
+      "loss": 1.4928,
+      "step": 250
+    },
+    {
+      "epoch": 0.022030165308523798,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.000753,
+      "loss": 1.5167,
+      "step": 251
+    },
+    {
+      "epoch": 0.022117934891426282,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.000756,
+      "loss": 1.4861,
+      "step": 252
+    },
+    {
+      "epoch": 0.022205704474328767,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.000759,
+      "loss": 1.4847,
+      "step": 253
+    },
+    {
+      "epoch": 0.022293474057231255,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.000762,
+      "loss": 1.5386,
+      "step": 254
+    },
+    {
+      "epoch": 0.02238124364013374,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0007650000000000001,
+      "loss": 1.5287,
+      "step": 255
+    },
+    {
+      "epoch": 0.022469013223036224,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.000768,
+      "loss": 1.5219,
+      "step": 256
+    },
+    {
+      "epoch": 0.02255678280593871,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.000771,
+      "loss": 1.5138,
+      "step": 257
+    },
+    {
+      "epoch": 0.022644552388841194,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0007740000000000001,
+      "loss": 1.5172,
+      "step": 258
+    },
+    {
+      "epoch": 0.02273232197174368,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.000777,
+      "loss": 1.4874,
+      "step": 259
+    },
+    {
+      "epoch": 0.022820091554646166,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0007800000000000001,
+      "loss": 1.5269,
+      "step": 260
+    },
+    {
+      "epoch": 0.02290786113754865,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.0007830000000000001,
+      "loss": 1.5049,
+      "step": 261
+    },
+    {
+      "epoch": 0.022995630720451136,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.000786,
+      "loss": 1.5297,
+      "step": 262
+    },
+    {
+      "epoch": 0.02308340030335362,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0007890000000000001,
+      "loss": 1.516,
+      "step": 263
+    },
+    {
+      "epoch": 0.023171169886256105,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.0007920000000000001,
+      "loss": 1.5185,
+      "step": 264
+    },
+    {
+      "epoch": 0.02325893946915859,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.000795,
+      "loss": 1.4899,
+      "step": 265
+    },
+    {
+      "epoch": 0.023346709052061078,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.0007980000000000001,
+      "loss": 1.5057,
+      "step": 266
+    },
+    {
+      "epoch": 0.023434478634963563,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0008010000000000001,
+      "loss": 1.518,
+      "step": 267
+    },
+    {
+      "epoch": 0.023522248217866047,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.000804,
+      "loss": 1.4779,
+      "step": 268
+    },
+    {
+      "epoch": 0.023610017800768532,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0008070000000000001,
+      "loss": 1.483,
+      "step": 269
+    },
+    {
+      "epoch": 0.023697787383671017,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0008100000000000001,
+      "loss": 1.5232,
+      "step": 270
+    },
+    {
+      "epoch": 0.023785556966573505,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.000813,
+      "loss": 1.5,
+      "step": 271
+    },
+    {
+      "epoch": 0.02387332654947599,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0008160000000000001,
+      "loss": 1.4608,
+      "step": 272
+    },
+    {
+      "epoch": 0.023961096132378474,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0008190000000000001,
+      "loss": 1.4699,
+      "step": 273
+    },
+    {
+      "epoch": 0.02404886571528096,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.000822,
+      "loss": 1.493,
+      "step": 274
+    },
+    {
+      "epoch": 0.024136635298183443,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0008250000000000001,
+      "loss": 1.5142,
+      "step": 275
+    },
+    {
+      "epoch": 0.024224404881085928,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0008280000000000001,
+      "loss": 1.5442,
+      "step": 276
+    },
+    {
+      "epoch": 0.024312174463988416,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0008310000000000001,
+      "loss": 1.5254,
+      "step": 277
+    },
+    {
+      "epoch": 0.0243999440468909,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0008340000000000001,
+      "loss": 1.4906,
+      "step": 278
+    },
+    {
+      "epoch": 0.024487713629793385,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0008370000000000001,
+      "loss": 1.4933,
+      "step": 279
+    },
+    {
+      "epoch": 0.02457548321269587,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0008400000000000001,
+      "loss": 1.5112,
+      "step": 280
+    },
+    {
+      "epoch": 0.024663252795598355,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0008430000000000001,
+      "loss": 1.4743,
+      "step": 281
+    },
+    {
+      "epoch": 0.02475102237850084,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.000846,
+      "loss": 1.5153,
+      "step": 282
+    },
+    {
+      "epoch": 0.024838791961403327,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0008489999999999999,
+      "loss": 1.4923,
+      "step": 283
+    },
+    {
+      "epoch": 0.024926561544305812,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0008519999999999999,
+      "loss": 1.4636,
+      "step": 284
+    },
+    {
+      "epoch": 0.025014331127208297,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.000855,
+      "loss": 1.5013,
+      "step": 285
+    },
+    {
+      "epoch": 0.02510210071011078,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0008579999999999999,
+      "loss": 1.4789,
+      "step": 286
+    },
+    {
+      "epoch": 0.025189870293013266,
+      "grad_norm": 0.25,
+      "learning_rate": 0.000861,
+      "loss": 1.4871,
+      "step": 287
+    },
+    {
+      "epoch": 0.02527763987591575,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.000864,
+      "loss": 1.4758,
+      "step": 288
+    },
+    {
+      "epoch": 0.02536540945881824,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0008669999999999999,
+      "loss": 1.4956,
+      "step": 289
+    },
+    {
+      "epoch": 0.025453179041720723,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00087,
+      "loss": 1.5027,
+      "step": 290
+    },
+    {
+      "epoch": 0.025540948624623208,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.000873,
+      "loss": 1.503,
+      "step": 291
+    },
+    {
+      "epoch": 0.025628718207525693,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0008759999999999999,
+      "loss": 1.4981,
+      "step": 292
+    },
+    {
+      "epoch": 0.025716487790428177,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.000879,
+      "loss": 1.5098,
+      "step": 293
+    },
+    {
+      "epoch": 0.025804257373330662,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.000882,
+      "loss": 1.4731,
+      "step": 294
+    },
+    {
+      "epoch": 0.02589202695623315,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0008849999999999999,
+      "loss": 1.4688,
+      "step": 295
+    },
+    {
+      "epoch": 0.025979796539135635,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.000888,
+      "loss": 1.4874,
+      "step": 296
+    },
+    {
+      "epoch": 0.02606756612203812,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.000891,
+      "loss": 1.4911,
+      "step": 297
+    },
+    {
+      "epoch": 0.026155335704940604,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0008939999999999999,
+      "loss": 1.51,
+      "step": 298
+    },
+    {
+      "epoch": 0.02624310528784309,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.000897,
+      "loss": 1.509,
+      "step": 299
+    },
+    {
+      "epoch": 0.026330874870745574,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0009,
+      "loss": 1.5112,
+      "step": 300
+    },
+    {
+      "epoch": 0.02641864445364806,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.0009029999999999999,
+      "loss": 1.5105,
+      "step": 301
+    },
+    {
+      "epoch": 0.026506414036550546,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.000906,
+      "loss": 1.5134,
+      "step": 302
+    },
+    {
+      "epoch": 0.02659418361945303,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.000909,
+      "loss": 1.4693,
+      "step": 303
+    },
+    {
+      "epoch": 0.026681953202355516,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.000912,
+      "loss": 1.4779,
+      "step": 304
+    },
+    {
+      "epoch": 0.026769722785258,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.000915,
+      "loss": 1.4689,
+      "step": 305
+    },
+    {
+      "epoch": 0.02685749236816049,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.000918,
+      "loss": 1.4608,
+      "step": 306
+    },
+    {
+      "epoch": 0.026945261951062973,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.000921,
+      "loss": 1.5206,
+      "step": 307
+    },
+    {
+      "epoch": 0.027033031533965458,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.000924,
+      "loss": 1.4968,
+      "step": 308
+    },
+    {
+      "epoch": 0.027120801116867942,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.000927,
+      "loss": 1.4968,
+      "step": 309
+    },
+    {
+      "epoch": 0.027208570699770427,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00093,
+      "loss": 1.4736,
+      "step": 310
+    },
+    {
+      "epoch": 0.02729634028267291,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.000933,
+      "loss": 1.4864,
+      "step": 311
+    },
+    {
+      "epoch": 0.0273841098655754,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.000936,
+      "loss": 1.4975,
+      "step": 312
+    },
+    {
+      "epoch": 0.027471879448477884,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.0009390000000000001,
+      "loss": 1.4502,
+      "step": 313
+    },
+    {
+      "epoch": 0.02755964903138037,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.000942,
+      "loss": 1.4725,
+      "step": 314
+    },
+    {
+      "epoch": 0.027647418614282854,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.000945,
+      "loss": 1.505,
+      "step": 315
+    },
+    {
+      "epoch": 0.02773518819718534,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0009480000000000001,
+      "loss": 1.4774,
+      "step": 316
+    },
+    {
+      "epoch": 0.027822957780087823,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.000951,
+      "loss": 1.5004,
+      "step": 317
+    },
+    {
+      "epoch": 0.02791072736299031,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.000954,
+      "loss": 1.4844,
+      "step": 318
+    },
+    {
+      "epoch": 0.027998496945892796,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0009570000000000001,
+      "loss": 1.4924,
+      "step": 319
+    },
+    {
+      "epoch": 0.02808626652879528,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00096,
+      "loss": 1.4896,
+      "step": 320
+    },
+    {
+      "epoch": 0.028174036111697765,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.000963,
+      "loss": 1.5079,
+      "step": 321
+    },
+    {
+      "epoch": 0.02826180569460025,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0009660000000000001,
+      "loss": 1.4842,
+      "step": 322
+    },
+    {
+      "epoch": 0.028349575277502734,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.000969,
+      "loss": 1.4825,
+      "step": 323
+    },
+    {
+      "epoch": 0.028437344860405223,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0009720000000000001,
+      "loss": 1.502,
+      "step": 324
+    },
+    {
+      "epoch": 0.028525114443307707,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.0009750000000000001,
+      "loss": 1.4996,
+      "step": 325
+    },
+    {
+      "epoch": 0.028612884026210192,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.0009780000000000001,
+      "loss": 1.4486,
+      "step": 326
+    },
+    {
+      "epoch": 0.028700653609112677,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.000981,
+      "loss": 1.483,
+      "step": 327
+    },
+    {
+      "epoch": 0.02878842319201516,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.000984,
+      "loss": 1.465,
+      "step": 328
+    },
+    {
+      "epoch": 0.028876192774917646,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.000987,
+      "loss": 1.4744,
+      "step": 329
+    },
+    {
+      "epoch": 0.028963962357820134,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00099,
+      "loss": 1.4533,
+      "step": 330
+    },
+    {
+      "epoch": 0.02905173194072262,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0009930000000000002,
+      "loss": 1.4598,
+      "step": 331
+    },
+    {
+      "epoch": 0.029139501523625103,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0009960000000000001,
+      "loss": 1.4901,
+      "step": 332
+    },
+    {
+      "epoch": 0.029227271106527588,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.000999,
+      "loss": 1.5001,
+      "step": 333
+    },
+    {
+      "epoch": 0.029315040689430073,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001002,
+      "loss": 1.4961,
+      "step": 334
+    },
+    {
+      "epoch": 0.029402810272332557,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.001005,
+      "loss": 1.4744,
+      "step": 335
+    },
+    {
+      "epoch": 0.029490579855235045,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.001008,
+      "loss": 1.4319,
+      "step": 336
+    },
+    {
+      "epoch": 0.02957834943813753,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0010110000000000002,
+      "loss": 1.4544,
+      "step": 337
+    },
+    {
+      "epoch": 0.029666119021040015,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0010140000000000001,
+      "loss": 1.4933,
+      "step": 338
+    },
+    {
+      "epoch": 0.0297538886039425,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0010170000000000001,
+      "loss": 1.4171,
+      "step": 339
+    },
+    {
+      "epoch": 0.029841658186844984,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00102,
+      "loss": 1.4674,
+      "step": 340
+    },
+    {
+      "epoch": 0.029929427769747472,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.001023,
+      "loss": 1.4816,
+      "step": 341
+    },
+    {
+      "epoch": 0.030017197352649957,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.001026,
+      "loss": 1.5247,
+      "step": 342
+    },
+    {
+      "epoch": 0.03010496693555244,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0010290000000000002,
+      "loss": 1.5058,
+      "step": 343
+    },
+    {
+      "epoch": 0.030192736518454926,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.001032,
+      "loss": 1.4926,
+      "step": 344
+    },
+    {
+      "epoch": 0.03028050610135741,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001035,
+      "loss": 1.4171,
+      "step": 345
+    },
+    {
+      "epoch": 0.030368275684259895,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0010379999999999999,
+      "loss": 1.4589,
+      "step": 346
+    },
+    {
+      "epoch": 0.030456045267162384,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.001041,
+      "loss": 1.4681,
+      "step": 347
+    },
+    {
+      "epoch": 0.030543814850064868,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.001044,
+      "loss": 1.4715,
+      "step": 348
+    },
+    {
+      "epoch": 0.030631584432967353,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001047,
+      "loss": 1.4296,
+      "step": 349
+    },
+    {
+      "epoch": 0.030719354015869837,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00105,
+      "loss": 1.4862,
+      "step": 350
+    },
+    {
+      "epoch": 0.030807123598772322,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.001053,
+      "loss": 1.4369,
+      "step": 351
+    },
+    {
+      "epoch": 0.030894893181674807,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0010559999999999999,
+      "loss": 1.506,
+      "step": 352
+    },
+    {
+      "epoch": 0.030982662764577295,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.001059,
+      "loss": 1.4843,
+      "step": 353
+    },
+    {
+      "epoch": 0.03107043234747978,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.001062,
+      "loss": 1.4721,
+      "step": 354
+    },
+    {
+      "epoch": 0.031158201930382264,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.001065,
+      "loss": 1.4635,
+      "step": 355
+    },
+    {
+      "epoch": 0.03124597151328475,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.001068,
+      "loss": 1.5184,
+      "step": 356
+    },
+    {
+      "epoch": 0.03133374109618724,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.001071,
+      "loss": 1.4645,
+      "step": 357
+    },
+    {
+      "epoch": 0.03142151067908972,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.001074,
+      "loss": 1.4442,
+      "step": 358
+    },
+    {
+      "epoch": 0.031509280261992206,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.001077,
+      "loss": 1.4393,
+      "step": 359
+    },
+    {
+      "epoch": 0.03159704984489469,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00108,
+      "loss": 1.495,
+      "step": 360
+    },
+    {
+      "epoch": 0.031684819427797176,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.001083,
+      "loss": 1.4464,
+      "step": 361
+    },
+    {
+      "epoch": 0.03177258901069966,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.001086,
+      "loss": 1.4554,
+      "step": 362
+    },
+    {
+      "epoch": 0.031860358593602145,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.001089,
+      "loss": 1.4758,
+      "step": 363
+    },
+    {
+      "epoch": 0.03194812817650463,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.001092,
+      "loss": 1.4644,
+      "step": 364
+    },
+    {
+      "epoch": 0.032035897759407114,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.001095,
+      "loss": 1.4704,
+      "step": 365
+    },
+    {
+      "epoch": 0.0321236673423096,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.001098,
+      "loss": 1.4682,
+      "step": 366
+    },
+    {
+      "epoch": 0.032211436925212084,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.001101,
+      "loss": 1.4806,
+      "step": 367
+    },
+    {
+      "epoch": 0.032299206508114575,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001104,
+      "loss": 1.4375,
+      "step": 368
+    },
+    {
+      "epoch": 0.03238697609101706,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.001107,
+      "loss": 1.4962,
+      "step": 369
+    },
+    {
+      "epoch": 0.032474745673919544,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00111,
+      "loss": 1.4409,
+      "step": 370
+    },
+    {
+      "epoch": 0.03256251525682203,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.001113,
+      "loss": 1.5093,
+      "step": 371
+    },
+    {
+      "epoch": 0.032650284839724514,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.001116,
+      "loss": 1.4576,
+      "step": 372
+    },
+    {
+      "epoch": 0.032738054422627,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.001119,
+      "loss": 1.4691,
+      "step": 373
+    },
+    {
+      "epoch": 0.03282582400552948,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.001122,
+      "loss": 1.4446,
+      "step": 374
+    },
+    {
+      "epoch": 0.03291359358843197,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0011250000000000001,
+      "loss": 1.4638,
+      "step": 375
+    },
+    {
+      "epoch": 0.03300136317133445,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.001128,
+      "loss": 1.4647,
+      "step": 376
+    },
+    {
+      "epoch": 0.03308913275423694,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.001131,
+      "loss": 1.4811,
+      "step": 377
+    },
+    {
+      "epoch": 0.03317690233713942,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.001134,
+      "loss": 1.4713,
+      "step": 378
+    },
+    {
+      "epoch": 0.03326467192004191,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.001137,
+      "loss": 1.455,
+      "step": 379
+    },
+    {
+      "epoch": 0.0333524415029444,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00114,
+      "loss": 1.4806,
+      "step": 380
+    },
+    {
+      "epoch": 0.03344021108584688,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0011430000000000001,
+      "loss": 1.4486,
+      "step": 381
+    },
+    {
+      "epoch": 0.03352798066874937,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.001146,
+      "loss": 1.5052,
+      "step": 382
+    },
+    {
+      "epoch": 0.03361575025165185,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.001149,
+      "loss": 1.4378,
+      "step": 383
+    },
+    {
+      "epoch": 0.03370351983455434,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.001152,
+      "loss": 1.4448,
+      "step": 384
+    },
+    {
+      "epoch": 0.03379128941745682,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.001155,
+      "loss": 1.4563,
+      "step": 385
+    },
+    {
+      "epoch": 0.033879059000359306,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.001158,
+      "loss": 1.4746,
+      "step": 386
+    },
+    {
+      "epoch": 0.03396682858326179,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0011610000000000001,
+      "loss": 1.4195,
+      "step": 387
+    },
+    {
+      "epoch": 0.034054598166164275,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.001164,
+      "loss": 1.4748,
+      "step": 388
+    },
+    {
+      "epoch": 0.03414236774906676,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.001167,
+      "loss": 1.4435,
+      "step": 389
+    },
+    {
+      "epoch": 0.034230137331969245,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00117,
+      "loss": 1.5054,
+      "step": 390
+    },
+    {
+      "epoch": 0.034317906914871736,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.001173,
+      "loss": 1.4637,
+      "step": 391
+    },
+    {
+      "epoch": 0.03440567649777422,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.001176,
+      "loss": 1.4625,
+      "step": 392
+    },
+    {
+      "epoch": 0.034493446080676705,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0011790000000000001,
+      "loss": 1.4551,
+      "step": 393
+    },
+    {
+      "epoch": 0.03458121566357919,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.001182,
+      "loss": 1.4235,
+      "step": 394
+    },
+    {
+      "epoch": 0.034668985246481675,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.001185,
+      "loss": 1.4662,
+      "step": 395
+    },
+    {
+      "epoch": 0.03475675482938416,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.001188,
+      "loss": 1.4661,
+      "step": 396
+    },
+    {
+      "epoch": 0.034844524412286644,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.001191,
+      "loss": 1.4496,
+      "step": 397
+    },
+    {
+      "epoch": 0.03493229399518913,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.0011940000000000002,
+      "loss": 1.4556,
+      "step": 398
+    },
+    {
+      "epoch": 0.03502006357809161,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.0011970000000000001,
+      "loss": 1.4646,
+      "step": 399
+    },
+    {
+      "epoch": 0.0351078331609941,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0012000000000000001,
+      "loss": 1.4353,
+      "step": 400
+    },
+    {
+      "epoch": 0.03519560274389658,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001203,
+      "loss": 1.4552,
+      "step": 401
+    },
+    {
+      "epoch": 0.03528337232679907,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.001206,
+      "loss": 1.4238,
+      "step": 402
+    },
+    {
+      "epoch": 0.03537114190970156,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.001209,
+      "loss": 1.467,
+      "step": 403
+    },
+    {
+      "epoch": 0.035458911492604044,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0012120000000000002,
+      "loss": 1.488,
+      "step": 404
+    },
+    {
+      "epoch": 0.03554668107550653,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.0012150000000000002,
+      "loss": 1.505,
+      "step": 405
+    },
+    {
+      "epoch": 0.03563445065840901,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.0012180000000000001,
+      "loss": 1.4489,
+      "step": 406
+    },
+    {
+      "epoch": 0.0357222202413115,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.0012209999999999999,
+      "loss": 1.4478,
+      "step": 407
+    },
+    {
+      "epoch": 0.03580998982421398,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.001224,
+      "loss": 1.5016,
+      "step": 408
+    },
+    {
+      "epoch": 0.03589775940711647,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.001227,
+      "loss": 1.4721,
+      "step": 409
+    },
+    {
+      "epoch": 0.03598552899001895,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00123,
+      "loss": 1.4401,
+      "step": 410
+    },
+    {
+      "epoch": 0.036073298572921436,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.001233,
+      "loss": 1.4316,
+      "step": 411
+    },
+    {
+      "epoch": 0.03616106815582392,
+      "grad_norm": 0.375,
+      "learning_rate": 0.001236,
+      "loss": 1.4899,
+      "step": 412
+    },
+    {
+      "epoch": 0.036248837738726405,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0012389999999999999,
+      "loss": 1.459,
+      "step": 413
+    },
+    {
+      "epoch": 0.0363366073216289,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.001242,
+      "loss": 1.4614,
+      "step": 414
+    },
+    {
+      "epoch": 0.03642437690453138,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.001245,
+      "loss": 1.4686,
+      "step": 415
+    },
+    {
+      "epoch": 0.036512146487433866,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.001248,
+      "loss": 1.4314,
+      "step": 416
+    },
+    {
+      "epoch": 0.03659991607033635,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.001251,
+      "loss": 1.4779,
+      "step": 417
+    },
+    {
+      "epoch": 0.036687685653238836,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.001254,
+      "loss": 1.4588,
+      "step": 418
+    },
+    {
+      "epoch": 0.03677545523614132,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0012569999999999999,
+      "loss": 1.4517,
+      "step": 419
+    },
+    {
+      "epoch": 0.036863224819043805,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.00126,
+      "loss": 1.4708,
+      "step": 420
+    },
+    {
+      "epoch": 0.03695099440194629,
+      "grad_norm": 0.25,
+      "learning_rate": 0.001263,
+      "loss": 1.3873,
+      "step": 421
+    },
+    {
+      "epoch": 0.037038763984848774,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.001266,
+      "loss": 1.4315,
+      "step": 422
+    },
+    {
+      "epoch": 0.03712653356775126,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001269,
+      "loss": 1.4852,
+      "step": 423
+    },
+    {
+      "epoch": 0.037214303150653744,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.001272,
+      "loss": 1.4162,
+      "step": 424
+    },
+    {
+      "epoch": 0.03730207273355623,
+      "grad_norm": 0.375,
+      "learning_rate": 0.001275,
+      "loss": 1.4792,
+      "step": 425
+    },
+    {
+      "epoch": 0.03738984231645872,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.001278,
+      "loss": 1.4004,
+      "step": 426
+    },
+    {
+      "epoch": 0.037477611899361205,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.001281,
+      "loss": 1.4025,
+      "step": 427
+    },
+    {
+      "epoch": 0.03756538148226369,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.001284,
+      "loss": 1.4255,
+      "step": 428
+    },
+    {
+      "epoch": 0.037653151065166174,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.001287,
+      "loss": 1.4704,
+      "step": 429
+    },
+    {
+      "epoch": 0.03774092064806866,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00129,
+      "loss": 1.3953,
+      "step": 430
+    },
+    {
+      "epoch": 0.03782869023097114,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.001293,
+      "loss": 1.4279,
+      "step": 431
+    },
+    {
+      "epoch": 0.03791645981387363,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.001296,
+      "loss": 1.4645,
+      "step": 432
+    },
+    {
+      "epoch": 0.03800422939677611,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001299,
+      "loss": 1.4312,
+      "step": 433
+    },
+    {
+      "epoch": 0.0380919989796786,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.001302,
+      "loss": 1.4486,
+      "step": 434
+    },
+    {
+      "epoch": 0.03817976856258108,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.001305,
+      "loss": 1.4482,
+      "step": 435
+    },
+    {
+      "epoch": 0.038267538145483566,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.001308,
+      "loss": 1.4681,
+      "step": 436
+    },
+    {
+      "epoch": 0.03835530772838605,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001311,
+      "loss": 1.4597,
+      "step": 437
+    },
+    {
+      "epoch": 0.03844307731128854,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.001314,
+      "loss": 1.4637,
+      "step": 438
+    },
+    {
+      "epoch": 0.03853084689419103,
+      "grad_norm": 2.390625,
+      "learning_rate": 0.001317,
+      "loss": 1.4723,
+      "step": 439
+    },
+    {
+      "epoch": 0.03861861647709351,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00132,
+      "loss": 1.4858,
+      "step": 440
+    },
+    {
+      "epoch": 0.038706386059996,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.001323,
+      "loss": 1.4264,
+      "step": 441
+    },
+    {
+      "epoch": 0.03879415564289848,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0013260000000000001,
+      "loss": 1.4633,
+      "step": 442
+    },
+    {
+      "epoch": 0.038881925225800966,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.001329,
+      "loss": 1.4634,
+      "step": 443
+    },
+    {
+      "epoch": 0.03896969480870345,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.001332,
+      "loss": 1.4455,
+      "step": 444
+    },
+    {
+      "epoch": 0.039057464391605935,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.001335,
+      "loss": 1.4572,
+      "step": 445
+    },
+    {
+      "epoch": 0.03914523397450842,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001338,
+      "loss": 1.4889,
+      "step": 446
+    },
+    {
+      "epoch": 0.039233003557410905,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001341,
+      "loss": 1.4454,
+      "step": 447
+    },
+    {
+      "epoch": 0.03932077314031339,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0013440000000000001,
+      "loss": 1.4824,
+      "step": 448
+    },
+    {
+      "epoch": 0.03940854272321588,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.001347,
+      "loss": 1.4214,
+      "step": 449
+    },
+    {
+      "epoch": 0.039496312306118365,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00135,
+      "loss": 1.4663,
+      "step": 450
+    },
+    {
+      "epoch": 0.03958408188902085,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.001353,
+      "loss": 1.4223,
+      "step": 451
+    },
+    {
+      "epoch": 0.039671851471923335,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.001356,
+      "loss": 1.4393,
+      "step": 452
+    },
+    {
+      "epoch": 0.03975962105482582,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.001359,
+      "loss": 1.4849,
+      "step": 453
+    },
+    {
+      "epoch": 0.039847390637728304,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0013620000000000001,
+      "loss": 1.4257,
+      "step": 454
+    },
+    {
+      "epoch": 0.03993516022063079,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0013650000000000001,
+      "loss": 1.451,
+      "step": 455
+    },
+    {
+      "epoch": 0.04002292980353327,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.001368,
+      "loss": 1.466,
+      "step": 456
+    },
+    {
+      "epoch": 0.04011069938643576,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.001371,
+      "loss": 1.4471,
+      "step": 457
+    },
+    {
+      "epoch": 0.04019846896933824,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.001374,
+      "loss": 1.3836,
+      "step": 458
+    },
+    {
+      "epoch": 0.04028623855224073,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0013770000000000002,
+      "loss": 1.4219,
+      "step": 459
+    },
+    {
+      "epoch": 0.04037400813514321,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0013800000000000002,
+      "loss": 1.4534,
+      "step": 460
+    },
+    {
+      "epoch": 0.040461777718045704,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0013830000000000001,
+      "loss": 1.4593,
+      "step": 461
+    },
+    {
+      "epoch": 0.04054954730094819,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.001386,
+      "loss": 1.4534,
+      "step": 462
+    },
+    {
+      "epoch": 0.04063731688385067,
+      "grad_norm": 1.625,
+      "learning_rate": 0.001389,
+      "loss": 1.4809,
+      "step": 463
+    },
+    {
+      "epoch": 0.04072508646675316,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.001392,
+      "loss": 1.452,
+      "step": 464
+    },
+    {
+      "epoch": 0.04081285604965564,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.0013950000000000002,
+      "loss": 1.4777,
+      "step": 465
+    },
+    {
+      "epoch": 0.04090062563255813,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0013980000000000002,
+      "loss": 1.4226,
+      "step": 466
+    },
+    {
+      "epoch": 0.04098839521546061,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0014010000000000001,
+      "loss": 1.4141,
+      "step": 467
+    },
+    {
+      "epoch": 0.041076164798363096,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.001404,
+      "loss": 1.4417,
+      "step": 468
+    },
+    {
+      "epoch": 0.04116393438126558,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.001407,
+      "loss": 1.4534,
+      "step": 469
+    },
+    {
+      "epoch": 0.041251703964168066,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00141,
+      "loss": 1.4575,
+      "step": 470
+    },
+    {
+      "epoch": 0.04133947354707055,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.001413,
+      "loss": 1.4112,
+      "step": 471
+    },
+    {
+      "epoch": 0.041427243129973035,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.001416,
+      "loss": 1.4343,
+      "step": 472
+    },
+    {
+      "epoch": 0.041515012712875526,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.001419,
+      "loss": 1.4372,
+      "step": 473
+    },
+    {
+      "epoch": 0.04160278229577801,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0014219999999999999,
+      "loss": 1.4323,
+      "step": 474
+    },
+    {
+      "epoch": 0.041690551878680496,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.001425,
+      "loss": 1.4271,
+      "step": 475
+    },
+    {
+      "epoch": 0.04177832146158298,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.001428,
+      "loss": 1.4238,
+      "step": 476
+    },
+    {
+      "epoch": 0.041866091044485465,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.001431,
+      "loss": 1.4322,
+      "step": 477
+    },
+    {
+      "epoch": 0.04195386062738795,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.001434,
+      "loss": 1.4554,
+      "step": 478
+    },
+    {
+      "epoch": 0.042041630210290434,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.001437,
+      "loss": 1.4249,
+      "step": 479
+    },
+    {
+      "epoch": 0.04212939979319292,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.0014399999999999999,
+      "loss": 1.4723,
+      "step": 480
+    },
+    {
+      "epoch": 0.042217169376095404,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.001443,
+      "loss": 1.4306,
+      "step": 481
+    },
+    {
+      "epoch": 0.04230493895899789,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.001446,
+      "loss": 1.454,
+      "step": 482
+    },
+    {
+      "epoch": 0.04239270854190037,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.001449,
+      "loss": 1.4104,
+      "step": 483
+    },
+    {
+      "epoch": 0.042480478124802865,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.001452,
+      "loss": 1.4182,
+      "step": 484
+    },
+    {
+      "epoch": 0.04256824770770535,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.001455,
+      "loss": 1.456,
+      "step": 485
+    },
+    {
+      "epoch": 0.042656017290607834,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.001458,
+      "loss": 1.4492,
+      "step": 486
+    },
+    {
+      "epoch": 0.04274378687351032,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.001461,
+      "loss": 1.4531,
+      "step": 487
+    },
+    {
+      "epoch": 0.0428315564564128,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.001464,
+      "loss": 1.4238,
+      "step": 488
+    },
+    {
+      "epoch": 0.04291932603931529,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.001467,
+      "loss": 1.3839,
+      "step": 489
+    },
+    {
+      "epoch": 0.04300709562221777,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00147,
+      "loss": 1.4123,
+      "step": 490
+    },
+    {
+      "epoch": 0.04309486520512026,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.001473,
+      "loss": 1.499,
+      "step": 491
+    },
+    {
+      "epoch": 0.04318263478802274,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.001476,
+      "loss": 1.4202,
+      "step": 492
+    },
+    {
+      "epoch": 0.043270404370925226,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.001479,
+      "loss": 1.4173,
+      "step": 493
+    },
+    {
+      "epoch": 0.04335817395382771,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.001482,
+      "loss": 1.4403,
+      "step": 494
+    },
+    {
+      "epoch": 0.043445943536730196,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.001485,
+      "loss": 1.4078,
+      "step": 495
+    },
+    {
+      "epoch": 0.04353371311963269,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.001488,
+      "loss": 1.4551,
+      "step": 496
+    },
+    {
+      "epoch": 0.04362148270253517,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.001491,
+      "loss": 1.4303,
+      "step": 497
+    },
+    {
+      "epoch": 0.04370925228543766,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.001494,
+      "loss": 1.3937,
+      "step": 498
+    },
+    {
+      "epoch": 0.04379702186834014,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001497,
+      "loss": 1.464,
+      "step": 499
+    },
+    {
+      "epoch": 0.043884791451242626,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.0015,
+      "loss": 1.4162,
+      "step": 500
+    },
+    {
+      "epoch": 0.043884791451242626,
+      "eval_loss": 1.4264767169952393,
+      "eval_runtime": 436.4842,
+      "eval_samples_per_second": 33.77,
+      "eval_steps_per_second": 8.442,
+      "step": 500
+    },
+    {
+      "epoch": 0.04397256103414511,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.001503,
+      "loss": 1.4449,
+      "step": 501
+    },
+    {
+      "epoch": 0.044060330617047595,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001506,
+      "loss": 1.4514,
+      "step": 502
+    },
+    {
+      "epoch": 0.04414810019995008,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0015090000000000001,
+      "loss": 1.394,
+      "step": 503
+    },
+    {
+      "epoch": 0.044235869782852565,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.001512,
+      "loss": 1.4512,
+      "step": 504
+    },
+    {
+      "epoch": 0.04432363936575505,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.001515,
+      "loss": 1.4325,
+      "step": 505
+    },
+    {
+      "epoch": 0.044411408948657534,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.001518,
+      "loss": 1.4595,
+      "step": 506
+    },
+    {
+      "epoch": 0.044499178531560026,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.001521,
+      "loss": 1.4304,
+      "step": 507
+    },
+    {
+      "epoch": 0.04458694811446251,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.001524,
+      "loss": 1.4345,
+      "step": 508
+    },
+    {
+      "epoch": 0.044674717697364995,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0015270000000000001,
+      "loss": 1.4321,
+      "step": 509
+    },
+    {
+      "epoch": 0.04476248728026748,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0015300000000000001,
+      "loss": 1.4396,
+      "step": 510
+    },
+    {
+      "epoch": 0.044850256863169964,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.001533,
+      "loss": 1.4269,
+      "step": 511
+    },
+    {
+      "epoch": 0.04493802644607245,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.001536,
+      "loss": 1.424,
+      "step": 512
+    },
+    {
+      "epoch": 0.04502579602897493,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.001539,
+      "loss": 1.4528,
+      "step": 513
+    },
+    {
+      "epoch": 0.04511356561187742,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.001542,
+      "loss": 1.4619,
+      "step": 514
+    },
+    {
+      "epoch": 0.0452013351947799,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0015450000000000001,
+      "loss": 1.3829,
+      "step": 515
+    },
+    {
+      "epoch": 0.04528910477768239,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0015480000000000001,
+      "loss": 1.3826,
+      "step": 516
+    },
+    {
+      "epoch": 0.04537687436058487,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.001551,
+      "loss": 1.4539,
+      "step": 517
+    },
+    {
+      "epoch": 0.04546464394348736,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.001554,
+      "loss": 1.4542,
+      "step": 518
+    },
+    {
+      "epoch": 0.04555241352638985,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.001557,
+      "loss": 1.3974,
+      "step": 519
+    },
+    {
+      "epoch": 0.04564018310929233,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0015600000000000002,
+      "loss": 1.3831,
+      "step": 520
+    },
+    {
+      "epoch": 0.04572795269219482,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0015630000000000002,
+      "loss": 1.4389,
+      "step": 521
+    },
+    {
+      "epoch": 0.0458157222750973,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0015660000000000001,
+      "loss": 1.4368,
+      "step": 522
+    },
+    {
+      "epoch": 0.04590349185799979,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.001569,
+      "loss": 1.4652,
+      "step": 523
+    },
+    {
+      "epoch": 0.04599126144090227,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.001572,
+      "loss": 1.4344,
+      "step": 524
+    },
+    {
+      "epoch": 0.046079031023804756,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.001575,
+      "loss": 1.413,
+      "step": 525
+    },
+    {
+      "epoch": 0.04616680060670724,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0015780000000000002,
+      "loss": 1.4295,
+      "step": 526
+    },
+    {
+      "epoch": 0.046254570189609726,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0015810000000000002,
+      "loss": 1.4291,
+      "step": 527
+    },
+    {
+      "epoch": 0.04634233977251221,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0015840000000000001,
+      "loss": 1.4636,
+      "step": 528
+    },
+    {
+      "epoch": 0.046430109355414695,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.001587,
+      "loss": 1.4616,
+      "step": 529
+    },
+    {
+      "epoch": 0.04651787893831718,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00159,
+      "loss": 1.4599,
+      "step": 530
+    },
+    {
+      "epoch": 0.04660564852121967,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.001593,
+      "loss": 1.4816,
+      "step": 531
+    },
+    {
+      "epoch": 0.046693418104122156,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0015960000000000002,
+      "loss": 1.4006,
+      "step": 532
+    },
+    {
+      "epoch": 0.04678118768702464,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.0015990000000000002,
+      "loss": 1.4182,
+      "step": 533
+    },
+    {
+      "epoch": 0.046868957269927125,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0016020000000000001,
+      "loss": 1.4418,
+      "step": 534
+    },
+    {
+      "epoch": 0.04695672685282961,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.001605,
+      "loss": 1.3954,
+      "step": 535
+    },
+    {
+      "epoch": 0.047044496435732094,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.001608,
+      "loss": 1.4086,
+      "step": 536
+    },
+    {
+      "epoch": 0.04713226601863458,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0016110000000000002,
+      "loss": 1.4538,
+      "step": 537
+    },
+    {
+      "epoch": 0.047220035601537064,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0016140000000000002,
+      "loss": 1.4082,
+      "step": 538
+    },
+    {
+      "epoch": 0.04730780518443955,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0016170000000000002,
+      "loss": 1.4373,
+      "step": 539
+    },
+    {
+      "epoch": 0.04739557476734203,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0016200000000000001,
+      "loss": 1.4836,
+      "step": 540
+    },
+    {
+      "epoch": 0.04748334435024452,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.001623,
+      "loss": 1.4136,
+      "step": 541
+    },
+    {
+      "epoch": 0.04757111393314701,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.001626,
+      "loss": 1.4191,
+      "step": 542
+    },
+    {
+      "epoch": 0.047658883516049494,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0016290000000000002,
+      "loss": 1.4583,
+      "step": 543
+    },
+    {
+      "epoch": 0.04774665309895198,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0016320000000000002,
+      "loss": 1.4471,
+      "step": 544
+    },
+    {
+      "epoch": 0.04783442268185446,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0016350000000000002,
+      "loss": 1.5006,
+      "step": 545
+    },
+    {
+      "epoch": 0.04792219226475695,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.0016380000000000001,
+      "loss": 1.4461,
+      "step": 546
+    },
+    {
+      "epoch": 0.04800996184765943,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.001641,
+      "loss": 1.4521,
+      "step": 547
+    },
+    {
+      "epoch": 0.04809773143056192,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.001644,
+      "loss": 1.4113,
+      "step": 548
+    },
+    {
+      "epoch": 0.0481855010134644,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.0016470000000000002,
+      "loss": 1.4362,
+      "step": 549
+    },
+    {
+      "epoch": 0.048273270596366886,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.0016500000000000002,
+      "loss": 1.3773,
+      "step": 550
+    },
+    {
+      "epoch": 0.04836104017926937,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0016530000000000002,
+      "loss": 1.4137,
+      "step": 551
+    },
+    {
+      "epoch": 0.048448809762171856,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0016560000000000001,
+      "loss": 1.3973,
+      "step": 552
+    },
+    {
+      "epoch": 0.04853657934507434,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.001659,
+      "loss": 1.4353,
+      "step": 553
+    },
+    {
+      "epoch": 0.04862434892797683,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.0016620000000000003,
+      "loss": 1.3787,
+      "step": 554
+    },
+    {
+      "epoch": 0.04871211851087932,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.0016650000000000002,
+      "loss": 1.4034,
+      "step": 555
+    },
+    {
+      "epoch": 0.0487998880937818,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0016680000000000002,
+      "loss": 1.4361,
+      "step": 556
+    },
+    {
+      "epoch": 0.048887657676684286,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0016710000000000002,
+      "loss": 1.4113,
+      "step": 557
+    },
+    {
+      "epoch": 0.04897542725958677,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0016740000000000001,
+      "loss": 1.4935,
+      "step": 558
+    },
+    {
+      "epoch": 0.049063196842489255,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.001677,
+      "loss": 1.4107,
+      "step": 559
+    },
+    {
+      "epoch": 0.04915096642539174,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0016800000000000003,
+      "loss": 1.4258,
+      "step": 560
+    },
+    {
+      "epoch": 0.049238736008294225,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.0016830000000000003,
+      "loss": 1.4335,
+      "step": 561
+    },
+    {
+      "epoch": 0.04932650559119671,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0016860000000000002,
+      "loss": 1.432,
+      "step": 562
+    },
+    {
+      "epoch": 0.049414275174099194,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.001689,
+      "loss": 1.4373,
+      "step": 563
+    },
+    {
+      "epoch": 0.04950204475700168,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.001692,
+      "loss": 1.3948,
+      "step": 564
+    },
+    {
+      "epoch": 0.04958981433990416,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.001695,
+      "loss": 1.4371,
+      "step": 565
+    },
+    {
+      "epoch": 0.049677583922806655,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0016979999999999999,
+      "loss": 1.3975,
+      "step": 566
+    },
+    {
+      "epoch": 0.04976535350570914,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0017009999999999998,
+      "loss": 1.4313,
+      "step": 567
+    },
+    {
+      "epoch": 0.049853123088611624,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0017039999999999998,
+      "loss": 1.4414,
+      "step": 568
+    },
+    {
+      "epoch": 0.04994089267151411,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.001707,
+      "loss": 1.4192,
+      "step": 569
+    },
+    {
+      "epoch": 0.050028662254416593,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00171,
+      "loss": 1.425,
+      "step": 570
+    },
+    {
+      "epoch": 0.05011643183731908,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.001713,
+      "loss": 1.421,
+      "step": 571
+    },
+    {
+      "epoch": 0.05020420142022156,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0017159999999999999,
+      "loss": 1.4101,
+      "step": 572
+    },
+    {
+      "epoch": 0.05029197100312405,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0017189999999999998,
+      "loss": 1.4414,
+      "step": 573
+    },
+    {
+      "epoch": 0.05037974058602653,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.001722,
+      "loss": 1.4248,
+      "step": 574
+    },
+    {
+      "epoch": 0.05046751016892902,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.001725,
+      "loss": 1.4058,
+      "step": 575
+    },
+    {
+      "epoch": 0.0505552797518315,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.001728,
+      "loss": 1.4156,
+      "step": 576
+    },
+    {
+      "epoch": 0.05064304933473399,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.001731,
+      "loss": 1.4011,
+      "step": 577
+    },
+    {
+      "epoch": 0.05073081891763648,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0017339999999999999,
+      "loss": 1.4422,
+      "step": 578
+    },
+    {
+      "epoch": 0.05081858850053896,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0017369999999999998,
+      "loss": 1.4312,
+      "step": 579
+    },
+    {
+      "epoch": 0.05090635808344145,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00174,
+      "loss": 1.4641,
+      "step": 580
+    },
+    {
+      "epoch": 0.05099412766634393,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.001743,
+      "loss": 1.4104,
+      "step": 581
+    },
+    {
+      "epoch": 0.051081897249246416,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.001746,
+      "loss": 1.3759,
+      "step": 582
+    },
+    {
+      "epoch": 0.0511696668321489,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.001749,
+      "loss": 1.4461,
+      "step": 583
+    },
+    {
+      "epoch": 0.051257436415051386,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.0017519999999999999,
+      "loss": 1.4052,
+      "step": 584
+    },
+    {
+      "epoch": 0.05134520599795387,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0017549999999999998,
+      "loss": 1.4361,
+      "step": 585
+    },
+    {
+      "epoch": 0.051432975580856355,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001758,
+      "loss": 1.4081,
+      "step": 586
+    },
+    {
+      "epoch": 0.05152074516375884,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.001761,
+      "loss": 1.4229,
+      "step": 587
+    },
+    {
+      "epoch": 0.051608514746661324,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.001764,
+      "loss": 1.4724,
+      "step": 588
+    },
+    {
+      "epoch": 0.051696284329563816,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.001767,
+      "loss": 1.4288,
+      "step": 589
+    },
+    {
+      "epoch": 0.0517840539124663,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.0017699999999999999,
+      "loss": 1.5109,
+      "step": 590
+    },
+    {
+      "epoch": 0.051871823495368785,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.001773,
+      "loss": 1.4384,
+      "step": 591
+    },
+    {
+      "epoch": 0.05195959307827127,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.001776,
+      "loss": 1.3817,
+      "step": 592
+    },
+    {
+      "epoch": 0.052047362661173754,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.001779,
+      "loss": 1.4257,
+      "step": 593
+    },
+    {
+      "epoch": 0.05213513224407624,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.001782,
+      "loss": 1.4052,
+      "step": 594
+    },
+    {
+      "epoch": 0.052222901826978724,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.001785,
+      "loss": 1.4482,
+      "step": 595
+    },
+    {
+      "epoch": 0.05231067140988121,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0017879999999999999,
+      "loss": 1.4112,
+      "step": 596
+    },
+    {
+      "epoch": 0.05239844099278369,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.001791,
+      "loss": 1.4303,
+      "step": 597
+    },
+    {
+      "epoch": 0.05248621057568618,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.001794,
+      "loss": 1.4169,
+      "step": 598
+    },
+    {
+      "epoch": 0.05257398015858866,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.001797,
+      "loss": 1.408,
+      "step": 599
+    },
+    {
+      "epoch": 0.05266174974149115,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0018,
+      "loss": 1.4476,
+      "step": 600
+    },
+    {
+      "epoch": 0.05274951932439364,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.001803,
+      "loss": 1.4048,
+      "step": 601
+    },
+    {
+      "epoch": 0.05283728890729612,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.0018059999999999999,
+      "loss": 1.3852,
+      "step": 602
+    },
+    {
+      "epoch": 0.05292505849019861,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.001809,
+      "loss": 1.4454,
+      "step": 603
+    },
+    {
+      "epoch": 0.05301282807310109,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.001812,
+      "loss": 1.4336,
+      "step": 604
+    },
+    {
+      "epoch": 0.05310059765600358,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.001815,
+      "loss": 1.4297,
+      "step": 605
+    },
+    {
+      "epoch": 0.05318836723890606,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.001818,
+      "loss": 1.4429,
+      "step": 606
+    },
+    {
+      "epoch": 0.05327613682180855,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.001821,
+      "loss": 1.4271,
+      "step": 607
+    },
+    {
+      "epoch": 0.05336390640471103,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001824,
+      "loss": 1.4233,
+      "step": 608
+    },
+    {
+      "epoch": 0.053451675987613516,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.001827,
+      "loss": 1.3934,
+      "step": 609
+    },
+    {
+      "epoch": 0.053539445570516,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.00183,
+      "loss": 1.4198,
+      "step": 610
+    },
+    {
+      "epoch": 0.053627215153418485,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.001833,
+      "loss": 1.3836,
+      "step": 611
+    },
+    {
+      "epoch": 0.05371498473632098,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.001836,
+      "loss": 1.4215,
+      "step": 612
+    },
+    {
+      "epoch": 0.05380275431922346,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.001839,
+      "loss": 1.4335,
+      "step": 613
+    },
+    {
+      "epoch": 0.053890523902125946,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.001842,
+      "loss": 1.4146,
+      "step": 614
+    },
+    {
+      "epoch": 0.05397829348502843,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.001845,
+      "loss": 1.3886,
+      "step": 615
+    },
+    {
+      "epoch": 0.054066063067930915,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.001848,
+      "loss": 1.4687,
+      "step": 616
+    },
+    {
+      "epoch": 0.0541538326508334,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.001851,
+      "loss": 1.4417,
+      "step": 617
+    },
+    {
+      "epoch": 0.054241602233735885,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.001854,
+      "loss": 1.4007,
+      "step": 618
+    },
+    {
+      "epoch": 0.05432937181663837,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.001857,
+      "loss": 1.3945,
+      "step": 619
+    },
+    {
+      "epoch": 0.054417141399540854,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00186,
+      "loss": 1.4588,
+      "step": 620
+    },
+    {
+      "epoch": 0.05450491098244334,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.001863,
+      "loss": 1.3761,
+      "step": 621
+    },
+    {
+      "epoch": 0.05459268056534582,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001866,
+      "loss": 1.4146,
+      "step": 622
+    },
+    {
+      "epoch": 0.05468045014824831,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.001869,
+      "loss": 1.4337,
+      "step": 623
+    },
+    {
+      "epoch": 0.0547682197311508,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.001872,
+      "loss": 1.4874,
+      "step": 624
+    },
+    {
+      "epoch": 0.054855989314053284,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.001875,
+      "loss": 1.4406,
+      "step": 625
+    },
+    {
+      "epoch": 0.05494375889695577,
+      "grad_norm": 1.65625,
+      "learning_rate": 0.0018780000000000001,
+      "loss": 1.5081,
+      "step": 626
+    },
+    {
+      "epoch": 0.055031528479858254,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001881,
+      "loss": 1.4169,
+      "step": 627
+    },
+    {
+      "epoch": 0.05511929806276074,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.001884,
+      "loss": 1.4722,
+      "step": 628
+    },
+    {
+      "epoch": 0.05520706764566322,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.001887,
+      "loss": 1.4118,
+      "step": 629
+    },
+    {
+      "epoch": 0.05529483722856571,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00189,
+      "loss": 1.4665,
+      "step": 630
+    },
+    {
+      "epoch": 0.05538260681146819,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0018930000000000002,
+      "loss": 1.4355,
+      "step": 631
+    },
+    {
+      "epoch": 0.05547037639437068,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0018960000000000001,
+      "loss": 1.43,
+      "step": 632
+    },
+    {
+      "epoch": 0.05555814597727316,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.001899,
+      "loss": 1.3824,
+      "step": 633
+    },
+    {
+      "epoch": 0.055645915560175646,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.001902,
+      "loss": 1.4145,
+      "step": 634
+    },
+    {
+      "epoch": 0.05573368514307813,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.001905,
+      "loss": 1.4662,
+      "step": 635
+    },
+    {
+      "epoch": 0.05582145472598062,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.001908,
+      "loss": 1.4029,
+      "step": 636
+    },
+    {
+      "epoch": 0.05590922430888311,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0019110000000000002,
+      "loss": 1.385,
+      "step": 637
+    },
+    {
+      "epoch": 0.05599699389178559,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0019140000000000001,
+      "loss": 1.4255,
+      "step": 638
+    },
+    {
+      "epoch": 0.056084763474688076,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.001917,
+      "loss": 1.4397,
+      "step": 639
+    },
+    {
+      "epoch": 0.05617253305759056,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.00192,
+      "loss": 1.4382,
+      "step": 640
+    },
+    {
+      "epoch": 0.056260302640493046,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.001923,
+      "loss": 1.4125,
+      "step": 641
+    },
+    {
+      "epoch": 0.05634807222339553,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.001926,
+      "loss": 1.4425,
+      "step": 642
+    },
+    {
+      "epoch": 0.056435841806298015,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0019290000000000002,
+      "loss": 1.4217,
+      "step": 643
+    },
+    {
+      "epoch": 0.0565236113892005,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0019320000000000001,
+      "loss": 1.3992,
+      "step": 644
+    },
+    {
+      "epoch": 0.056611380972102984,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.001935,
+      "loss": 1.3734,
+      "step": 645
+    },
+    {
+      "epoch": 0.05669915055500547,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.001938,
+      "loss": 1.4064,
+      "step": 646
+    },
+    {
+      "epoch": 0.05678692013790796,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.001941,
+      "loss": 1.4006,
+      "step": 647
+    },
+    {
+      "epoch": 0.056874689720810445,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0019440000000000002,
+      "loss": 1.3646,
+      "step": 648
+    },
+    {
+      "epoch": 0.05696245930371293,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0019470000000000002,
+      "loss": 1.4171,
+      "step": 649
+    },
+    {
+      "epoch": 0.057050228886615414,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0019500000000000001,
+      "loss": 1.4012,
+      "step": 650
+    },
+    {
+      "epoch": 0.0571379984695179,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.001953,
+      "loss": 1.434,
+      "step": 651
+    },
+    {
+      "epoch": 0.057225768052420384,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0019560000000000003,
+      "loss": 1.4133,
+      "step": 652
+    },
+    {
+      "epoch": 0.05731353763532287,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.0019590000000000002,
+      "loss": 1.3874,
+      "step": 653
+    },
+    {
+      "epoch": 0.05740130721822535,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.001962,
+      "loss": 1.4022,
+      "step": 654
+    },
+    {
+      "epoch": 0.05748907680112784,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.001965,
+      "loss": 1.4111,
+      "step": 655
+    },
+    {
+      "epoch": 0.05757684638403032,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.001968,
+      "loss": 1.454,
+      "step": 656
+    },
+    {
+      "epoch": 0.05766461596693281,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.001971,
+      "loss": 1.4611,
+      "step": 657
+    },
+    {
+      "epoch": 0.05775238554983529,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.001974,
+      "loss": 1.4277,
+      "step": 658
+    },
+    {
+      "epoch": 0.05784015513273778,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.001977,
+      "loss": 1.4443,
+      "step": 659
+    },
+    {
+      "epoch": 0.05792792471564027,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.00198,
+      "loss": 1.3763,
+      "step": 660
+    },
+    {
+      "epoch": 0.05801569429854275,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.001983,
+      "loss": 1.3979,
+      "step": 661
+    },
+    {
+      "epoch": 0.05810346388144524,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0019860000000000004,
+      "loss": 1.4839,
+      "step": 662
+    },
+    {
+      "epoch": 0.05819123346434772,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0019890000000000003,
+      "loss": 1.3999,
+      "step": 663
+    },
+    {
+      "epoch": 0.05827900304725021,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0019920000000000003,
+      "loss": 1.3618,
+      "step": 664
+    },
+    {
+      "epoch": 0.05836677263015269,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0019950000000000002,
+      "loss": 1.4032,
+      "step": 665
+    },
+    {
+      "epoch": 0.058454542213055176,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.001998,
+      "loss": 1.4011,
+      "step": 666
+    },
+    {
+      "epoch": 0.05854231179595766,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.002001,
+      "loss": 1.4109,
+      "step": 667
+    },
+    {
+      "epoch": 0.058630081378860145,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.002004,
+      "loss": 1.4714,
+      "step": 668
+    },
+    {
+      "epoch": 0.05871785096176263,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.002007,
+      "loss": 1.3933,
+      "step": 669
+    },
+    {
+      "epoch": 0.058805620544665115,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.00201,
+      "loss": 1.3808,
+      "step": 670
+    },
+    {
+      "epoch": 0.058893390127567606,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.002013,
+      "loss": 1.3896,
+      "step": 671
+    },
+    {
+      "epoch": 0.05898115971047009,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.002016,
+      "loss": 1.391,
+      "step": 672
+    },
+    {
+      "epoch": 0.059068929293372575,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002019,
+      "loss": 1.378,
+      "step": 673
+    },
+    {
+      "epoch": 0.05915669887627506,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.0020220000000000004,
+      "loss": 1.4274,
+      "step": 674
+    },
+    {
+      "epoch": 0.059244468459177545,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0020250000000000003,
+      "loss": 1.4078,
+      "step": 675
+    },
+    {
+      "epoch": 0.05933223804208003,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0020280000000000003,
+      "loss": 1.4316,
+      "step": 676
+    },
+    {
+      "epoch": 0.059420007624982514,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0020310000000000003,
+      "loss": 1.4221,
+      "step": 677
+    },
+    {
+      "epoch": 0.059507777207885,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0020340000000000002,
+      "loss": 1.4176,
+      "step": 678
+    },
+    {
+      "epoch": 0.05959554679078748,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.002037,
+      "loss": 1.4378,
+      "step": 679
+    },
+    {
+      "epoch": 0.05968331637368997,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00204,
+      "loss": 1.4158,
+      "step": 680
+    },
+    {
+      "epoch": 0.05977108595659245,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.002043,
+      "loss": 1.4449,
+      "step": 681
+    },
+    {
+      "epoch": 0.059858855539494944,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.002046,
+      "loss": 1.4289,
+      "step": 682
+    },
+    {
+      "epoch": 0.05994662512239743,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.002049,
+      "loss": 1.4242,
+      "step": 683
+    },
+    {
+      "epoch": 0.060034394705299914,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.002052,
+      "loss": 1.3802,
+      "step": 684
+    },
+    {
+      "epoch": 0.0601221642882024,
+      "grad_norm": 1.703125,
+      "learning_rate": 0.0020550000000000004,
+      "loss": 1.4551,
+      "step": 685
+    },
+    {
+      "epoch": 0.06020993387110488,
+      "grad_norm": 2.828125,
+      "learning_rate": 0.0020580000000000004,
+      "loss": 1.9919,
+      "step": 686
+    },
+    {
+      "epoch": 0.06029770345400737,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0020610000000000003,
+      "loss": 1.469,
+      "step": 687
+    },
+    {
+      "epoch": 0.06038547303690985,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.002064,
+      "loss": 1.4125,
+      "step": 688
+    },
+    {
+      "epoch": 0.06047324261981234,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.002067,
+      "loss": 1.4094,
+      "step": 689
+    },
+    {
+      "epoch": 0.06056101220271482,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00207,
+      "loss": 1.3779,
+      "step": 690
+    },
+    {
+      "epoch": 0.060648781785617306,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0020729999999999998,
+      "loss": 1.4507,
+      "step": 691
+    },
+    {
+      "epoch": 0.06073655136851979,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.0020759999999999997,
+      "loss": 1.4563,
+      "step": 692
+    },
+    {
+      "epoch": 0.060824320951422275,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0020789999999999997,
+      "loss": 1.4445,
+      "step": 693
+    },
+    {
+      "epoch": 0.06091209053432477,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.002082,
+      "loss": 1.432,
+      "step": 694
+    },
+    {
+      "epoch": 0.06099986011722725,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.002085,
+      "loss": 1.3819,
+      "step": 695
+    },
+    {
+      "epoch": 0.061087629700129736,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.002088,
+      "loss": 1.4557,
+      "step": 696
+    },
+    {
+      "epoch": 0.06117539928303222,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.002091,
+      "loss": 1.442,
+      "step": 697
+    },
+    {
+      "epoch": 0.061263168865934706,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.002094,
+      "loss": 1.4193,
+      "step": 698
+    },
+    {
+      "epoch": 0.06135093844883719,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002097,
+      "loss": 1.4591,
+      "step": 699
+    },
+    {
+      "epoch": 0.061438708031739675,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.0021,
+      "loss": 1.4352,
+      "step": 700
+    },
+    {
+      "epoch": 0.06152647761464216,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002103,
+      "loss": 1.4033,
+      "step": 701
+    },
+    {
+      "epoch": 0.061614247197544644,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.002106,
+      "loss": 1.4272,
+      "step": 702
+    },
+    {
+      "epoch": 0.06170201678044713,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0021089999999999998,
+      "loss": 1.39,
+      "step": 703
+    },
+    {
+      "epoch": 0.061789786363349614,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0021119999999999997,
+      "loss": 1.3744,
+      "step": 704
+    },
+    {
+      "epoch": 0.0618775559462521,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.002115,
+      "loss": 1.4513,
+      "step": 705
+    },
+    {
+      "epoch": 0.06196532552915459,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.002118,
+      "loss": 1.428,
+      "step": 706
+    },
+    {
+      "epoch": 0.062053095112057075,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.002121,
+      "loss": 1.4149,
+      "step": 707
+    },
+    {
+      "epoch": 0.06214086469495956,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.002124,
+      "loss": 1.4335,
+      "step": 708
+    },
+    {
+      "epoch": 0.062228634277862044,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.002127,
+      "loss": 1.4268,
+      "step": 709
+    },
+    {
+      "epoch": 0.06231640386076453,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.00213,
+      "loss": 1.4299,
+      "step": 710
+    },
+    {
+      "epoch": 0.06240417344366701,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.002133,
+      "loss": 1.428,
+      "step": 711
+    },
+    {
+      "epoch": 0.0624919430265695,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.002136,
+      "loss": 1.377,
+      "step": 712
+    },
+    {
+      "epoch": 0.06257971260947198,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.002139,
+      "loss": 1.3745,
+      "step": 713
+    },
+    {
+      "epoch": 0.06266748219237447,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.002142,
+      "loss": 1.4129,
+      "step": 714
+    },
+    {
+      "epoch": 0.06275525177527695,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.0021449999999999998,
+      "loss": 1.4237,
+      "step": 715
+    },
+    {
+      "epoch": 0.06284302135817944,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.002148,
+      "loss": 1.4042,
+      "step": 716
+    },
+    {
+      "epoch": 0.06293079094108192,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.002151,
+      "loss": 1.401,
+      "step": 717
+    },
+    {
+      "epoch": 0.06301856052398441,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.002154,
+      "loss": 1.3501,
+      "step": 718
+    },
+    {
+      "epoch": 0.06310633010688689,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.002157,
+      "loss": 1.4434,
+      "step": 719
+    },
+    {
+      "epoch": 0.06319409968978938,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00216,
+      "loss": 1.4139,
+      "step": 720
+    },
+    {
+      "epoch": 0.06328186927269186,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.002163,
+      "loss": 1.3906,
+      "step": 721
+    },
+    {
+      "epoch": 0.06336963885559435,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002166,
+      "loss": 1.3987,
+      "step": 722
+    },
+    {
+      "epoch": 0.06345740843849684,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.002169,
+      "loss": 1.4363,
+      "step": 723
+    },
+    {
+      "epoch": 0.06354517802139932,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002172,
+      "loss": 1.3764,
+      "step": 724
+    },
+    {
+      "epoch": 0.06363294760430181,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.002175,
+      "loss": 1.4358,
+      "step": 725
+    },
+    {
+      "epoch": 0.06372071718720429,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.002178,
+      "loss": 1.3627,
+      "step": 726
+    },
+    {
+      "epoch": 0.06380848677010678,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0021809999999999998,
+      "loss": 1.4233,
+      "step": 727
+    },
+    {
+      "epoch": 0.06389625635300926,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.002184,
+      "loss": 1.3925,
+      "step": 728
+    },
+    {
+      "epoch": 0.06398402593591175,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.002187,
+      "loss": 1.4323,
+      "step": 729
+    },
+    {
+      "epoch": 0.06407179551881423,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.00219,
+      "loss": 1.4255,
+      "step": 730
+    },
+    {
+      "epoch": 0.06415956510171672,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.002193,
+      "loss": 1.4131,
+      "step": 731
+    },
+    {
+      "epoch": 0.0642473346846192,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.002196,
+      "loss": 1.415,
+      "step": 732
+    },
+    {
+      "epoch": 0.06433510426752169,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.002199,
+      "loss": 1.4095,
+      "step": 733
+    },
+    {
+      "epoch": 0.06442287385042417,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.002202,
+      "loss": 1.4265,
+      "step": 734
+    },
+    {
+      "epoch": 0.06451064343332666,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002205,
+      "loss": 1.3979,
+      "step": 735
+    },
+    {
+      "epoch": 0.06459841301622915,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.002208,
+      "loss": 1.3926,
+      "step": 736
+    },
+    {
+      "epoch": 0.06468618259913163,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.002211,
+      "loss": 1.3872,
+      "step": 737
+    },
+    {
+      "epoch": 0.06477395218203412,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.002214,
+      "loss": 1.4486,
+      "step": 738
+    },
+    {
+      "epoch": 0.0648617217649366,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.0022170000000000002,
+      "loss": 1.3547,
+      "step": 739
+    },
+    {
+      "epoch": 0.06494949134783909,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00222,
+      "loss": 1.4189,
+      "step": 740
+    },
+    {
+      "epoch": 0.06503726093074157,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002223,
+      "loss": 1.4166,
+      "step": 741
+    },
+    {
+      "epoch": 0.06512503051364406,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.002226,
+      "loss": 1.3736,
+      "step": 742
+    },
+    {
+      "epoch": 0.06521280009654654,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.002229,
+      "loss": 1.4195,
+      "step": 743
+    },
+    {
+      "epoch": 0.06530056967944903,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.002232,
+      "loss": 1.4215,
+      "step": 744
+    },
+    {
+      "epoch": 0.0653883392623515,
+      "grad_norm": 0.75,
+      "learning_rate": 0.002235,
+      "loss": 1.4432,
+      "step": 745
+    },
+    {
+      "epoch": 0.065476108845254,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.002238,
+      "loss": 1.372,
+      "step": 746
+    },
+    {
+      "epoch": 0.06556387842815649,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.002241,
+      "loss": 1.4103,
+      "step": 747
+    },
+    {
+      "epoch": 0.06565164801105897,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.002244,
+      "loss": 1.4322,
+      "step": 748
+    },
+    {
+      "epoch": 0.06573941759396146,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.002247,
+      "loss": 1.3891,
+      "step": 749
+    },
+    {
+      "epoch": 0.06582718717686394,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0022500000000000003,
+      "loss": 1.4126,
+      "step": 750
+    },
+    {
+      "epoch": 0.06591495675976643,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0022530000000000002,
+      "loss": 1.3994,
+      "step": 751
+    },
+    {
+      "epoch": 0.0660027263426689,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002256,
+      "loss": 1.3941,
+      "step": 752
+    },
+    {
+      "epoch": 0.0660904959255714,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002259,
+      "loss": 1.3345,
+      "step": 753
+    },
+    {
+      "epoch": 0.06617826550847387,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.002262,
+      "loss": 1.4136,
+      "step": 754
+    },
+    {
+      "epoch": 0.06626603509137637,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002265,
+      "loss": 1.3705,
+      "step": 755
+    },
+    {
+      "epoch": 0.06635380467427884,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.002268,
+      "loss": 1.4587,
+      "step": 756
+    },
+    {
+      "epoch": 0.06644157425718134,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.002271,
+      "loss": 1.432,
+      "step": 757
+    },
+    {
+      "epoch": 0.06652934384008383,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002274,
+      "loss": 1.3662,
+      "step": 758
+    },
+    {
+      "epoch": 0.0666171134229863,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.002277,
+      "loss": 1.387,
+      "step": 759
+    },
+    {
+      "epoch": 0.0667048830058888,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00228,
+      "loss": 1.3745,
+      "step": 760
+    },
+    {
+      "epoch": 0.06679265258879127,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.002283,
+      "loss": 1.4119,
+      "step": 761
+    },
+    {
+      "epoch": 0.06688042217169377,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0022860000000000003,
+      "loss": 1.3951,
+      "step": 762
+    },
+    {
+      "epoch": 0.06696819175459624,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0022890000000000002,
+      "loss": 1.4563,
+      "step": 763
+    },
+    {
+      "epoch": 0.06705596133749873,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.002292,
+      "loss": 1.3564,
+      "step": 764
+    },
+    {
+      "epoch": 0.06714373092040121,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002295,
+      "loss": 1.4115,
+      "step": 765
+    },
+    {
+      "epoch": 0.0672315005033037,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002298,
+      "loss": 1.4028,
+      "step": 766
+    },
+    {
+      "epoch": 0.06731927008620618,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002301,
+      "loss": 1.3478,
+      "step": 767
+    },
+    {
+      "epoch": 0.06740703966910867,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.002304,
+      "loss": 1.3808,
+      "step": 768
+    },
+    {
+      "epoch": 0.06749480925201115,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002307,
+      "loss": 1.3547,
+      "step": 769
+    },
+    {
+      "epoch": 0.06758257883491364,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00231,
+      "loss": 1.4304,
+      "step": 770
+    },
+    {
+      "epoch": 0.06767034841781613,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.002313,
+      "loss": 1.4404,
+      "step": 771
+    },
+    {
+      "epoch": 0.06775811800071861,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002316,
+      "loss": 1.4134,
+      "step": 772
+    },
+    {
+      "epoch": 0.0678458875836211,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0023190000000000003,
+      "loss": 1.4424,
+      "step": 773
+    },
+    {
+      "epoch": 0.06793365716652358,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0023220000000000003,
+      "loss": 1.4063,
+      "step": 774
+    },
+    {
+      "epoch": 0.06802142674942607,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0023250000000000002,
+      "loss": 1.444,
+      "step": 775
+    },
+    {
+      "epoch": 0.06810919633232855,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.002328,
+      "loss": 1.4046,
+      "step": 776
+    },
+    {
+      "epoch": 0.06819696591523104,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.002331,
+      "loss": 1.3615,
+      "step": 777
+    },
+    {
+      "epoch": 0.06828473549813352,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.002334,
+      "loss": 1.3941,
+      "step": 778
+    },
+    {
+      "epoch": 0.06837250508103601,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002337,
+      "loss": 1.4284,
+      "step": 779
+    },
+    {
+      "epoch": 0.06846027466393849,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.00234,
+      "loss": 1.4889,
+      "step": 780
+    },
+    {
+      "epoch": 0.06854804424684098,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.002343,
+      "loss": 1.4139,
+      "step": 781
+    },
+    {
+      "epoch": 0.06863581382974347,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002346,
+      "loss": 1.3882,
+      "step": 782
+    },
+    {
+      "epoch": 0.06872358341264595,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.002349,
+      "loss": 1.3539,
+      "step": 783
+    },
+    {
+      "epoch": 0.06881135299554844,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.002352,
+      "loss": 1.3903,
+      "step": 784
+    },
+    {
+      "epoch": 0.06889912257845092,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0023550000000000003,
+      "loss": 1.4078,
+      "step": 785
+    },
+    {
+      "epoch": 0.06898689216135341,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0023580000000000003,
+      "loss": 1.4124,
+      "step": 786
+    },
+    {
+      "epoch": 0.06907466174425589,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.0023610000000000003,
+      "loss": 1.404,
+      "step": 787
+    },
+    {
+      "epoch": 0.06916243132715838,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002364,
+      "loss": 1.4122,
+      "step": 788
+    },
+    {
+      "epoch": 0.06925020091006086,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.002367,
+      "loss": 1.3725,
+      "step": 789
+    },
+    {
+      "epoch": 0.06933797049296335,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00237,
+      "loss": 1.379,
+      "step": 790
+    },
+    {
+      "epoch": 0.06942574007586583,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.002373,
+      "loss": 1.3556,
+      "step": 791
+    },
+    {
+      "epoch": 0.06951350965876832,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.002376,
+      "loss": 1.361,
+      "step": 792
+    },
+    {
+      "epoch": 0.06960127924167081,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.002379,
+      "loss": 1.4277,
+      "step": 793
+    },
+    {
+      "epoch": 0.06968904882457329,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.002382,
+      "loss": 1.3974,
+      "step": 794
+    },
+    {
+      "epoch": 0.06977681840747578,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.002385,
+      "loss": 1.4055,
+      "step": 795
+    },
+    {
+      "epoch": 0.06986458799037826,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0023880000000000004,
+      "loss": 1.3921,
+      "step": 796
+    },
+    {
+      "epoch": 0.06995235757328075,
+      "grad_norm": 0.85546875,
+      "learning_rate": 0.0023910000000000003,
+      "loss": 1.4157,
+      "step": 797
+    },
+    {
+      "epoch": 0.07004012715618323,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0023940000000000003,
+      "loss": 1.3959,
+      "step": 798
+    },
+    {
+      "epoch": 0.07012789673908572,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0023970000000000003,
+      "loss": 1.3814,
+      "step": 799
+    },
+    {
+      "epoch": 0.0702156663219882,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.0024000000000000002,
+      "loss": 1.3708,
+      "step": 800
+    },
+    {
+      "epoch": 0.07030343590489069,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.002403,
+      "loss": 1.4212,
+      "step": 801
+    },
+    {
+      "epoch": 0.07039120548779317,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.002406,
+      "loss": 1.3823,
+      "step": 802
+    },
+    {
+      "epoch": 0.07047897507069566,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.002409,
+      "loss": 1.4311,
+      "step": 803
+    },
+    {
+      "epoch": 0.07056674465359813,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002412,
+      "loss": 1.3919,
+      "step": 804
+    },
+    {
+      "epoch": 0.07065451423650063,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.002415,
+      "loss": 1.4003,
+      "step": 805
+    },
+    {
+      "epoch": 0.07074228381940312,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002418,
+      "loss": 1.4254,
+      "step": 806
+    },
+    {
+      "epoch": 0.0708300534023056,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0024210000000000004,
+      "loss": 1.3777,
+      "step": 807
+    },
+    {
+      "epoch": 0.07091782298520809,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0024240000000000004,
+      "loss": 1.4008,
+      "step": 808
+    },
+    {
+      "epoch": 0.07100559256811056,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.0024270000000000003,
+      "loss": 1.3804,
+      "step": 809
+    },
+    {
+      "epoch": 0.07109336215101306,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0024300000000000003,
+      "loss": 1.4175,
+      "step": 810
+    },
+    {
+      "epoch": 0.07118113173391553,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0024330000000000003,
+      "loss": 1.398,
+      "step": 811
+    },
+    {
+      "epoch": 0.07126890131681803,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0024360000000000002,
+      "loss": 1.3582,
+      "step": 812
+    },
+    {
+      "epoch": 0.0713566708997205,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0024389999999999998,
+      "loss": 1.4143,
+      "step": 813
+    },
+    {
+      "epoch": 0.071444440482623,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0024419999999999997,
+      "loss": 1.4149,
+      "step": 814
+    },
+    {
+      "epoch": 0.07153221006552547,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0024449999999999997,
+      "loss": 1.3686,
+      "step": 815
+    },
+    {
+      "epoch": 0.07161997964842796,
+      "grad_norm": 3.828125,
+      "learning_rate": 0.002448,
+      "loss": 1.4106,
+      "step": 816
+    },
+    {
+      "epoch": 0.07170774923133046,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.002451,
+      "loss": 1.4604,
+      "step": 817
+    },
+    {
+      "epoch": 0.07179551881423293,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.002454,
+      "loss": 1.4668,
+      "step": 818
+    },
+    {
+      "epoch": 0.07188328839713543,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.002457,
+      "loss": 1.3775,
+      "step": 819
+    },
+    {
+      "epoch": 0.0719710579800379,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00246,
+      "loss": 1.3806,
+      "step": 820
+    },
+    {
+      "epoch": 0.0720588275629404,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.002463,
+      "loss": 1.4512,
+      "step": 821
+    },
+    {
+      "epoch": 0.07214659714584287,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.002466,
+      "loss": 1.4076,
+      "step": 822
+    },
+    {
+      "epoch": 0.07223436672874536,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.002469,
+      "loss": 1.3972,
+      "step": 823
+    },
+    {
+      "epoch": 0.07232213631164784,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.002472,
+      "loss": 1.4491,
+      "step": 824
+    },
+    {
+      "epoch": 0.07240990589455033,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0024749999999999998,
+      "loss": 1.4254,
+      "step": 825
+    },
+    {
+      "epoch": 0.07249767547745281,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0024779999999999997,
+      "loss": 1.3944,
+      "step": 826
+    },
+    {
+      "epoch": 0.0725854450603553,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.002481,
+      "loss": 1.4159,
+      "step": 827
+    },
+    {
+      "epoch": 0.0726732146432578,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.002484,
+      "loss": 1.3851,
+      "step": 828
+    },
+    {
+      "epoch": 0.07276098422616027,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.002487,
+      "loss": 1.3845,
+      "step": 829
+    },
+    {
+      "epoch": 0.07284875380906276,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.00249,
+      "loss": 1.4024,
+      "step": 830
+    },
+    {
+      "epoch": 0.07293652339196524,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002493,
+      "loss": 1.3894,
+      "step": 831
+    },
+    {
+      "epoch": 0.07302429297486773,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.002496,
+      "loss": 1.4546,
+      "step": 832
+    },
+    {
+      "epoch": 0.07311206255777021,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002499,
+      "loss": 1.4338,
+      "step": 833
+    },
+    {
+      "epoch": 0.0731998321406727,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.002502,
+      "loss": 1.4251,
+      "step": 834
+    },
+    {
+      "epoch": 0.07328760172357518,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.002505,
+      "loss": 1.4107,
+      "step": 835
+    },
+    {
+      "epoch": 0.07337537130647767,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.002508,
+      "loss": 1.3467,
+      "step": 836
+    },
+    {
+      "epoch": 0.07346314088938015,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0025109999999999998,
+      "loss": 1.3859,
+      "step": 837
+    },
+    {
+      "epoch": 0.07355091047228264,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0025139999999999997,
+      "loss": 1.3957,
+      "step": 838
+    },
+    {
+      "epoch": 0.07363868005518512,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002517,
+      "loss": 1.3665,
+      "step": 839
+    },
+    {
+      "epoch": 0.07372644963808761,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.00252,
+      "loss": 1.3961,
+      "step": 840
+    },
+    {
+      "epoch": 0.0738142192209901,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.002523,
+      "loss": 1.3849,
+      "step": 841
+    },
+    {
+      "epoch": 0.07390198880389258,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.002526,
+      "loss": 1.352,
+      "step": 842
+    },
+    {
+      "epoch": 0.07398975838679507,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002529,
+      "loss": 1.4165,
+      "step": 843
+    },
+    {
+      "epoch": 0.07407752796969755,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002532,
+      "loss": 1.4083,
+      "step": 844
+    },
+    {
+      "epoch": 0.07416529755260004,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.002535,
+      "loss": 1.4197,
+      "step": 845
+    },
+    {
+      "epoch": 0.07425306713550252,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.002538,
+      "loss": 1.4065,
+      "step": 846
+    },
+    {
+      "epoch": 0.07434083671840501,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.002541,
+      "loss": 1.3912,
+      "step": 847
+    },
+    {
+      "epoch": 0.07442860630130749,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.002544,
+      "loss": 1.4331,
+      "step": 848
+    },
+    {
+      "epoch": 0.07451637588420998,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002547,
+      "loss": 1.3746,
+      "step": 849
+    },
+    {
+      "epoch": 0.07460414546711246,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.00255,
+      "loss": 1.3978,
+      "step": 850
+    },
+    {
+      "epoch": 0.07469191505001495,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.002553,
+      "loss": 1.4342,
+      "step": 851
+    },
+    {
+      "epoch": 0.07477968463291744,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.002556,
+      "loss": 1.369,
+      "step": 852
+    },
+    {
+      "epoch": 0.07486745421581992,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.002559,
+      "loss": 1.4062,
+      "step": 853
+    },
+    {
+      "epoch": 0.07495522379872241,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.002562,
+      "loss": 1.3848,
+      "step": 854
+    },
+    {
+      "epoch": 0.07504299338162489,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.002565,
+      "loss": 1.3805,
+      "step": 855
+    },
+    {
+      "epoch": 0.07513076296452738,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.002568,
+      "loss": 1.3979,
+      "step": 856
+    },
+    {
+      "epoch": 0.07521853254742986,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.002571,
+      "loss": 1.3539,
+      "step": 857
+    },
+    {
+      "epoch": 0.07530630213033235,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.002574,
+      "loss": 1.3399,
+      "step": 858
+    },
+    {
+      "epoch": 0.07539407171323483,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.002577,
+      "loss": 1.4018,
+      "step": 859
+    },
+    {
+      "epoch": 0.07548184129613732,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.00258,
+      "loss": 1.4543,
+      "step": 860
+    },
+    {
+      "epoch": 0.0755696108790398,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0025830000000000002,
+      "loss": 1.4085,
+      "step": 861
+    },
+    {
+      "epoch": 0.07565738046194229,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.002586,
+      "loss": 1.4448,
+      "step": 862
+    },
+    {
+      "epoch": 0.07574515004484478,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.002589,
+      "loss": 1.4506,
+      "step": 863
+    },
+    {
+      "epoch": 0.07583291962774726,
+      "grad_norm": 0.25,
+      "learning_rate": 0.002592,
+      "loss": 1.4231,
+      "step": 864
+    },
+    {
+      "epoch": 0.07592068921064975,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.002595,
+      "loss": 1.3962,
+      "step": 865
+    },
+    {
+      "epoch": 0.07600845879355222,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.002598,
+      "loss": 1.3933,
+      "step": 866
+    },
+    {
+      "epoch": 0.07609622837645472,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.002601,
+      "loss": 1.4323,
+      "step": 867
+    },
+    {
+      "epoch": 0.0761839979593572,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.002604,
+      "loss": 1.4321,
+      "step": 868
+    },
+    {
+      "epoch": 0.07627176754225969,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.002607,
+      "loss": 1.403,
+      "step": 869
+    },
+    {
+      "epoch": 0.07635953712516216,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00261,
+      "loss": 1.3849,
+      "step": 870
+    },
+    {
+      "epoch": 0.07644730670806466,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.002613,
+      "loss": 1.4362,
+      "step": 871
+    },
+    {
+      "epoch": 0.07653507629096713,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.002616,
+      "loss": 1.3857,
+      "step": 872
+    },
+    {
+      "epoch": 0.07662284587386962,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.0026190000000000002,
+      "loss": 1.424,
+      "step": 873
+    },
+    {
+      "epoch": 0.0767106154567721,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.002622,
+      "loss": 1.4226,
+      "step": 874
+    },
+    {
+      "epoch": 0.0767983850396746,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.002625,
+      "loss": 1.4058,
+      "step": 875
+    },
+    {
+      "epoch": 0.07688615462257709,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.002628,
+      "loss": 1.3957,
+      "step": 876
+    },
+    {
+      "epoch": 0.07697392420547956,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.002631,
+      "loss": 1.4533,
+      "step": 877
+    },
+    {
+      "epoch": 0.07706169378838205,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.002634,
+      "loss": 1.3528,
+      "step": 878
+    },
+    {
+      "epoch": 0.07714946337128453,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.002637,
+      "loss": 1.3916,
+      "step": 879
+    },
+    {
+      "epoch": 0.07723723295418702,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.00264,
+      "loss": 1.449,
+      "step": 880
+    },
+    {
+      "epoch": 0.0773250025370895,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.002643,
+      "loss": 1.4141,
+      "step": 881
+    },
+    {
+      "epoch": 0.077412772119992,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.002646,
+      "loss": 1.4228,
+      "step": 882
+    },
+    {
+      "epoch": 0.07750054170289447,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.002649,
+      "loss": 1.3928,
+      "step": 883
+    },
+    {
+      "epoch": 0.07758831128579696,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0026520000000000003,
+      "loss": 1.3593,
+      "step": 884
+    },
+    {
+      "epoch": 0.07767608086869944,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0026550000000000002,
+      "loss": 1.3993,
+      "step": 885
+    },
+    {
+      "epoch": 0.07776385045160193,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.002658,
+      "loss": 1.3602,
+      "step": 886
+    },
+    {
+      "epoch": 0.07785162003450442,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.002661,
+      "loss": 1.4019,
+      "step": 887
+    },
+    {
+      "epoch": 0.0779393896174069,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.002664,
+      "loss": 1.3701,
+      "step": 888
+    },
+    {
+      "epoch": 0.07802715920030939,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.002667,
+      "loss": 1.4444,
+      "step": 889
+    },
+    {
+      "epoch": 0.07811492878321187,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00267,
+      "loss": 1.401,
+      "step": 890
+    },
+    {
+      "epoch": 0.07820269836611436,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002673,
+      "loss": 1.409,
+      "step": 891
+    },
+    {
+      "epoch": 0.07829046794901684,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.002676,
+      "loss": 1.3678,
+      "step": 892
+    },
+    {
+      "epoch": 0.07837823753191933,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.002679,
+      "loss": 1.4032,
+      "step": 893
+    },
+    {
+      "epoch": 0.07846600711482181,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.002682,
+      "loss": 1.3723,
+      "step": 894
+    },
+    {
+      "epoch": 0.0785537766977243,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0026850000000000003,
+      "loss": 1.3817,
+      "step": 895
+    },
+    {
+      "epoch": 0.07864154628062678,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0026880000000000003,
+      "loss": 1.4086,
+      "step": 896
+    },
+    {
+      "epoch": 0.07872931586352927,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0026910000000000002,
+      "loss": 1.3719,
+      "step": 897
+    },
+    {
+      "epoch": 0.07881708544643176,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002694,
+      "loss": 1.3896,
+      "step": 898
+    },
+    {
+      "epoch": 0.07890485502933424,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.002697,
+      "loss": 1.4354,
+      "step": 899
+    },
+    {
+      "epoch": 0.07899262461223673,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0027,
+      "loss": 1.4313,
+      "step": 900
+    },
+    {
+      "epoch": 0.07908039419513921,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002703,
+      "loss": 1.4053,
+      "step": 901
+    },
+    {
+      "epoch": 0.0791681637780417,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.002706,
+      "loss": 1.4214,
+      "step": 902
+    },
+    {
+      "epoch": 0.07925593336094418,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.002709,
+      "loss": 1.3898,
+      "step": 903
+    },
+    {
+      "epoch": 0.07934370294384667,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.002712,
+      "loss": 1.4251,
+      "step": 904
+    },
+    {
+      "epoch": 0.07943147252674915,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.002715,
+      "loss": 1.3688,
+      "step": 905
+    },
+    {
+      "epoch": 0.07951924210965164,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.002718,
+      "loss": 1.377,
+      "step": 906
+    },
+    {
+      "epoch": 0.07960701169255412,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0027210000000000003,
+      "loss": 1.4229,
+      "step": 907
+    },
+    {
+      "epoch": 0.07969478127545661,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0027240000000000003,
+      "loss": 1.4327,
+      "step": 908
+    },
+    {
+      "epoch": 0.07978255085835909,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0027270000000000003,
+      "loss": 1.3982,
+      "step": 909
+    },
+    {
+      "epoch": 0.07987032044126158,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0027300000000000002,
+      "loss": 1.392,
+      "step": 910
+    },
+    {
+      "epoch": 0.07995809002416407,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.002733,
+      "loss": 1.4287,
+      "step": 911
+    },
+    {
+      "epoch": 0.08004585960706655,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.002736,
+      "loss": 1.3982,
+      "step": 912
+    },
+    {
+      "epoch": 0.08013362918996904,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.002739,
+      "loss": 1.3988,
+      "step": 913
+    },
+    {
+      "epoch": 0.08022139877287152,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.002742,
+      "loss": 1.3755,
+      "step": 914
+    },
+    {
+      "epoch": 0.08030916835577401,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.002745,
+      "loss": 1.4671,
+      "step": 915
+    },
+    {
+      "epoch": 0.08039693793867649,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.002748,
+      "loss": 1.4094,
+      "step": 916
+    },
+    {
+      "epoch": 0.08048470752157898,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.002751,
+      "loss": 1.4038,
+      "step": 917
+    },
+    {
+      "epoch": 0.08057247710448145,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0027540000000000004,
+      "loss": 1.4396,
+      "step": 918
+    },
+    {
+      "epoch": 0.08066024668738395,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0027570000000000003,
+      "loss": 1.3817,
+      "step": 919
+    },
+    {
+      "epoch": 0.08074801627028642,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0027600000000000003,
+      "loss": 1.382,
+      "step": 920
+    },
+    {
+      "epoch": 0.08083578585318892,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0027630000000000003,
+      "loss": 1.4015,
+      "step": 921
+    },
+    {
+      "epoch": 0.08092355543609141,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0027660000000000002,
+      "loss": 1.4056,
+      "step": 922
+    },
+    {
+      "epoch": 0.08101132501899388,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.002769,
+      "loss": 1.4484,
+      "step": 923
+    },
+    {
+      "epoch": 0.08109909460189638,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.002772,
+      "loss": 1.4534,
+      "step": 924
+    },
+    {
+      "epoch": 0.08118686418479885,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002775,
+      "loss": 1.3962,
+      "step": 925
+    },
+    {
+      "epoch": 0.08127463376770135,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.002778,
+      "loss": 1.3683,
+      "step": 926
+    },
+    {
+      "epoch": 0.08136240335060382,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002781,
+      "loss": 1.3443,
+      "step": 927
+    },
+    {
+      "epoch": 0.08145017293350632,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002784,
+      "loss": 1.3491,
+      "step": 928
+    },
+    {
+      "epoch": 0.08153794251640879,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0027870000000000004,
+      "loss": 1.3446,
+      "step": 929
+    },
+    {
+      "epoch": 0.08162571209931128,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0027900000000000004,
+      "loss": 1.3753,
+      "step": 930
+    },
+    {
+      "epoch": 0.08171348168221376,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.0027930000000000003,
+      "loss": 1.4317,
+      "step": 931
+    },
+    {
+      "epoch": 0.08180125126511625,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0027960000000000003,
+      "loss": 1.4003,
+      "step": 932
+    },
+    {
+      "epoch": 0.08188902084801875,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0027990000000000003,
+      "loss": 1.3705,
+      "step": 933
+    },
+    {
+      "epoch": 0.08197679043092122,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0028020000000000002,
+      "loss": 1.4637,
+      "step": 934
+    },
+    {
+      "epoch": 0.08206456001382371,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.002805,
+      "loss": 1.3788,
+      "step": 935
+    },
+    {
+      "epoch": 0.08215232959672619,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.002808,
+      "loss": 1.4177,
+      "step": 936
+    },
+    {
+      "epoch": 0.08224009917962868,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.002811,
+      "loss": 1.365,
+      "step": 937
+    },
+    {
+      "epoch": 0.08232786876253116,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.002814,
+      "loss": 1.403,
+      "step": 938
+    },
+    {
+      "epoch": 0.08241563834543365,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002817,
+      "loss": 1.4323,
+      "step": 939
+    },
+    {
+      "epoch": 0.08250340792833613,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00282,
+      "loss": 1.3889,
+      "step": 940
+    },
+    {
+      "epoch": 0.08259117751123862,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.002823,
+      "loss": 1.3413,
+      "step": 941
+    },
+    {
+      "epoch": 0.0826789470941411,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.002826,
+      "loss": 1.4498,
+      "step": 942
+    },
+    {
+      "epoch": 0.08276671667704359,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002829,
+      "loss": 1.4469,
+      "step": 943
+    },
+    {
+      "epoch": 0.08285448625994607,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.002832,
+      "loss": 1.3705,
+      "step": 944
+    },
+    {
+      "epoch": 0.08294225584284856,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.002835,
+      "loss": 1.3415,
+      "step": 945
+    },
+    {
+      "epoch": 0.08303002542575105,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.002838,
+      "loss": 1.4412,
+      "step": 946
+    },
+    {
+      "epoch": 0.08311779500865353,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0028409999999999998,
+      "loss": 1.4073,
+      "step": 947
+    },
+    {
+      "epoch": 0.08320556459155602,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0028439999999999997,
+      "loss": 1.4343,
+      "step": 948
+    },
+    {
+      "epoch": 0.0832933341744585,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.002847,
+      "loss": 1.427,
+      "step": 949
+    },
+    {
+      "epoch": 0.08338110375736099,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.00285,
+      "loss": 1.3348,
+      "step": 950
+    },
+    {
+      "epoch": 0.08346887334026347,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.002853,
+      "loss": 1.3768,
+      "step": 951
+    },
+    {
+      "epoch": 0.08355664292316596,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.002856,
+      "loss": 1.4057,
+      "step": 952
+    },
+    {
+      "epoch": 0.08364441250606844,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.002859,
+      "loss": 1.4168,
+      "step": 953
+    },
+    {
+      "epoch": 0.08373218208897093,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.002862,
+      "loss": 1.3865,
+      "step": 954
+    },
+    {
+      "epoch": 0.08381995167187341,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.002865,
+      "loss": 1.447,
+      "step": 955
+    },
+    {
+      "epoch": 0.0839077212547759,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.002868,
+      "loss": 1.3856,
+      "step": 956
+    },
+    {
+      "epoch": 0.08399549083767839,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.002871,
+      "loss": 1.4196,
+      "step": 957
+    },
+    {
+      "epoch": 0.08408326042058087,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.002874,
+      "loss": 1.3786,
+      "step": 958
+    },
+    {
+      "epoch": 0.08417103000348336,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002877,
+      "loss": 1.3898,
+      "step": 959
+    },
+    {
+      "epoch": 0.08425879958638584,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0028799999999999997,
+      "loss": 1.4289,
+      "step": 960
+    },
+    {
+      "epoch": 0.08434656916928833,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002883,
+      "loss": 1.3777,
+      "step": 961
+    },
+    {
+      "epoch": 0.08443433875219081,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.002886,
+      "loss": 1.4581,
+      "step": 962
+    },
+    {
+      "epoch": 0.0845221083350933,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.002889,
+      "loss": 1.3526,
+      "step": 963
+    },
+    {
+      "epoch": 0.08460987791799578,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.002892,
+      "loss": 1.404,
+      "step": 964
+    },
+    {
+      "epoch": 0.08469764750089827,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.002895,
+      "loss": 1.3726,
+      "step": 965
+    },
+    {
+      "epoch": 0.08478541708380075,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.002898,
+      "loss": 1.423,
+      "step": 966
+    },
+    {
+      "epoch": 0.08487318666670324,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.002901,
+      "loss": 1.4026,
+      "step": 967
+    },
+    {
+      "epoch": 0.08496095624960573,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.002904,
+      "loss": 1.4215,
+      "step": 968
+    },
+    {
+      "epoch": 0.0850487258325082,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.002907,
+      "loss": 1.3334,
+      "step": 969
+    },
+    {
+      "epoch": 0.0851364954154107,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00291,
+      "loss": 1.4332,
+      "step": 970
+    },
+    {
+      "epoch": 0.08522426499831318,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002913,
+      "loss": 1.3709,
+      "step": 971
+    },
+    {
+      "epoch": 0.08531203458121567,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.002916,
+      "loss": 1.378,
+      "step": 972
+    },
+    {
+      "epoch": 0.08539980416411815,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.002919,
+      "loss": 1.3483,
+      "step": 973
+    },
+    {
+      "epoch": 0.08548757374702064,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.002922,
+      "loss": 1.3419,
+      "step": 974
+    },
+    {
+      "epoch": 0.08557534332992311,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002925,
+      "loss": 1.4303,
+      "step": 975
+    },
+    {
+      "epoch": 0.0856631129128256,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002928,
+      "loss": 1.3703,
+      "step": 976
+    },
+    {
+      "epoch": 0.08575088249572808,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002931,
+      "loss": 1.3757,
+      "step": 977
+    },
+    {
+      "epoch": 0.08583865207863058,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.002934,
+      "loss": 1.4041,
+      "step": 978
+    },
+    {
+      "epoch": 0.08592642166153307,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002937,
+      "loss": 1.3838,
+      "step": 979
+    },
+    {
+      "epoch": 0.08601419124443554,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.00294,
+      "loss": 1.3663,
+      "step": 980
+    },
+    {
+      "epoch": 0.08610196082733804,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.002943,
+      "loss": 1.3884,
+      "step": 981
+    },
+    {
+      "epoch": 0.08618973041024051,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002946,
+      "loss": 1.4175,
+      "step": 982
+    },
+    {
+      "epoch": 0.086277499993143,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0029490000000000002,
+      "loss": 1.3364,
+      "step": 983
+    },
+    {
+      "epoch": 0.08636526957604548,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.002952,
+      "loss": 1.3714,
+      "step": 984
+    },
+    {
+      "epoch": 0.08645303915894798,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.002955,
+      "loss": 1.3786,
+      "step": 985
+    },
+    {
+      "epoch": 0.08654080874185045,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.002958,
+      "loss": 1.4111,
+      "step": 986
+    },
+    {
+      "epoch": 0.08662857832475294,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002961,
+      "loss": 1.4381,
+      "step": 987
+    },
+    {
+      "epoch": 0.08671634790765542,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.002964,
+      "loss": 1.3721,
+      "step": 988
+    },
+    {
+      "epoch": 0.08680411749055791,
+      "grad_norm": 0.25,
+      "learning_rate": 0.002967,
+      "loss": 1.4126,
+      "step": 989
+    },
+    {
+      "epoch": 0.08689188707346039,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00297,
+      "loss": 1.4663,
+      "step": 990
+    },
+    {
+      "epoch": 0.08697965665636288,
+      "grad_norm": 0.25,
+      "learning_rate": 0.002973,
+      "loss": 1.382,
+      "step": 991
+    },
+    {
+      "epoch": 0.08706742623926537,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.002976,
+      "loss": 1.4019,
+      "step": 992
+    },
+    {
+      "epoch": 0.08715519582216785,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.002979,
+      "loss": 1.462,
+      "step": 993
+    },
+    {
+      "epoch": 0.08724296540507034,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.002982,
+      "loss": 1.421,
+      "step": 994
+    },
+    {
+      "epoch": 0.08733073498797282,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0029850000000000002,
+      "loss": 1.3908,
+      "step": 995
+    },
+    {
+      "epoch": 0.08741850457087531,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002988,
+      "loss": 1.3896,
+      "step": 996
+    },
+    {
+      "epoch": 0.08750627415377779,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.002991,
+      "loss": 1.4022,
+      "step": 997
+    },
+    {
+      "epoch": 0.08759404373668028,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.002994,
+      "loss": 1.3985,
+      "step": 998
+    },
+    {
+      "epoch": 0.08768181331958276,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.002997,
+      "loss": 1.3715,
+      "step": 999
+    },
+    {
+      "epoch": 0.08776958290248525,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.003,
+      "loss": 1.3632,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08776958290248525,
+      "eval_loss": 1.3825047016143799,
+      "eval_runtime": 437.7802,
+      "eval_samples_per_second": 33.67,
+      "eval_steps_per_second": 8.417,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08785735248538773,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0029999999333801706,
+      "loss": 1.4102,
+      "step": 1001
+    },
+    {
+      "epoch": 0.08794512206829022,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0029999997335206897,
+      "loss": 1.4553,
+      "step": 1002
+    },
+    {
+      "epoch": 0.08803289165119271,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0029999994004215767,
+      "loss": 1.3328,
+      "step": 1003
+    },
+    {
+      "epoch": 0.08812066123409519,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.002999998934082865,
+      "loss": 1.4013,
+      "step": 1004
+    },
+    {
+      "epoch": 0.08820843081699768,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0029999983345045996,
+      "loss": 1.3431,
+      "step": 1005
+    },
+    {
+      "epoch": 0.08829620039990016,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0029999976016868404,
+      "loss": 1.3888,
+      "step": 1006
+    },
+    {
+      "epoch": 0.08838396998280265,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0029999967356296596,
+      "loss": 1.4219,
+      "step": 1007
+    },
+    {
+      "epoch": 0.08847173956570513,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.002999995736333143,
+      "loss": 1.4155,
+      "step": 1008
+    },
+    {
+      "epoch": 0.08855950914860762,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0029999946037973886,
+      "loss": 1.3674,
+      "step": 1009
+    },
+    {
+      "epoch": 0.0886472787315101,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0029999933380225087,
+      "loss": 1.3753,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08873504831441259,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0029999919390086274,
+      "loss": 1.3757,
+      "step": 1011
+    },
+    {
+      "epoch": 0.08882281789731507,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0029999904067558834,
+      "loss": 1.3702,
+      "step": 1012
+    },
+    {
+      "epoch": 0.08891058748021756,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.002999988741264429,
+      "loss": 1.3923,
+      "step": 1013
+    },
+    {
+      "epoch": 0.08899835706312005,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0029999869425344264,
+      "loss": 1.37,
+      "step": 1014
+    },
+    {
+      "epoch": 0.08908612664602253,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0029999850105660547,
+      "loss": 1.4014,
+      "step": 1015
+    },
+    {
+      "epoch": 0.08917389622892502,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0029999829453595038,
+      "loss": 1.3861,
+      "step": 1016
+    },
+    {
+      "epoch": 0.0892616658118275,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.002999980746914978,
+      "loss": 1.4112,
+      "step": 1017
+    },
+    {
+      "epoch": 0.08934943539472999,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002999978415232694,
+      "loss": 1.3873,
+      "step": 1018
+    },
+    {
+      "epoch": 0.08943720497763247,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.002999975950312882,
+      "loss": 1.404,
+      "step": 1019
+    },
+    {
+      "epoch": 0.08952497456053496,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.002999973352155785,
+      "loss": 1.4123,
+      "step": 1020
+    },
+    {
+      "epoch": 0.08961274414343744,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00299997062076166,
+      "loss": 1.4009,
+      "step": 1021
+    },
+    {
+      "epoch": 0.08970051372633993,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0029999677561307766,
+      "loss": 1.352,
+      "step": 1022
+    },
+    {
+      "epoch": 0.0897882833092424,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.002999964758263417,
+      "loss": 1.4426,
+      "step": 1023
+    },
+    {
+      "epoch": 0.0898760528921449,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.002999961627159877,
+      "loss": 1.4257,
+      "step": 1024
+    },
+    {
+      "epoch": 0.08996382247504738,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.002999958362820466,
+      "loss": 1.4002,
+      "step": 1025
+    },
+    {
+      "epoch": 0.09005159205794987,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.0029999549652455063,
+      "loss": 1.413,
+      "step": 1026
+    },
+    {
+      "epoch": 0.09013936164085236,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002999951434435333,
+      "loss": 1.3596,
+      "step": 1027
+    },
+    {
+      "epoch": 0.09022713122375484,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0029999477703902947,
+      "loss": 1.4353,
+      "step": 1028
+    },
+    {
+      "epoch": 0.09031490080665733,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0029999439731107527,
+      "loss": 1.3752,
+      "step": 1029
+    },
+    {
+      "epoch": 0.0904026703895598,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002999940042597082,
+      "loss": 1.386,
+      "step": 1030
+    },
+    {
+      "epoch": 0.0904904399724623,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0029999359788496715,
+      "loss": 1.3796,
+      "step": 1031
+    },
+    {
+      "epoch": 0.09057820955536477,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00299993178186892,
+      "loss": 1.3642,
+      "step": 1032
+    },
+    {
+      "epoch": 0.09066597913826727,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0029999274516552435,
+      "loss": 1.3919,
+      "step": 1033
+    },
+    {
+      "epoch": 0.09075374872116974,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002999922988209069,
+      "loss": 1.3464,
+      "step": 1034
+    },
+    {
+      "epoch": 0.09084151830407224,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.002999918391530837,
+      "loss": 1.3758,
+      "step": 1035
+    },
+    {
+      "epoch": 0.09092928788697471,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.002999913661621001,
+      "loss": 1.408,
+      "step": 1036
+    },
+    {
+      "epoch": 0.0910170574698772,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0029999087984800277,
+      "loss": 1.3728,
+      "step": 1037
+    },
+    {
+      "epoch": 0.0911048270527797,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002999903802108398,
+      "loss": 1.3839,
+      "step": 1038
+    },
+    {
+      "epoch": 0.09119259663568217,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0029998986725066033,
+      "loss": 1.3796,
+      "step": 1039
+    },
+    {
+      "epoch": 0.09128036621858467,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002999893409675152,
+      "loss": 1.3723,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09136813580148714,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0029998880136145613,
+      "loss": 1.3627,
+      "step": 1041
+    },
+    {
+      "epoch": 0.09145590538438964,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.002999882484325366,
+      "loss": 1.4139,
+      "step": 1042
+    },
+    {
+      "epoch": 0.09154367496729211,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.00299987682180811,
+      "loss": 1.4144,
+      "step": 1043
+    },
+    {
+      "epoch": 0.0916314445501946,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002999871026063353,
+      "loss": 1.335,
+      "step": 1044
+    },
+    {
+      "epoch": 0.09171921413309708,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.002999865097091667,
+      "loss": 1.36,
+      "step": 1045
+    },
+    {
+      "epoch": 0.09180698371599957,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0029998590348936376,
+      "loss": 1.3949,
+      "step": 1046
+    },
+    {
+      "epoch": 0.09189475329890205,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.0029998528394698616,
+      "loss": 1.3764,
+      "step": 1047
+    },
+    {
+      "epoch": 0.09198252288180454,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002999846510820952,
+      "loss": 1.3687,
+      "step": 1048
+    },
+    {
+      "epoch": 0.09207029246470703,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0029998400489475332,
+      "loss": 1.3591,
+      "step": 1049
+    },
+    {
+      "epoch": 0.09215806204760951,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0029998334538502416,
+      "loss": 1.3819,
+      "step": 1050
+    },
+    {
+      "epoch": 0.092245831630512,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0029998267255297298,
+      "loss": 1.3957,
+      "step": 1051
+    },
+    {
+      "epoch": 0.09233360121341448,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029998198639866612,
+      "loss": 1.3797,
+      "step": 1052
+    },
+    {
+      "epoch": 0.09242137079631697,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0029998128692217125,
+      "loss": 1.4328,
+      "step": 1053
+    },
+    {
+      "epoch": 0.09250914037921945,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0029998057412355754,
+      "loss": 1.425,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09259690996212194,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002999798480028952,
+      "loss": 1.4552,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09268467954502442,
+      "grad_norm": 4.5625,
+      "learning_rate": 0.0029997910856025597,
+      "loss": 1.4378,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09277244912792691,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0029997835579571278,
+      "loss": 1.397,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09286021871082939,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0029997758970933997,
+      "loss": 1.3356,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09294798829373188,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0029997681030121313,
+      "loss": 1.422,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09303575787663436,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.002999760175714092,
+      "loss": 1.4501,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09312352745953685,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0029997521152000644,
+      "loss": 1.4169,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09321129704243934,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0029997439214708432,
+      "loss": 1.3878,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09329906662534182,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002999735594527238,
+      "loss": 1.406,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09338683620824431,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.00299972713437007,
+      "loss": 1.3747,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09347460579114679,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0029997185410001745,
+      "loss": 1.3693,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09356237537404928,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0029997098144183996,
+      "loss": 1.4782,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09365014495695176,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.0029997009546256065,
+      "loss": 1.3834,
+      "step": 1067
+    },
+    {
+      "epoch": 0.09373791453985425,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00299969196162267,
+      "loss": 1.3232,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09382568412275673,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0029996828354104765,
+      "loss": 1.4099,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09391345370565922,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.002999673575989929,
+      "loss": 1.3691,
+      "step": 1070
+    },
+    {
+      "epoch": 0.0940012232885617,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.002999664183361939,
+      "loss": 1.4001,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09408899287146419,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002999654657527434,
+      "loss": 1.3827,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09417676245436668,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.002999644998487355,
+      "loss": 1.3673,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09426453203726916,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0029996352062426554,
+      "loss": 1.3937,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09435230162017165,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0029996252807943007,
+      "loss": 1.359,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09444007120307413,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0029996152221432714,
+      "loss": 1.4212,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09452784078597662,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0029996050302905594,
+      "loss": 1.3969,
+      "step": 1077
+    },
+    {
+      "epoch": 0.0946156103688791,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002999594705237171,
+      "loss": 1.4725,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09470337995178159,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.002999584246984126,
+      "loss": 1.3726,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09479114953468407,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.002999573655532455,
+      "loss": 1.3471,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09487891911758656,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0029995629308832046,
+      "loss": 1.3358,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09496668870048904,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002999552073037433,
+      "loss": 1.3811,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09505445828339153,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.002999541081996211,
+      "loss": 1.4059,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09514222786629402,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0029995299577606246,
+      "loss": 1.411,
+      "step": 1084
+    },
+    {
+      "epoch": 0.0952299974491965,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0029995187003317714,
+      "loss": 1.344,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09531776703209899,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0029995073097107616,
+      "loss": 1.4106,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09540553661500147,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0029994957858987207,
+      "loss": 1.41,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09549330619790396,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0029994841288967854,
+      "loss": 1.3824,
+      "step": 1088
+    },
+    {
+      "epoch": 0.09558107578080643,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.002999472338706106,
+      "loss": 1.3149,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09566884536370893,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.002999460415327846,
+      "loss": 1.4327,
+      "step": 1090
+    },
+    {
+      "epoch": 0.0957566149466114,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0029994483587631833,
+      "loss": 1.3614,
+      "step": 1091
+    },
+    {
+      "epoch": 0.0958443845295139,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0029994361690133067,
+      "loss": 1.4151,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09593215411241637,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0029994238460794194,
+      "loss": 1.4326,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09601992369531887,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.0029994113899627383,
+      "loss": 1.4586,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09610769327822134,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0029993988006644925,
+      "loss": 1.4072,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09619546286112383,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002999386078185924,
+      "loss": 1.3702,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09628323244402633,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.002999373222528289,
+      "loss": 1.3838,
+      "step": 1097
+    },
+    {
+      "epoch": 0.0963710020269288,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0029993602336928563,
+      "loss": 1.3674,
+      "step": 1098
+    },
+    {
+      "epoch": 0.0964587716098313,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0029993471116809074,
+      "loss": 1.4215,
+      "step": 1099
+    },
+    {
+      "epoch": 0.09654654119273377,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.0029993338564937378,
+      "loss": 1.3732,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09663431077563626,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0029993204681326557,
+      "loss": 1.3703,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09672208035853874,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002999306946598982,
+      "loss": 1.3311,
+      "step": 1102
+    },
+    {
+      "epoch": 0.09680984994144123,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.002999293291894052,
+      "loss": 1.4013,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09689761952434371,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0029992795040192127,
+      "loss": 1.3463,
+      "step": 1104
+    },
+    {
+      "epoch": 0.0969853891072462,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0029992655829758253,
+      "loss": 1.3948,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09707315869014868,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002999251528765264,
+      "loss": 1.3679,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09716092827305117,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0029992373413889146,
+      "loss": 1.3756,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09724869785595366,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0029992230208481787,
+      "loss": 1.4107,
+      "step": 1108
+    },
+    {
+      "epoch": 0.09733646743885614,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.00299920856714447,
+      "loss": 1.3858,
+      "step": 1109
+    },
+    {
+      "epoch": 0.09742423702175863,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002999193980279213,
+      "loss": 1.3736,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09751200660466111,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0029991792602538495,
+      "loss": 1.4024,
+      "step": 1111
+    },
+    {
+      "epoch": 0.0975997761875636,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0029991644070698317,
+      "loss": 1.342,
+      "step": 1112
+    },
+    {
+      "epoch": 0.09768754577046608,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0029991494207286247,
+      "loss": 1.3804,
+      "step": 1113
+    },
+    {
+      "epoch": 0.09777531535336857,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002999134301231708,
+      "loss": 1.4002,
+      "step": 1114
+    },
+    {
+      "epoch": 0.09786308493627105,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0029991190485805746,
+      "loss": 1.406,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09795085451917354,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0029991036627767293,
+      "loss": 1.4043,
+      "step": 1116
+    },
+    {
+      "epoch": 0.09803862410207602,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0029990881438216902,
+      "loss": 1.3829,
+      "step": 1117
+    },
+    {
+      "epoch": 0.09812639368497851,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00299907249171699,
+      "loss": 1.4053,
+      "step": 1118
+    },
+    {
+      "epoch": 0.098214163267881,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0029990567064641723,
+      "loss": 1.4058,
+      "step": 1119
+    },
+    {
+      "epoch": 0.09830193285078348,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0029990407880647954,
+      "loss": 1.3976,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09838970243368597,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0029990247365204313,
+      "loss": 1.3871,
+      "step": 1121
+    },
+    {
+      "epoch": 0.09847747201658845,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002999008551832663,
+      "loss": 1.3627,
+      "step": 1122
+    },
+    {
+      "epoch": 0.09856524159949094,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0029989922340030884,
+      "loss": 1.4073,
+      "step": 1123
+    },
+    {
+      "epoch": 0.09865301118239342,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0029989757830333184,
+      "loss": 1.4222,
+      "step": 1124
+    },
+    {
+      "epoch": 0.09874078076529591,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.002998959198924976,
+      "loss": 1.416,
+      "step": 1125
+    },
+    {
+      "epoch": 0.09882855034819839,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0029989424816796984,
+      "loss": 1.3457,
+      "step": 1126
+    },
+    {
+      "epoch": 0.09891631993110088,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.0029989256312991353,
+      "loss": 1.3658,
+      "step": 1127
+    },
+    {
+      "epoch": 0.09900408951400336,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002998908647784949,
+      "loss": 1.4118,
+      "step": 1128
+    },
+    {
+      "epoch": 0.09909185909690585,
+      "grad_norm": 0.25,
+      "learning_rate": 0.002998891531138818,
+      "loss": 1.3455,
+      "step": 1129
+    },
+    {
+      "epoch": 0.09917962867980833,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002998874281362429,
+      "loss": 1.3987,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09926739826271082,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.002998856898457486,
+      "loss": 1.3671,
+      "step": 1131
+    },
+    {
+      "epoch": 0.09935516784561331,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0029988393824257045,
+      "loss": 1.4391,
+      "step": 1132
+    },
+    {
+      "epoch": 0.09944293742851579,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0029988217332688133,
+      "loss": 1.3633,
+      "step": 1133
+    },
+    {
+      "epoch": 0.09953070701141828,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0029988039509885535,
+      "loss": 1.3812,
+      "step": 1134
+    },
+    {
+      "epoch": 0.09961847659432076,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.002998786035586681,
+      "loss": 1.3976,
+      "step": 1135
+    },
+    {
+      "epoch": 0.09970624617722325,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0029987679870649635,
+      "loss": 1.3556,
+      "step": 1136
+    },
+    {
+      "epoch": 0.09979401576012573,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0029987498054251825,
+      "loss": 1.3307,
+      "step": 1137
+    },
+    {
+      "epoch": 0.09988178534302822,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0029987314906691326,
+      "loss": 1.3842,
+      "step": 1138
+    },
+    {
+      "epoch": 0.0999695549259307,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0029987130427986206,
+      "loss": 1.3949,
+      "step": 1139
+    },
+    {
+      "epoch": 0.10005732450883319,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0029986944618154688,
+      "loss": 1.4001,
+      "step": 1140
+    },
+    {
+      "epoch": 0.10014509409173566,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0029986757477215094,
+      "loss": 1.3963,
+      "step": 1141
+    },
+    {
+      "epoch": 0.10023286367463816,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0029986569005185904,
+      "loss": 1.3784,
+      "step": 1142
+    },
+    {
+      "epoch": 0.10032063325754065,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0029986379202085714,
+      "loss": 1.3445,
+      "step": 1143
+    },
+    {
+      "epoch": 0.10040840284044313,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.0029986188067933264,
+      "loss": 1.3893,
+      "step": 1144
+    },
+    {
+      "epoch": 0.10049617242334562,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0029985995602747413,
+      "loss": 1.3833,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1005839420062481,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0029985801806547154,
+      "loss": 1.412,
+      "step": 1146
+    },
+    {
+      "epoch": 0.10067171158915059,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002998560667935162,
+      "loss": 1.3944,
+      "step": 1147
+    },
+    {
+      "epoch": 0.10075948117205306,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.002998541022118007,
+      "loss": 1.3138,
+      "step": 1148
+    },
+    {
+      "epoch": 0.10084725075495556,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0029985212432051883,
+      "loss": 1.383,
+      "step": 1149
+    },
+    {
+      "epoch": 0.10093502033785803,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.00299850133119866,
+      "loss": 1.293,
+      "step": 1150
+    },
+    {
+      "epoch": 0.10102278992076053,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.002998481286100385,
+      "loss": 1.464,
+      "step": 1151
+    },
+    {
+      "epoch": 0.101110559503663,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0029984611079123427,
+      "loss": 1.3577,
+      "step": 1152
+    },
+    {
+      "epoch": 0.1011983290865655,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0029984407966365248,
+      "loss": 1.3531,
+      "step": 1153
+    },
+    {
+      "epoch": 0.10128609866946799,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002998420352274936,
+      "loss": 1.3539,
+      "step": 1154
+    },
+    {
+      "epoch": 0.10137386825237046,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0029983997748295934,
+      "loss": 1.4027,
+      "step": 1155
+    },
+    {
+      "epoch": 0.10146163783527296,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002998379064302529,
+      "loss": 1.3267,
+      "step": 1156
+    },
+    {
+      "epoch": 0.10154940741817543,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002998358220695786,
+      "loss": 1.3353,
+      "step": 1157
+    },
+    {
+      "epoch": 0.10163717700107792,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.0029983372440114215,
+      "loss": 1.3713,
+      "step": 1158
+    },
+    {
+      "epoch": 0.1017249465839804,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.002998316134251506,
+      "loss": 1.3866,
+      "step": 1159
+    },
+    {
+      "epoch": 0.1018127161668829,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0029982948914181237,
+      "loss": 1.418,
+      "step": 1160
+    },
+    {
+      "epoch": 0.10190048574978537,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.00299827351551337,
+      "loss": 1.3636,
+      "step": 1161
+    },
+    {
+      "epoch": 0.10198825533268786,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0029982520065393557,
+      "loss": 1.3576,
+      "step": 1162
+    },
+    {
+      "epoch": 0.10207602491559034,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.0029982303644982026,
+      "loss": 1.406,
+      "step": 1163
+    },
+    {
+      "epoch": 0.10216379449849283,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0029982085893920474,
+      "loss": 1.4094,
+      "step": 1164
+    },
+    {
+      "epoch": 0.10225156408139531,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0029981866812230393,
+      "loss": 1.3641,
+      "step": 1165
+    },
+    {
+      "epoch": 0.1023393336642978,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0029981646399933402,
+      "loss": 1.3666,
+      "step": 1166
+    },
+    {
+      "epoch": 0.1024271032472003,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.002998142465705125,
+      "loss": 1.3237,
+      "step": 1167
+    },
+    {
+      "epoch": 0.10251487283010277,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0029981201583605833,
+      "loss": 1.3855,
+      "step": 1168
+    },
+    {
+      "epoch": 0.10260264241300526,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.002998097717961916,
+      "loss": 1.3694,
+      "step": 1169
+    },
+    {
+      "epoch": 0.10269041199590774,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002998075144511338,
+      "loss": 1.4139,
+      "step": 1170
+    },
+    {
+      "epoch": 0.10277818157881023,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0029980524380110777,
+      "loss": 1.3603,
+      "step": 1171
+    },
+    {
+      "epoch": 0.10286595116171271,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0029980295984633754,
+      "loss": 1.3754,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1029537207446152,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0029980066258704858,
+      "loss": 1.4013,
+      "step": 1173
+    },
+    {
+      "epoch": 0.10304149032751768,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.0029979835202346757,
+      "loss": 1.3714,
+      "step": 1174
+    },
+    {
+      "epoch": 0.10312925991042017,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0029979602815582264,
+      "loss": 1.393,
+      "step": 1175
+    },
+    {
+      "epoch": 0.10321702949332265,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0029979369098434307,
+      "loss": 1.371,
+      "step": 1176
+    },
+    {
+      "epoch": 0.10330479907622514,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0029979134050925954,
+      "loss": 1.3506,
+      "step": 1177
+    },
+    {
+      "epoch": 0.10339256865912763,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.00299788976730804,
+      "loss": 1.3718,
+      "step": 1178
+    },
+    {
+      "epoch": 0.10348033824203011,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0029978659964920986,
+      "loss": 1.3942,
+      "step": 1179
+    },
+    {
+      "epoch": 0.1035681078249326,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0029978420926471167,
+      "loss": 1.3182,
+      "step": 1180
+    },
+    {
+      "epoch": 0.10365587740783508,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0029978180557754526,
+      "loss": 1.3925,
+      "step": 1181
+    },
+    {
+      "epoch": 0.10374364699073757,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.00299779388587948,
+      "loss": 1.4286,
+      "step": 1182
+    },
+    {
+      "epoch": 0.10383141657364005,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0029977695829615835,
+      "loss": 1.3791,
+      "step": 1183
+    },
+    {
+      "epoch": 0.10391918615654254,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.0029977451470241624,
+      "loss": 1.3654,
+      "step": 1184
+    },
+    {
+      "epoch": 0.10400695573944502,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0029977205780696275,
+      "loss": 1.3322,
+      "step": 1185
+    },
+    {
+      "epoch": 0.10409472532234751,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002997695876100405,
+      "loss": 1.3478,
+      "step": 1186
+    },
+    {
+      "epoch": 0.10418249490524999,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0029976710411189315,
+      "loss": 1.4218,
+      "step": 1187
+    },
+    {
+      "epoch": 0.10427026448815248,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.002997646073127659,
+      "loss": 1.4149,
+      "step": 1188
+    },
+    {
+      "epoch": 0.10435803407105497,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002997620972129051,
+      "loss": 1.386,
+      "step": 1189
+    },
+    {
+      "epoch": 0.10444580365395745,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0029975957381255863,
+      "loss": 1.3308,
+      "step": 1190
+    },
+    {
+      "epoch": 0.10453357323685994,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.0029975703711197536,
+      "loss": 1.3653,
+      "step": 1191
+    },
+    {
+      "epoch": 0.10462134281976242,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002997544871114057,
+      "loss": 1.3197,
+      "step": 1192
+    },
+    {
+      "epoch": 0.10470911240266491,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.002997519238111014,
+      "loss": 1.3795,
+      "step": 1193
+    },
+    {
+      "epoch": 0.10479688198556739,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0029974934721131545,
+      "loss": 1.3968,
+      "step": 1194
+    },
+    {
+      "epoch": 0.10488465156846988,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0029974675731230206,
+      "loss": 1.358,
+      "step": 1195
+    },
+    {
+      "epoch": 0.10497242115137236,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.002997441541143169,
+      "loss": 1.3964,
+      "step": 1196
+    },
+    {
+      "epoch": 0.10506019073427485,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.002997415376176169,
+      "loss": 1.3614,
+      "step": 1197
+    },
+    {
+      "epoch": 0.10514796031717732,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.0029973890782246026,
+      "loss": 1.358,
+      "step": 1198
+    },
+    {
+      "epoch": 0.10523572990007982,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0029973626472910657,
+      "loss": 1.384,
+      "step": 1199
+    },
+    {
+      "epoch": 0.1053234994829823,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002997336083378167,
+      "loss": 1.401,
+      "step": 1200
+    },
+    {
+      "epoch": 0.10541126906588479,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002997309386488527,
+      "loss": 1.3944,
+      "step": 1201
+    },
+    {
+      "epoch": 0.10549903864878728,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0029972825566247823,
+      "loss": 1.4017,
+      "step": 1202
+    },
+    {
+      "epoch": 0.10558680823168975,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0029972555937895803,
+      "loss": 1.3158,
+      "step": 1203
+    },
+    {
+      "epoch": 0.10567457781459225,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.002997228497985582,
+      "loss": 1.4139,
+      "step": 1204
+    },
+    {
+      "epoch": 0.10576234739749472,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0029972012692154612,
+      "loss": 1.3973,
+      "step": 1205
+    },
+    {
+      "epoch": 0.10585011698039722,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0029971739074819066,
+      "loss": 1.3658,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1059378865632997,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.002997146412787617,
+      "loss": 1.3152,
+      "step": 1207
+    },
+    {
+      "epoch": 0.10602565614620219,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0029971187851353075,
+      "loss": 1.4202,
+      "step": 1208
+    },
+    {
+      "epoch": 0.10611342572910466,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0029970910245277036,
+      "loss": 1.3919,
+      "step": 1209
+    },
+    {
+      "epoch": 0.10620119531200715,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.002997063130967546,
+      "loss": 1.3949,
+      "step": 1210
+    },
+    {
+      "epoch": 0.10628896489490963,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0029970351044575874,
+      "loss": 1.3882,
+      "step": 1211
+    },
+    {
+      "epoch": 0.10637673447781212,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.002997006945000594,
+      "loss": 1.3757,
+      "step": 1212
+    },
+    {
+      "epoch": 0.10646450406071462,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0029969786525993448,
+      "loss": 1.4256,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1065522736436171,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0029969502272566327,
+      "loss": 1.4044,
+      "step": 1214
+    },
+    {
+      "epoch": 0.10664004322651958,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0029969216689752624,
+      "loss": 1.4365,
+      "step": 1215
+    },
+    {
+      "epoch": 0.10672781280942206,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029968929777580526,
+      "loss": 1.3836,
+      "step": 1216
+    },
+    {
+      "epoch": 0.10681558239232455,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.002996864153607836,
+      "loss": 1.3908,
+      "step": 1217
+    },
+    {
+      "epoch": 0.10690335197522703,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0029968351965274564,
+      "loss": 1.3522,
+      "step": 1218
+    },
+    {
+      "epoch": 0.10699112155812952,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0029968061065197725,
+      "loss": 1.391,
+      "step": 1219
+    },
+    {
+      "epoch": 0.107078891141032,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.002996776883587654,
+      "loss": 1.389,
+      "step": 1220
+    },
+    {
+      "epoch": 0.10716666072393449,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0029967475277339864,
+      "loss": 1.376,
+      "step": 1221
+    },
+    {
+      "epoch": 0.10725443030683697,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.002996718038961667,
+      "loss": 1.3482,
+      "step": 1222
+    },
+    {
+      "epoch": 0.10734219988973946,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0029966884172736054,
+      "loss": 1.3258,
+      "step": 1223
+    },
+    {
+      "epoch": 0.10742996947264195,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002996658662672726,
+      "loss": 1.3769,
+      "step": 1224
+    },
+    {
+      "epoch": 0.10751773905554443,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029966287751619644,
+      "loss": 1.3529,
+      "step": 1225
+    },
+    {
+      "epoch": 0.10760550863844692,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0029965987547442715,
+      "loss": 1.3632,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1076932782213494,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.00299656860142261,
+      "loss": 1.3678,
+      "step": 1227
+    },
+    {
+      "epoch": 0.10778104780425189,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.002996538315199955,
+      "loss": 1.302,
+      "step": 1228
+    },
+    {
+      "epoch": 0.10786881738715437,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.002996507896079297,
+      "loss": 1.4266,
+      "step": 1229
+    },
+    {
+      "epoch": 0.10795658697005686,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002996477344063637,
+      "loss": 1.3641,
+      "step": 1230
+    },
+    {
+      "epoch": 0.10804435655295934,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.002996446659155991,
+      "loss": 1.3934,
+      "step": 1231
+    },
+    {
+      "epoch": 0.10813212613586183,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002996415841359387,
+      "loss": 1.3668,
+      "step": 1232
+    },
+    {
+      "epoch": 0.10821989571876431,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0029963848906768677,
+      "loss": 1.445,
+      "step": 1233
+    },
+    {
+      "epoch": 0.1083076653016668,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0029963538071114864,
+      "loss": 1.3484,
+      "step": 1234
+    },
+    {
+      "epoch": 0.10839543488456928,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002996322590666312,
+      "loss": 1.3515,
+      "step": 1235
+    },
+    {
+      "epoch": 0.10848320446747177,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.002996291241344425,
+      "loss": 1.3772,
+      "step": 1236
+    },
+    {
+      "epoch": 0.10857097405037426,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0029962597591489198,
+      "loss": 1.38,
+      "step": 1237
+    },
+    {
+      "epoch": 0.10865874363327674,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.002996228144082903,
+      "loss": 1.435,
+      "step": 1238
+    },
+    {
+      "epoch": 0.10874651321617923,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0029961963961494947,
+      "loss": 1.4508,
+      "step": 1239
+    },
+    {
+      "epoch": 0.10883428279908171,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002996164515351829,
+      "loss": 1.3645,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1089220523819842,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.002996132501693052,
+      "loss": 1.3712,
+      "step": 1241
+    },
+    {
+      "epoch": 0.10900982196488668,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0029961003551763235,
+      "loss": 1.3633,
+      "step": 1242
+    },
+    {
+      "epoch": 0.10909759154778917,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0029960680758048166,
+      "loss": 1.3638,
+      "step": 1243
+    },
+    {
+      "epoch": 0.10918536113069165,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0029960356635817162,
+      "loss": 1.3499,
+      "step": 1244
+    },
+    {
+      "epoch": 0.10927313071359414,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002996003118510222,
+      "loss": 1.3687,
+      "step": 1245
+    },
+    {
+      "epoch": 0.10936090029649662,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002995970440593546,
+      "loss": 1.3434,
+      "step": 1246
+    },
+    {
+      "epoch": 0.10944866987939911,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0029959376298349136,
+      "loss": 1.3438,
+      "step": 1247
+    },
+    {
+      "epoch": 0.1095364394623016,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.002995904686237562,
+      "loss": 1.3487,
+      "step": 1248
+    },
+    {
+      "epoch": 0.10962420904520408,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0029958716098047434,
+      "loss": 1.3443,
+      "step": 1249
+    },
+    {
+      "epoch": 0.10971197862810657,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0029958384005397227,
+      "loss": 1.4267,
+      "step": 1250
+    },
+    {
+      "epoch": 0.10979974821100905,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002995805058445777,
+      "loss": 1.4072,
+      "step": 1251
+    },
+    {
+      "epoch": 0.10988751779391154,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.002995771583526197,
+      "loss": 1.4223,
+      "step": 1252
+    },
+    {
+      "epoch": 0.10997528737681402,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0029957379757842868,
+      "loss": 1.4047,
+      "step": 1253
+    },
+    {
+      "epoch": 0.11006305695971651,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0029957042352233634,
+      "loss": 1.3857,
+      "step": 1254
+    },
+    {
+      "epoch": 0.11015082654261898,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.002995670361846756,
+      "loss": 1.3768,
+      "step": 1255
+    },
+    {
+      "epoch": 0.11023859612552148,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0029956363556578086,
+      "loss": 1.3833,
+      "step": 1256
+    },
+    {
+      "epoch": 0.11032636570842395,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0029956022166598775,
+      "loss": 1.3569,
+      "step": 1257
+    },
+    {
+      "epoch": 0.11041413529132645,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0029955679448563323,
+      "loss": 1.3996,
+      "step": 1258
+    },
+    {
+      "epoch": 0.11050190487422894,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0029955335402505547,
+      "loss": 1.3399,
+      "step": 1259
+    },
+    {
+      "epoch": 0.11058967445713141,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00299549900284594,
+      "loss": 1.4107,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1106774440400339,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002995464332645899,
+      "loss": 1.3458,
+      "step": 1261
+    },
+    {
+      "epoch": 0.11076521362293638,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.0029954295296538507,
+      "loss": 1.2937,
+      "step": 1262
+    },
+    {
+      "epoch": 0.11085298320583888,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0029953945938732325,
+      "loss": 1.3608,
+      "step": 1263
+    },
+    {
+      "epoch": 0.11094075278874135,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002995359525307491,
+      "loss": 1.463,
+      "step": 1264
+    },
+    {
+      "epoch": 0.11102852237164385,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0029953243239600872,
+      "loss": 1.4157,
+      "step": 1265
+    },
+    {
+      "epoch": 0.11111629195454632,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0029952889898344965,
+      "loss": 1.3454,
+      "step": 1266
+    },
+    {
+      "epoch": 0.11120406153744881,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0029952535229342053,
+      "loss": 1.379,
+      "step": 1267
+    },
+    {
+      "epoch": 0.11129183112035129,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.002995217923262714,
+      "loss": 1.3982,
+      "step": 1268
+    },
+    {
+      "epoch": 0.11137960070325378,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0029951821908235365,
+      "loss": 1.3906,
+      "step": 1269
+    },
+    {
+      "epoch": 0.11146737028615626,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0029951463256201994,
+      "loss": 1.3601,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11155513986905875,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0029951103276562433,
+      "loss": 1.3265,
+      "step": 1271
+    },
+    {
+      "epoch": 0.11164290945196124,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.002995074196935219,
+      "loss": 1.3922,
+      "step": 1272
+    },
+    {
+      "epoch": 0.11173067903486372,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.002995037933460694,
+      "loss": 1.3893,
+      "step": 1273
+    },
+    {
+      "epoch": 0.11181844861776621,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0029950015372362474,
+      "loss": 1.3614,
+      "step": 1274
+    },
+    {
+      "epoch": 0.11190621820066869,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002994965008265471,
+      "loss": 1.3425,
+      "step": 1275
+    },
+    {
+      "epoch": 0.11199398778357118,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00299492834655197,
+      "loss": 1.329,
+      "step": 1276
+    },
+    {
+      "epoch": 0.11208175736647366,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002994891552099362,
+      "loss": 1.3674,
+      "step": 1277
+    },
+    {
+      "epoch": 0.11216952694937615,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0029948546249112798,
+      "loss": 1.3518,
+      "step": 1278
+    },
+    {
+      "epoch": 0.11225729653227863,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.002994817564991368,
+      "loss": 1.3763,
+      "step": 1279
+    },
+    {
+      "epoch": 0.11234506611518112,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.002994780372343283,
+      "loss": 1.3855,
+      "step": 1280
+    },
+    {
+      "epoch": 0.1124328356980836,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0029947430469706962,
+      "loss": 1.3847,
+      "step": 1281
+    },
+    {
+      "epoch": 0.11252060528098609,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.002994705588877292,
+      "loss": 1.3432,
+      "step": 1282
+    },
+    {
+      "epoch": 0.11260837486388858,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0029946679980667667,
+      "loss": 1.3996,
+      "step": 1283
+    },
+    {
+      "epoch": 0.11269614444679106,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0029946302745428306,
+      "loss": 1.3817,
+      "step": 1284
+    },
+    {
+      "epoch": 0.11278391402969355,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0029945924183092066,
+      "loss": 1.3386,
+      "step": 1285
+    },
+    {
+      "epoch": 0.11287168361259603,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.002994554429369632,
+      "loss": 1.4025,
+      "step": 1286
+    },
+    {
+      "epoch": 0.11295945319549852,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002994516307727855,
+      "loss": 1.3984,
+      "step": 1287
+    },
+    {
+      "epoch": 0.113047222778401,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0029944780533876377,
+      "loss": 1.3728,
+      "step": 1288
+    },
+    {
+      "epoch": 0.11313499236130349,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0029944396663527566,
+      "loss": 1.3559,
+      "step": 1289
+    },
+    {
+      "epoch": 0.11322276194420597,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0029944011466270007,
+      "loss": 1.3149,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11331053152710846,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0029943624942141704,
+      "loss": 1.3477,
+      "step": 1291
+    },
+    {
+      "epoch": 0.11339830111001094,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002994323709118082,
+      "loss": 1.3523,
+      "step": 1292
+    },
+    {
+      "epoch": 0.11348607069291343,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002994284791342563,
+      "loss": 1.3462,
+      "step": 1293
+    },
+    {
+      "epoch": 0.11357384027581592,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0029942457408914536,
+      "loss": 1.3958,
+      "step": 1294
+    },
+    {
+      "epoch": 0.1136616098587184,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0029942065577686083,
+      "loss": 1.3697,
+      "step": 1295
+    },
+    {
+      "epoch": 0.11374937944162089,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0029941672419778946,
+      "loss": 1.3842,
+      "step": 1296
+    },
+    {
+      "epoch": 0.11383714902452337,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0029941277935231925,
+      "loss": 1.3605,
+      "step": 1297
+    },
+    {
+      "epoch": 0.11392491860742586,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.002994088212408396,
+      "loss": 1.4183,
+      "step": 1298
+    },
+    {
+      "epoch": 0.11401268819032834,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0029940484986374113,
+      "loss": 1.4284,
+      "step": 1299
+    },
+    {
+      "epoch": 0.11410045777323083,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0029940086522141584,
+      "loss": 1.3648,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1141882273561333,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0029939686731425682,
+      "loss": 1.4495,
+      "step": 1301
+    },
+    {
+      "epoch": 0.1142759969390358,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0029939285614265887,
+      "loss": 1.3307,
+      "step": 1302
+    },
+    {
+      "epoch": 0.11436376652193828,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002993888317070178,
+      "loss": 1.3453,
+      "step": 1303
+    },
+    {
+      "epoch": 0.11445153610484077,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0029938479400773072,
+      "loss": 1.3246,
+      "step": 1304
+    },
+    {
+      "epoch": 0.11453930568774325,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002993807430451963,
+      "loss": 1.3524,
+      "step": 1305
+    },
+    {
+      "epoch": 0.11462707527064574,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0029937667881981414,
+      "loss": 1.3968,
+      "step": 1306
+    },
+    {
+      "epoch": 0.11471484485354823,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0029937260133198556,
+      "loss": 1.3437,
+      "step": 1307
+    },
+    {
+      "epoch": 0.1148026144364507,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.002993685105821129,
+      "loss": 1.4529,
+      "step": 1308
+    },
+    {
+      "epoch": 0.1148903840193532,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.002993644065705999,
+      "loss": 1.3231,
+      "step": 1309
+    },
+    {
+      "epoch": 0.11497815360225568,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0029936028929785164,
+      "loss": 1.4442,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11506592318515817,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0029935615876427444,
+      "loss": 1.3429,
+      "step": 1311
+    },
+    {
+      "epoch": 0.11515369276806064,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.00299352014970276,
+      "loss": 1.3844,
+      "step": 1312
+    },
+    {
+      "epoch": 0.11524146235096314,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0029934785791626527,
+      "loss": 1.3531,
+      "step": 1313
+    },
+    {
+      "epoch": 0.11532923193386561,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.002993436876026526,
+      "loss": 1.3314,
+      "step": 1314
+    },
+    {
+      "epoch": 0.1154170015167681,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0029933950402984957,
+      "loss": 1.3913,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11550477109967058,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.002993353071982689,
+      "loss": 1.4322,
+      "step": 1316
+    },
+    {
+      "epoch": 0.11559254068257308,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0029933109710832506,
+      "loss": 1.3768,
+      "step": 1317
+    },
+    {
+      "epoch": 0.11568031026547557,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0029932687376043347,
+      "loss": 1.3952,
+      "step": 1318
+    },
+    {
+      "epoch": 0.11576807984837804,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0029932263715501087,
+      "loss": 1.3741,
+      "step": 1319
+    },
+    {
+      "epoch": 0.11585584943128054,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0029931838729247554,
+      "loss": 1.4054,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11594361901418301,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0029931412417324678,
+      "loss": 1.3497,
+      "step": 1321
+    },
+    {
+      "epoch": 0.1160313885970855,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.002993098477977455,
+      "loss": 1.404,
+      "step": 1322
+    },
+    {
+      "epoch": 0.11611915817998798,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0029930555816639363,
+      "loss": 1.4049,
+      "step": 1323
+    },
+    {
+      "epoch": 0.11620692776289047,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.002993012552796146,
+      "loss": 1.3405,
+      "step": 1324
+    },
+    {
+      "epoch": 0.11629469734579295,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002992969391378331,
+      "loss": 1.3917,
+      "step": 1325
+    },
+    {
+      "epoch": 0.11638246692869544,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0029929260974147502,
+      "loss": 1.396,
+      "step": 1326
+    },
+    {
+      "epoch": 0.11647023651159792,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.002992882670909678,
+      "loss": 1.3071,
+      "step": 1327
+    },
+    {
+      "epoch": 0.11655800609450041,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.002992839111867399,
+      "loss": 1.3579,
+      "step": 1328
+    },
+    {
+      "epoch": 0.1166457756774029,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.002992795420292214,
+      "loss": 1.3896,
+      "step": 1329
+    },
+    {
+      "epoch": 0.11673354526030538,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0029927515961884338,
+      "loss": 1.3201,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11682131484320787,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.002992707639560384,
+      "loss": 1.3282,
+      "step": 1331
+    },
+    {
+      "epoch": 0.11690908442611035,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0029926635504124027,
+      "loss": 1.3491,
+      "step": 1332
+    },
+    {
+      "epoch": 0.11699685400901284,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.002992619328748842,
+      "loss": 1.4268,
+      "step": 1333
+    },
+    {
+      "epoch": 0.11708462359191532,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.002992574974574066,
+      "loss": 1.3757,
+      "step": 1334
+    },
+    {
+      "epoch": 0.11717239317481781,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002992530487892452,
+      "loss": 1.3592,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11726016275772029,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0029924858687083913,
+      "loss": 1.349,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11734793234062278,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002992441117026287,
+      "loss": 1.367,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11743570192352526,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0029923962328505566,
+      "loss": 1.3852,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11752347150642775,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0029923512161856296,
+      "loss": 1.3678,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11761124108933023,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0029923060670359485,
+      "loss": 1.3732,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11769901067223272,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.00299226078540597,
+      "loss": 1.3959,
+      "step": 1341
+    },
+    {
+      "epoch": 0.11778678025513521,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0029922153713001633,
+      "loss": 1.3595,
+      "step": 1342
+    },
+    {
+      "epoch": 0.11787454983803769,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0029921698247230106,
+      "loss": 1.3821,
+      "step": 1343
+    },
+    {
+      "epoch": 0.11796231942094018,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0029921241456790063,
+      "loss": 1.3625,
+      "step": 1344
+    },
+    {
+      "epoch": 0.11805008900384266,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0029920783341726596,
+      "loss": 1.4052,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11813785858674515,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0029920323902084918,
+      "loss": 1.3745,
+      "step": 1346
+    },
+    {
+      "epoch": 0.11822562816964763,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0029919863137910364,
+      "loss": 1.3355,
+      "step": 1347
+    },
+    {
+      "epoch": 0.11831339775255012,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002991940104924843,
+      "loss": 1.3472,
+      "step": 1348
+    },
+    {
+      "epoch": 0.1184011673354526,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00299189376361447,
+      "loss": 1.3406,
+      "step": 1349
+    },
+    {
+      "epoch": 0.11848893691835509,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.002991847289864493,
+      "loss": 1.3594,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11857670650125757,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0029918006836794973,
+      "loss": 1.3208,
+      "step": 1351
+    },
+    {
+      "epoch": 0.11866447608416006,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029917539450640836,
+      "loss": 1.3718,
+      "step": 1352
+    },
+    {
+      "epoch": 0.11875224566706255,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0029917070740228644,
+      "loss": 1.3373,
+      "step": 1353
+    },
+    {
+      "epoch": 0.11884001524996503,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.002991660070560466,
+      "loss": 1.3994,
+      "step": 1354
+    },
+    {
+      "epoch": 0.11892778483286752,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0029916129346815273,
+      "loss": 1.3803,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11901555441577,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0029915656663907003,
+      "loss": 1.3329,
+      "step": 1356
+    },
+    {
+      "epoch": 0.11910332399867249,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0029915182656926505,
+      "loss": 1.3175,
+      "step": 1357
+    },
+    {
+      "epoch": 0.11919109358157497,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.0029914707325920564,
+      "loss": 1.3346,
+      "step": 1358
+    },
+    {
+      "epoch": 0.11927886316447746,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0029914230670936084,
+      "loss": 1.3422,
+      "step": 1359
+    },
+    {
+      "epoch": 0.11936663274737994,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0029913752692020115,
+      "loss": 1.3332,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11945440233028243,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002991327338921983,
+      "loss": 1.3248,
+      "step": 1361
+    },
+    {
+      "epoch": 0.1195421719131849,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0029912792762582536,
+      "loss": 1.3769,
+      "step": 1362
+    },
+    {
+      "epoch": 0.1196299414960874,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.002991231081215567,
+      "loss": 1.3756,
+      "step": 1363
+    },
+    {
+      "epoch": 0.11971771107898989,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0029911827537986795,
+      "loss": 1.362,
+      "step": 1364
+    },
+    {
+      "epoch": 0.11980548066189237,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.002991134294012361,
+      "loss": 1.3265,
+      "step": 1365
+    },
+    {
+      "epoch": 0.11989325024479486,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.002991085701861394,
+      "loss": 1.3849,
+      "step": 1366
+    },
+    {
+      "epoch": 0.11998101982769734,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0029910369773505753,
+      "loss": 1.3654,
+      "step": 1367
+    },
+    {
+      "epoch": 0.12006878941059983,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0029909881204847125,
+      "loss": 1.3789,
+      "step": 1368
+    },
+    {
+      "epoch": 0.1201565589935023,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.002990939131268629,
+      "loss": 1.3777,
+      "step": 1369
+    },
+    {
+      "epoch": 0.1202443285764048,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002990890009707158,
+      "loss": 1.3323,
+      "step": 1370
+    },
+    {
+      "epoch": 0.12033209815930727,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0029908407558051496,
+      "loss": 1.3274,
+      "step": 1371
+    },
+    {
+      "epoch": 0.12041986774220977,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.002990791369567464,
+      "loss": 1.3721,
+      "step": 1372
+    },
+    {
+      "epoch": 0.12050763732511224,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0029907418509989747,
+      "loss": 1.3789,
+      "step": 1373
+    },
+    {
+      "epoch": 0.12059540690801474,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00299069220010457,
+      "loss": 1.3182,
+      "step": 1374
+    },
+    {
+      "epoch": 0.12068317649091721,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0029906424168891508,
+      "loss": 1.3797,
+      "step": 1375
+    },
+    {
+      "epoch": 0.1207709460738197,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.002990592501357629,
+      "loss": 1.4241,
+      "step": 1376
+    },
+    {
+      "epoch": 0.1208587156567222,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.002990542453514932,
+      "loss": 1.4091,
+      "step": 1377
+    },
+    {
+      "epoch": 0.12094648523962467,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0029904922733659987,
+      "loss": 1.3258,
+      "step": 1378
+    },
+    {
+      "epoch": 0.12103425482252717,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0029904419609157826,
+      "loss": 1.3287,
+      "step": 1379
+    },
+    {
+      "epoch": 0.12112202440542964,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0029903915161692486,
+      "loss": 1.3019,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12120979398833213,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0029903409391313753,
+      "loss": 1.3934,
+      "step": 1381
+    },
+    {
+      "epoch": 0.12129756357123461,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.002990290229807155,
+      "loss": 1.3374,
+      "step": 1382
+    },
+    {
+      "epoch": 0.1213853331541371,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0029902393882015924,
+      "loss": 1.3759,
+      "step": 1383
+    },
+    {
+      "epoch": 0.12147310273703958,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0029901884143197057,
+      "loss": 1.3208,
+      "step": 1384
+    },
+    {
+      "epoch": 0.12156087231994207,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0029901373081665248,
+      "loss": 1.3118,
+      "step": 1385
+    },
+    {
+      "epoch": 0.12164864190284455,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0029900860697470943,
+      "loss": 1.3979,
+      "step": 1386
+    },
+    {
+      "epoch": 0.12173641148574704,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002990034699066471,
+      "loss": 1.3388,
+      "step": 1387
+    },
+    {
+      "epoch": 0.12182418106864953,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002989983196129725,
+      "loss": 1.3613,
+      "step": 1388
+    },
+    {
+      "epoch": 0.12191195065155201,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0029899315609419404,
+      "loss": 1.4004,
+      "step": 1389
+    },
+    {
+      "epoch": 0.1219997202344545,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0029898797935082122,
+      "loss": 1.3582,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12208748981735698,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.00298982789383365,
+      "loss": 1.2863,
+      "step": 1391
+    },
+    {
+      "epoch": 0.12217525940025947,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002989775861923376,
+      "loss": 1.352,
+      "step": 1392
+    },
+    {
+      "epoch": 0.12226302898316195,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0029897236977825255,
+      "loss": 1.3793,
+      "step": 1393
+    },
+    {
+      "epoch": 0.12235079856606444,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002989671401416247,
+      "loss": 1.4031,
+      "step": 1394
+    },
+    {
+      "epoch": 0.12243856814896692,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0029896189728297025,
+      "loss": 1.3277,
+      "step": 1395
+    },
+    {
+      "epoch": 0.12252633773186941,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0029895664120280657,
+      "loss": 1.3364,
+      "step": 1396
+    },
+    {
+      "epoch": 0.12261410731477189,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0029895137190165243,
+      "loss": 1.3499,
+      "step": 1397
+    },
+    {
+      "epoch": 0.12270187689767438,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0029894608938002792,
+      "loss": 1.2889,
+      "step": 1398
+    },
+    {
+      "epoch": 0.12278964648057687,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002989407936384544,
+      "loss": 1.374,
+      "step": 1399
+    },
+    {
+      "epoch": 0.12287741606347935,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.002989354846774545,
+      "loss": 1.3264,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12296518564638184,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0029893016249755225,
+      "loss": 1.3311,
+      "step": 1401
+    },
+    {
+      "epoch": 0.12305295522928432,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0029892482709927286,
+      "loss": 1.3802,
+      "step": 1402
+    },
+    {
+      "epoch": 0.12314072481218681,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0029891947848314296,
+      "loss": 1.3593,
+      "step": 1403
+    },
+    {
+      "epoch": 0.12322849439508929,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0029891411664969042,
+      "loss": 1.3706,
+      "step": 1404
+    },
+    {
+      "epoch": 0.12331626397799178,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0029890874159944447,
+      "loss": 1.3851,
+      "step": 1405
+    },
+    {
+      "epoch": 0.12340403356089426,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0029890335333293554,
+      "loss": 1.3679,
+      "step": 1406
+    },
+    {
+      "epoch": 0.12349180314379675,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0029889795185069545,
+      "loss": 1.4016,
+      "step": 1407
+    },
+    {
+      "epoch": 0.12357957272669923,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0029889253715325737,
+      "loss": 1.359,
+      "step": 1408
+    },
+    {
+      "epoch": 0.12366734230960172,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0029888710924115564,
+      "loss": 1.3669,
+      "step": 1409
+    },
+    {
+      "epoch": 0.1237551118925042,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.0029888166811492603,
+      "loss": 1.3607,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12384288147540669,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0029887621377510543,
+      "loss": 1.373,
+      "step": 1411
+    },
+    {
+      "epoch": 0.12393065105830918,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002988707462222323,
+      "loss": 1.3402,
+      "step": 1412
+    },
+    {
+      "epoch": 0.12401842064121166,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0029886526545684624,
+      "loss": 1.3743,
+      "step": 1413
+    },
+    {
+      "epoch": 0.12410619022411415,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0029885977147948817,
+      "loss": 1.4382,
+      "step": 1414
+    },
+    {
+      "epoch": 0.12419395980701663,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0029885426429070022,
+      "loss": 1.355,
+      "step": 1415
+    },
+    {
+      "epoch": 0.12428172938991912,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0029884874389102607,
+      "loss": 1.4003,
+      "step": 1416
+    },
+    {
+      "epoch": 0.1243694989728216,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0029884321028101055,
+      "loss": 1.3275,
+      "step": 1417
+    },
+    {
+      "epoch": 0.12445726855572409,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.002988376634611997,
+      "loss": 1.3964,
+      "step": 1418
+    },
+    {
+      "epoch": 0.12454503813862657,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0029883210343214107,
+      "loss": 1.3267,
+      "step": 1419
+    },
+    {
+      "epoch": 0.12463280772152906,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.002988265301943834,
+      "loss": 1.3603,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12472057730443153,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.002988209437484767,
+      "loss": 1.3409,
+      "step": 1421
+    },
+    {
+      "epoch": 0.12480834688733403,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0029881534409497233,
+      "loss": 1.3781,
+      "step": 1422
+    },
+    {
+      "epoch": 0.12489611647023652,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0029880973123442302,
+      "loss": 1.3592,
+      "step": 1423
+    },
+    {
+      "epoch": 0.124983886053139,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002988041051673827,
+      "loss": 1.3418,
+      "step": 1424
+    },
+    {
+      "epoch": 0.12507165563604147,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002987984658944066,
+      "loss": 1.359,
+      "step": 1425
+    },
+    {
+      "epoch": 0.12515942521894396,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002987928134160514,
+      "loss": 1.3482,
+      "step": 1426
+    },
+    {
+      "epoch": 0.12524719480184646,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029878714773287477,
+      "loss": 1.3765,
+      "step": 1427
+    },
+    {
+      "epoch": 0.12533496438474895,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0029878146884543618,
+      "loss": 1.3548,
+      "step": 1428
+    },
+    {
+      "epoch": 0.1254227339676514,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002987757767542959,
+      "loss": 1.3192,
+      "step": 1429
+    },
+    {
+      "epoch": 0.1255105035505539,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.002987700714600157,
+      "loss": 1.3591,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1255982731334564,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0029876435296315885,
+      "loss": 1.3376,
+      "step": 1431
+    },
+    {
+      "epoch": 0.1256860427163589,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002987586212642896,
+      "loss": 1.3401,
+      "step": 1432
+    },
+    {
+      "epoch": 0.12577381229926135,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0029875287636397367,
+      "loss": 1.3824,
+      "step": 1433
+    },
+    {
+      "epoch": 0.12586158188216384,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.002987471182627781,
+      "loss": 1.3772,
+      "step": 1434
+    },
+    {
+      "epoch": 0.12594935146506633,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.002987413469612712,
+      "loss": 1.2982,
+      "step": 1435
+    },
+    {
+      "epoch": 0.12603712104796883,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0029873556246002254,
+      "loss": 1.3317,
+      "step": 1436
+    },
+    {
+      "epoch": 0.12612489063087132,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.00298729764759603,
+      "loss": 1.3751,
+      "step": 1437
+    },
+    {
+      "epoch": 0.12621266021377378,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0029872395386058483,
+      "loss": 1.3396,
+      "step": 1438
+    },
+    {
+      "epoch": 0.12630042979667627,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0029871812976354157,
+      "loss": 1.3549,
+      "step": 1439
+    },
+    {
+      "epoch": 0.12638819937957876,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.002987122924690479,
+      "loss": 1.3742,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12647596896248126,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0029870644197768016,
+      "loss": 1.2948,
+      "step": 1441
+    },
+    {
+      "epoch": 0.12656373854538372,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0029870057829001553,
+      "loss": 1.3143,
+      "step": 1442
+    },
+    {
+      "epoch": 0.1266515081282862,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002986947014066329,
+      "loss": 1.3048,
+      "step": 1443
+    },
+    {
+      "epoch": 0.1267392777111887,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.002986888113281123,
+      "loss": 1.3771,
+      "step": 1444
+    },
+    {
+      "epoch": 0.1268270472940912,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002986829080550349,
+      "loss": 1.3069,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12691481687699369,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.002986769915879835,
+      "loss": 1.3741,
+      "step": 1446
+    },
+    {
+      "epoch": 0.12700258645989615,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002986710619275419,
+      "loss": 1.3294,
+      "step": 1447
+    },
+    {
+      "epoch": 0.12709035604279864,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0029866511907429547,
+      "loss": 1.3173,
+      "step": 1448
+    },
+    {
+      "epoch": 0.12717812562570113,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0029865916302883066,
+      "loss": 1.328,
+      "step": 1449
+    },
+    {
+      "epoch": 0.12726589520860362,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0029865319379173525,
+      "loss": 1.3647,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1273536647915061,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002986472113635985,
+      "loss": 1.3442,
+      "step": 1451
+    },
+    {
+      "epoch": 0.12744143437440858,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0029864121574501084,
+      "loss": 1.3617,
+      "step": 1452
+    },
+    {
+      "epoch": 0.12752920395731107,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0029863520693656394,
+      "loss": 1.3258,
+      "step": 1453
+    },
+    {
+      "epoch": 0.12761697354021356,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.002986291849388509,
+      "loss": 1.326,
+      "step": 1454
+    },
+    {
+      "epoch": 0.12770474312311603,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0029862314975246606,
+      "loss": 1.3416,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12779251270601852,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.00298617101378005,
+      "loss": 1.3525,
+      "step": 1456
+    },
+    {
+      "epoch": 0.127880282288921,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002986110398160648,
+      "loss": 1.3637,
+      "step": 1457
+    },
+    {
+      "epoch": 0.1279680518718235,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0029860496506724362,
+      "loss": 1.3911,
+      "step": 1458
+    },
+    {
+      "epoch": 0.128055821454726,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00298598877132141,
+      "loss": 1.3446,
+      "step": 1459
+    },
+    {
+      "epoch": 0.12814359103762846,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0029859277601135786,
+      "loss": 1.3192,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12823136062053095,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0029858666170549638,
+      "loss": 1.3302,
+      "step": 1461
+    },
+    {
+      "epoch": 0.12831913020343344,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.002985805342151599,
+      "loss": 1.3877,
+      "step": 1462
+    },
+    {
+      "epoch": 0.12840689978633593,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0029857439354095323,
+      "loss": 1.3456,
+      "step": 1463
+    },
+    {
+      "epoch": 0.1284946693692384,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.002985682396834825,
+      "loss": 1.3953,
+      "step": 1464
+    },
+    {
+      "epoch": 0.1285824389521409,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0029856207264335496,
+      "loss": 1.3483,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12867020853504338,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0029855589242117936,
+      "loss": 1.3472,
+      "step": 1466
+    },
+    {
+      "epoch": 0.12875797811794587,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0029854969901756566,
+      "loss": 1.3891,
+      "step": 1467
+    },
+    {
+      "epoch": 0.12884574770084833,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0029854349243312508,
+      "loss": 1.3622,
+      "step": 1468
+    },
+    {
+      "epoch": 0.12893351728375083,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.002985372726684702,
+      "loss": 1.2893,
+      "step": 1469
+    },
+    {
+      "epoch": 0.12902128686665332,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002985310397242148,
+      "loss": 1.3801,
+      "step": 1470
+    },
+    {
+      "epoch": 0.1291090564495558,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002985247936009743,
+      "loss": 1.3415,
+      "step": 1471
+    },
+    {
+      "epoch": 0.1291968260324583,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0029851853429936487,
+      "loss": 1.3737,
+      "step": 1472
+    },
+    {
+      "epoch": 0.12928459561536076,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0029851226182000448,
+      "loss": 1.371,
+      "step": 1473
+    },
+    {
+      "epoch": 0.12937236519826326,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.002985059761635121,
+      "loss": 1.3949,
+      "step": 1474
+    },
+    {
+      "epoch": 0.12946013478116575,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0029849967733050816,
+      "loss": 1.411,
+      "step": 1475
+    },
+    {
+      "epoch": 0.12954790436406824,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0029849336532161426,
+      "loss": 1.3172,
+      "step": 1476
+    },
+    {
+      "epoch": 0.1296356739469707,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.002984870401374534,
+      "loss": 1.3485,
+      "step": 1477
+    },
+    {
+      "epoch": 0.1297234435298732,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0029848070177864987,
+      "loss": 1.3172,
+      "step": 1478
+    },
+    {
+      "epoch": 0.1298112131127757,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.002984743502458292,
+      "loss": 1.3329,
+      "step": 1479
+    },
+    {
+      "epoch": 0.12989898269567818,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002984679855396184,
+      "loss": 1.2902,
+      "step": 1480
+    },
+    {
+      "epoch": 0.12998675227858067,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002984616076606455,
+      "loss": 1.3072,
+      "step": 1481
+    },
+    {
+      "epoch": 0.13007452186148313,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0029845521660953993,
+      "loss": 1.4163,
+      "step": 1482
+    },
+    {
+      "epoch": 0.13016229144438562,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002984488123869326,
+      "loss": 1.3862,
+      "step": 1483
+    },
+    {
+      "epoch": 0.13025006102728812,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002984423949934555,
+      "loss": 1.4167,
+      "step": 1484
+    },
+    {
+      "epoch": 0.1303378306101906,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0029843596442974203,
+      "loss": 1.33,
+      "step": 1485
+    },
+    {
+      "epoch": 0.13042560019309307,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0029842952069642683,
+      "loss": 1.3811,
+      "step": 1486
+    },
+    {
+      "epoch": 0.13051336977599556,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.002984230637941459,
+      "loss": 1.3511,
+      "step": 1487
+    },
+    {
+      "epoch": 0.13060113935889806,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.002984165937235365,
+      "loss": 1.3482,
+      "step": 1488
+    },
+    {
+      "epoch": 0.13068890894180055,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.002984101104852372,
+      "loss": 1.3123,
+      "step": 1489
+    },
+    {
+      "epoch": 0.130776678524703,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.002984036140798879,
+      "loss": 1.3456,
+      "step": 1490
+    },
+    {
+      "epoch": 0.1308644481076055,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002983971045081297,
+      "loss": 1.2824,
+      "step": 1491
+    },
+    {
+      "epoch": 0.130952217690508,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002983905817706051,
+      "loss": 1.3123,
+      "step": 1492
+    },
+    {
+      "epoch": 0.13103998727341049,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0029838404586795793,
+      "loss": 1.3536,
+      "step": 1493
+    },
+    {
+      "epoch": 0.13112775685631298,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0029837749680083317,
+      "loss": 1.3014,
+      "step": 1494
+    },
+    {
+      "epoch": 0.13121552643921544,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0029837093456987727,
+      "loss": 1.427,
+      "step": 1495
+    },
+    {
+      "epoch": 0.13130329602211793,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.002983643591757378,
+      "loss": 1.3655,
+      "step": 1496
+    },
+    {
+      "epoch": 0.13139106560502042,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.002983577706190638,
+      "loss": 1.2918,
+      "step": 1497
+    },
+    {
+      "epoch": 0.13147883518792292,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.002983511689005055,
+      "loss": 1.3853,
+      "step": 1498
+    },
+    {
+      "epoch": 0.13156660477082538,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002983445540207145,
+      "loss": 1.3599,
+      "step": 1499
+    },
+    {
+      "epoch": 0.13165437435372787,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002983379259803436,
+      "loss": 1.3563,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13165437435372787,
+      "eval_loss": 1.3339033126831055,
+      "eval_runtime": 437.6205,
+      "eval_samples_per_second": 33.682,
+      "eval_steps_per_second": 8.421,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13174214393663036,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.00298331284780047,
+      "loss": 1.3152,
+      "step": 1501
+    },
+    {
+      "epoch": 0.13182991351953285,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0029832463042048017,
+      "loss": 1.3866,
+      "step": 1502
+    },
+    {
+      "epoch": 0.13191768310243532,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002983179629022999,
+      "loss": 1.3914,
+      "step": 1503
+    },
+    {
+      "epoch": 0.1320054526853378,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002983112822261641,
+      "loss": 1.3485,
+      "step": 1504
+    },
+    {
+      "epoch": 0.1320932222682403,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.002983045883927323,
+      "loss": 1.3199,
+      "step": 1505
+    },
+    {
+      "epoch": 0.1321809918511428,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0029829788140266506,
+      "loss": 1.3547,
+      "step": 1506
+    },
+    {
+      "epoch": 0.13226876143404528,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.002982911612566244,
+      "loss": 1.2855,
+      "step": 1507
+    },
+    {
+      "epoch": 0.13235653101694775,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0029828442795527352,
+      "loss": 1.3168,
+      "step": 1508
+    },
+    {
+      "epoch": 0.13244430059985024,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0029827768149927694,
+      "loss": 1.3254,
+      "step": 1509
+    },
+    {
+      "epoch": 0.13253207018275273,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0029827092188930058,
+      "loss": 1.3223,
+      "step": 1510
+    },
+    {
+      "epoch": 0.13261983976565522,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0029826414912601158,
+      "loss": 1.3559,
+      "step": 1511
+    },
+    {
+      "epoch": 0.1327076093485577,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0029825736321007833,
+      "loss": 1.3301,
+      "step": 1512
+    },
+    {
+      "epoch": 0.13279537893146018,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0029825056414217056,
+      "loss": 1.3512,
+      "step": 1513
+    },
+    {
+      "epoch": 0.13288314851436267,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0029824375192295943,
+      "loss": 1.351,
+      "step": 1514
+    },
+    {
+      "epoch": 0.13297091809726516,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029823692655311716,
+      "loss": 1.3494,
+      "step": 1515
+    },
+    {
+      "epoch": 0.13305868768016765,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0029823008803331746,
+      "loss": 1.3305,
+      "step": 1516
+    },
+    {
+      "epoch": 0.13314645726307012,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0029822323636423524,
+      "loss": 1.342,
+      "step": 1517
+    },
+    {
+      "epoch": 0.1332342268459726,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.002982163715465467,
+      "loss": 1.3594,
+      "step": 1518
+    },
+    {
+      "epoch": 0.1333219964288751,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0029820949358092944,
+      "loss": 1.3676,
+      "step": 1519
+    },
+    {
+      "epoch": 0.1334097660117776,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0029820260246806217,
+      "loss": 1.3476,
+      "step": 1520
+    },
+    {
+      "epoch": 0.13349753559468006,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0029819569820862517,
+      "loss": 1.3209,
+      "step": 1521
+    },
+    {
+      "epoch": 0.13358530517758255,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0029818878080329976,
+      "loss": 1.2706,
+      "step": 1522
+    },
+    {
+      "epoch": 0.13367307476048504,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0029818185025276872,
+      "loss": 1.3376,
+      "step": 1523
+    },
+    {
+      "epoch": 0.13376084434338753,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0029817490655771593,
+      "loss": 1.3505,
+      "step": 1524
+    },
+    {
+      "epoch": 0.13384861392629,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.002981679497188269,
+      "loss": 1.3944,
+      "step": 1525
+    },
+    {
+      "epoch": 0.13393638350919249,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0029816097973678814,
+      "loss": 1.3454,
+      "step": 1526
+    },
+    {
+      "epoch": 0.13402415309209498,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0029815399661228753,
+      "loss": 1.3408,
+      "step": 1527
+    },
+    {
+      "epoch": 0.13411192267499747,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0029814700034601436,
+      "loss": 1.3401,
+      "step": 1528
+    },
+    {
+      "epoch": 0.13419969225789996,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002981399909386591,
+      "loss": 1.342,
+      "step": 1529
+    },
+    {
+      "epoch": 0.13428746184080242,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.002981329683909135,
+      "loss": 1.3279,
+      "step": 1530
+    },
+    {
+      "epoch": 0.13437523142370492,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.002981259327034707,
+      "loss": 1.3395,
+      "step": 1531
+    },
+    {
+      "epoch": 0.1344630010066074,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0029811888387702512,
+      "loss": 1.3474,
+      "step": 1532
+    },
+    {
+      "epoch": 0.1345507705895099,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0029811182191227244,
+      "loss": 1.346,
+      "step": 1533
+    },
+    {
+      "epoch": 0.13463854017241236,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.002981047468099096,
+      "loss": 1.3413,
+      "step": 1534
+    },
+    {
+      "epoch": 0.13472630975531485,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0029809765857063495,
+      "loss": 1.2592,
+      "step": 1535
+    },
+    {
+      "epoch": 0.13481407933821735,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0029809055719514804,
+      "loss": 1.4012,
+      "step": 1536
+    },
+    {
+      "epoch": 0.13490184892111984,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0029808344268414974,
+      "loss": 1.3536,
+      "step": 1537
+    },
+    {
+      "epoch": 0.1349896185040223,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002980763150383422,
+      "loss": 1.3417,
+      "step": 1538
+    },
+    {
+      "epoch": 0.1350773880869248,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.00298069174258429,
+      "loss": 1.3449,
+      "step": 1539
+    },
+    {
+      "epoch": 0.13516515766982728,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0029806202034511472,
+      "loss": 1.3331,
+      "step": 1540
+    },
+    {
+      "epoch": 0.13525292725272978,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002980548532991056,
+      "loss": 1.3557,
+      "step": 1541
+    },
+    {
+      "epoch": 0.13534069683563227,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002980476731211089,
+      "loss": 1.371,
+      "step": 1542
+    },
+    {
+      "epoch": 0.13542846641853473,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0029804047981183334,
+      "loss": 1.358,
+      "step": 1543
+    },
+    {
+      "epoch": 0.13551623600143722,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0029803327337198887,
+      "loss": 1.334,
+      "step": 1544
+    },
+    {
+      "epoch": 0.13560400558433972,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002980260538022866,
+      "loss": 1.3648,
+      "step": 1545
+    },
+    {
+      "epoch": 0.1356917751672422,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0029801882110343926,
+      "loss": 1.3304,
+      "step": 1546
+    },
+    {
+      "epoch": 0.13577954475014467,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0029801157527616063,
+      "loss": 1.338,
+      "step": 1547
+    },
+    {
+      "epoch": 0.13586731433304716,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0029800431632116577,
+      "loss": 1.3448,
+      "step": 1548
+    },
+    {
+      "epoch": 0.13595508391594965,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.002979970442391711,
+      "loss": 1.3134,
+      "step": 1549
+    },
+    {
+      "epoch": 0.13604285349885215,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.002979897590308945,
+      "loss": 1.3217,
+      "step": 1550
+    },
+    {
+      "epoch": 0.13613062308175464,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0029798246069705483,
+      "loss": 1.32,
+      "step": 1551
+    },
+    {
+      "epoch": 0.1362183926646571,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.002979751492383725,
+      "loss": 1.3061,
+      "step": 1552
+    },
+    {
+      "epoch": 0.1363061622475596,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0029796782465556916,
+      "loss": 1.3187,
+      "step": 1553
+    },
+    {
+      "epoch": 0.13639393183046208,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.002979604869493676,
+      "loss": 1.3302,
+      "step": 1554
+    },
+    {
+      "epoch": 0.13648170141336458,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0029795313612049202,
+      "loss": 1.3247,
+      "step": 1555
+    },
+    {
+      "epoch": 0.13656947099626704,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0029794577216966807,
+      "loss": 1.3718,
+      "step": 1556
+    },
+    {
+      "epoch": 0.13665724057916953,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0029793839509762244,
+      "loss": 1.3535,
+      "step": 1557
+    },
+    {
+      "epoch": 0.13674501016207202,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0029793100490508313,
+      "loss": 1.2986,
+      "step": 1558
+    },
+    {
+      "epoch": 0.13683277974497451,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.002979236015927797,
+      "loss": 1.3521,
+      "step": 1559
+    },
+    {
+      "epoch": 0.13692054932787698,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0029791618516144276,
+      "loss": 1.3595,
+      "step": 1560
+    },
+    {
+      "epoch": 0.13700831891077947,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0029790875561180423,
+      "loss": 1.3786,
+      "step": 1561
+    },
+    {
+      "epoch": 0.13709608849368196,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0029790131294459743,
+      "loss": 1.3616,
+      "step": 1562
+    },
+    {
+      "epoch": 0.13718385807658445,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0029789385716055684,
+      "loss": 1.3719,
+      "step": 1563
+    },
+    {
+      "epoch": 0.13727162765948694,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0029788638826041846,
+      "loss": 1.4084,
+      "step": 1564
+    },
+    {
+      "epoch": 0.1373593972423894,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.0029787890624491933,
+      "loss": 1.341,
+      "step": 1565
+    },
+    {
+      "epoch": 0.1374471668252919,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0029787141111479796,
+      "loss": 1.3723,
+      "step": 1566
+    },
+    {
+      "epoch": 0.1375349364081944,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.00297863902870794,
+      "loss": 1.3145,
+      "step": 1567
+    },
+    {
+      "epoch": 0.13762270599109688,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0029785638151364857,
+      "loss": 1.3448,
+      "step": 1568
+    },
+    {
+      "epoch": 0.13771047557399935,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0029784884704410403,
+      "loss": 1.3786,
+      "step": 1569
+    },
+    {
+      "epoch": 0.13779824515690184,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0029784129946290386,
+      "loss": 1.3183,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13788601473980433,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002978337387707931,
+      "loss": 1.3635,
+      "step": 1571
+    },
+    {
+      "epoch": 0.13797378432270682,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002978261649685179,
+      "loss": 1.3174,
+      "step": 1572
+    },
+    {
+      "epoch": 0.13806155390560929,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.002978185780568258,
+      "loss": 1.35,
+      "step": 1573
+    },
+    {
+      "epoch": 0.13814932348851178,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0029781097803646558,
+      "loss": 1.3759,
+      "step": 1574
+    },
+    {
+      "epoch": 0.13823709307141427,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0029780336490818735,
+      "loss": 1.394,
+      "step": 1575
+    },
+    {
+      "epoch": 0.13832486265431676,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.002977957386727425,
+      "loss": 1.2821,
+      "step": 1576
+    },
+    {
+      "epoch": 0.13841263223721925,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.002977880993308836,
+      "loss": 1.3619,
+      "step": 1577
+    },
+    {
+      "epoch": 0.13850040182012172,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.002977804468833648,
+      "loss": 1.3,
+      "step": 1578
+    },
+    {
+      "epoch": 0.1385881714030242,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0029777278133094123,
+      "loss": 1.31,
+      "step": 1579
+    },
+    {
+      "epoch": 0.1386759409859267,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0029776510267436954,
+      "loss": 1.3326,
+      "step": 1580
+    },
+    {
+      "epoch": 0.1387637105688292,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.002977574109144075,
+      "loss": 1.4208,
+      "step": 1581
+    },
+    {
+      "epoch": 0.13885148015173165,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.002977497060518143,
+      "loss": 1.361,
+      "step": 1582
+    },
+    {
+      "epoch": 0.13893924973463415,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0029774198808735037,
+      "loss": 1.2836,
+      "step": 1583
+    },
+    {
+      "epoch": 0.13902701931753664,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002977342570217775,
+      "loss": 1.382,
+      "step": 1584
+    },
+    {
+      "epoch": 0.13911478890043913,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0029772651285585863,
+      "loss": 1.3055,
+      "step": 1585
+    },
+    {
+      "epoch": 0.13920255848334162,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.002977187555903581,
+      "loss": 1.4256,
+      "step": 1586
+    },
+    {
+      "epoch": 0.13929032806624408,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0029771098522604154,
+      "loss": 1.3969,
+      "step": 1587
+    },
+    {
+      "epoch": 0.13937809764914658,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.002977032017636759,
+      "loss": 1.3544,
+      "step": 1588
+    },
+    {
+      "epoch": 0.13946586723204907,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0029769540520402926,
+      "loss": 1.3442,
+      "step": 1589
+    },
+    {
+      "epoch": 0.13955363681495156,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002976875955478712,
+      "loss": 1.398,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13964140639785402,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.002976797727959725,
+      "loss": 1.3587,
+      "step": 1591
+    },
+    {
+      "epoch": 0.13972917598075651,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0029767193694910515,
+      "loss": 1.3348,
+      "step": 1592
+    },
+    {
+      "epoch": 0.139816945563659,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.0029766408800804265,
+      "loss": 1.3375,
+      "step": 1593
+    },
+    {
+      "epoch": 0.1399047151465615,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0029765622597355953,
+      "loss": 1.3374,
+      "step": 1594
+    },
+    {
+      "epoch": 0.13999248472946396,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.0029764835084643186,
+      "loss": 1.2843,
+      "step": 1595
+    },
+    {
+      "epoch": 0.14008025431236645,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.002976404626274368,
+      "loss": 1.304,
+      "step": 1596
+    },
+    {
+      "epoch": 0.14016802389526895,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0029763256131735286,
+      "loss": 1.3601,
+      "step": 1597
+    },
+    {
+      "epoch": 0.14025579347817144,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0029762464691696,
+      "loss": 1.3425,
+      "step": 1598
+    },
+    {
+      "epoch": 0.14034356306107393,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0029761671942703924,
+      "loss": 1.3645,
+      "step": 1599
+    },
+    {
+      "epoch": 0.1404313326439764,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.00297608778848373,
+      "loss": 1.3378,
+      "step": 1600
+    },
+    {
+      "epoch": 0.14051910222687888,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0029760082518174493,
+      "loss": 1.2775,
+      "step": 1601
+    },
+    {
+      "epoch": 0.14060687180978138,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002975928584279402,
+      "loss": 1.2902,
+      "step": 1602
+    },
+    {
+      "epoch": 0.14069464139268387,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.002975848785877449,
+      "loss": 1.2893,
+      "step": 1603
+    },
+    {
+      "epoch": 0.14078241097558633,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0029757688566194674,
+      "loss": 1.3126,
+      "step": 1604
+    },
+    {
+      "epoch": 0.14087018055848882,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.0029756887965133455,
+      "loss": 1.3202,
+      "step": 1605
+    },
+    {
+      "epoch": 0.1409579501413913,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0029756086055669844,
+      "loss": 1.2834,
+      "step": 1606
+    },
+    {
+      "epoch": 0.1410457197242938,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0029755282837882996,
+      "loss": 1.3656,
+      "step": 1607
+    },
+    {
+      "epoch": 0.14113348930719627,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.002975447831185218,
+      "loss": 1.3275,
+      "step": 1608
+    },
+    {
+      "epoch": 0.14122125889009876,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00297536724776568,
+      "loss": 1.3322,
+      "step": 1609
+    },
+    {
+      "epoch": 0.14130902847300125,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0029752865335376387,
+      "loss": 1.3148,
+      "step": 1610
+    },
+    {
+      "epoch": 0.14139679805590374,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00297520568850906,
+      "loss": 1.3834,
+      "step": 1611
+    },
+    {
+      "epoch": 0.14148456763880624,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.0029751247126879243,
+      "loss": 1.3611,
+      "step": 1612
+    },
+    {
+      "epoch": 0.1415723372217087,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0029750436060822223,
+      "loss": 1.3554,
+      "step": 1613
+    },
+    {
+      "epoch": 0.1416601068046112,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.002974962368699959,
+      "loss": 1.3455,
+      "step": 1614
+    },
+    {
+      "epoch": 0.14174787638751368,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.0029748810005491527,
+      "loss": 1.3609,
+      "step": 1615
+    },
+    {
+      "epoch": 0.14183564597041617,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0029747995016378348,
+      "loss": 1.2825,
+      "step": 1616
+    },
+    {
+      "epoch": 0.14192341555331864,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0029747178719740468,
+      "loss": 1.3947,
+      "step": 1617
+    },
+    {
+      "epoch": 0.14201118513622113,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.002974636111565847,
+      "loss": 1.3454,
+      "step": 1618
+    },
+    {
+      "epoch": 0.14209895471912362,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0029745542204213044,
+      "loss": 1.3789,
+      "step": 1619
+    },
+    {
+      "epoch": 0.1421867243020261,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.002974472198548501,
+      "loss": 1.3436,
+      "step": 1620
+    },
+    {
+      "epoch": 0.1422744938849286,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0029743900459555327,
+      "loss": 1.3593,
+      "step": 1621
+    },
+    {
+      "epoch": 0.14236226346783107,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.002974307762650507,
+      "loss": 1.3492,
+      "step": 1622
+    },
+    {
+      "epoch": 0.14245003305073356,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0029742253486415445,
+      "loss": 1.3053,
+      "step": 1623
+    },
+    {
+      "epoch": 0.14253780263363605,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.002974142803936781,
+      "loss": 1.3863,
+      "step": 1624
+    },
+    {
+      "epoch": 0.14262557221653854,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.002974060128544361,
+      "loss": 1.2817,
+      "step": 1625
+    },
+    {
+      "epoch": 0.142713341799441,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002973977322472446,
+      "loss": 1.3204,
+      "step": 1626
+    },
+    {
+      "epoch": 0.1428011113823435,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0029738943857292076,
+      "loss": 1.3363,
+      "step": 1627
+    },
+    {
+      "epoch": 0.142888880965246,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002973811318322832,
+      "loss": 1.3195,
+      "step": 1628
+    },
+    {
+      "epoch": 0.14297665054814848,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0029737281202615175,
+      "loss": 1.287,
+      "step": 1629
+    },
+    {
+      "epoch": 0.14306442013105095,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.002973644791553475,
+      "loss": 1.3488,
+      "step": 1630
+    },
+    {
+      "epoch": 0.14315218971395344,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0029735613322069286,
+      "loss": 1.3186,
+      "step": 1631
+    },
+    {
+      "epoch": 0.14323995929685593,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0029734777422301168,
+      "loss": 1.3029,
+      "step": 1632
+    },
+    {
+      "epoch": 0.14332772887975842,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0029733940216312878,
+      "loss": 1.3344,
+      "step": 1633
+    },
+    {
+      "epoch": 0.1434154984626609,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0029733101704187055,
+      "loss": 1.3609,
+      "step": 1634
+    },
+    {
+      "epoch": 0.14350326804556338,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0029732261886006453,
+      "loss": 1.3649,
+      "step": 1635
+    },
+    {
+      "epoch": 0.14359103762846587,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029731420761853963,
+      "loss": 1.3182,
+      "step": 1636
+    },
+    {
+      "epoch": 0.14367880721136836,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0029730578331812595,
+      "loss": 1.3874,
+      "step": 1637
+    },
+    {
+      "epoch": 0.14376657679427085,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00297297345959655,
+      "loss": 1.3422,
+      "step": 1638
+    },
+    {
+      "epoch": 0.14385434637717331,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.002972888955439594,
+      "loss": 1.4337,
+      "step": 1639
+    },
+    {
+      "epoch": 0.1439421159600758,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.002972804320718733,
+      "loss": 1.3749,
+      "step": 1640
+    },
+    {
+      "epoch": 0.1440298855429783,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0029727195554423197,
+      "loss": 1.3506,
+      "step": 1641
+    },
+    {
+      "epoch": 0.1441176551258808,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0029726346596187204,
+      "loss": 1.3125,
+      "step": 1642
+    },
+    {
+      "epoch": 0.14420542470878325,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0029725496332563134,
+      "loss": 1.3447,
+      "step": 1643
+    },
+    {
+      "epoch": 0.14429319429168574,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.00297246447636349,
+      "loss": 1.2968,
+      "step": 1644
+    },
+    {
+      "epoch": 0.14438096387458824,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.002972379188948656,
+      "loss": 1.315,
+      "step": 1645
+    },
+    {
+      "epoch": 0.14446873345749073,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.002972293771020228,
+      "loss": 1.3592,
+      "step": 1646
+    },
+    {
+      "epoch": 0.14455650304039322,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0029722082225866375,
+      "loss": 1.3075,
+      "step": 1647
+    },
+    {
+      "epoch": 0.14464427262329568,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.002972122543656327,
+      "loss": 1.3091,
+      "step": 1648
+    },
+    {
+      "epoch": 0.14473204220619817,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.0029720367342377526,
+      "loss": 1.3056,
+      "step": 1649
+    },
+    {
+      "epoch": 0.14481981178910067,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.002971950794339384,
+      "loss": 1.3202,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14490758137200316,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.002971864723969702,
+      "loss": 1.3596,
+      "step": 1651
+    },
+    {
+      "epoch": 0.14499535095490562,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.002971778523137203,
+      "loss": 1.3292,
+      "step": 1652
+    },
+    {
+      "epoch": 0.1450831205378081,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002971692191850393,
+      "loss": 1.3106,
+      "step": 1653
+    },
+    {
+      "epoch": 0.1451708901207106,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0029716057301177935,
+      "loss": 1.3506,
+      "step": 1654
+    },
+    {
+      "epoch": 0.1452586597036131,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.002971519137947938,
+      "loss": 1.3225,
+      "step": 1655
+    },
+    {
+      "epoch": 0.1453464292865156,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029714324153493725,
+      "loss": 1.3785,
+      "step": 1656
+    },
+    {
+      "epoch": 0.14543419886941805,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0029713455623306562,
+      "loss": 1.328,
+      "step": 1657
+    },
+    {
+      "epoch": 0.14552196845232054,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.002971258578900361,
+      "loss": 1.3539,
+      "step": 1658
+    },
+    {
+      "epoch": 0.14560973803522304,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.002971171465067072,
+      "loss": 1.3286,
+      "step": 1659
+    },
+    {
+      "epoch": 0.14569750761812553,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0029710842208393873,
+      "loss": 1.3292,
+      "step": 1660
+    },
+    {
+      "epoch": 0.145785277201028,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0029709968462259175,
+      "loss": 1.3318,
+      "step": 1661
+    },
+    {
+      "epoch": 0.14587304678393048,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0029709093412352856,
+      "loss": 1.304,
+      "step": 1662
+    },
+    {
+      "epoch": 0.14596081636683297,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002970821705876128,
+      "loss": 1.3398,
+      "step": 1663
+    },
+    {
+      "epoch": 0.14604858594973547,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0029707339401570947,
+      "loss": 1.3355,
+      "step": 1664
+    },
+    {
+      "epoch": 0.14613635553263793,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.002970646044086847,
+      "loss": 1.3437,
+      "step": 1665
+    },
+    {
+      "epoch": 0.14622412511554042,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0029705580176740604,
+      "loss": 1.3,
+      "step": 1666
+    },
+    {
+      "epoch": 0.1463118946984429,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0029704698609274226,
+      "loss": 1.3435,
+      "step": 1667
+    },
+    {
+      "epoch": 0.1463996642813454,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0029703815738556347,
+      "loss": 1.3546,
+      "step": 1668
+    },
+    {
+      "epoch": 0.1464874338642479,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.00297029315646741,
+      "loss": 1.2785,
+      "step": 1669
+    },
+    {
+      "epoch": 0.14657520344715036,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0029702046087714744,
+      "loss": 1.3239,
+      "step": 1670
+    },
+    {
+      "epoch": 0.14666297303005285,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.002970115930776568,
+      "loss": 1.3436,
+      "step": 1671
+    },
+    {
+      "epoch": 0.14675074261295534,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.002970027122491443,
+      "loss": 1.3654,
+      "step": 1672
+    },
+    {
+      "epoch": 0.14683851219585783,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0029699381839248633,
+      "loss": 1.3356,
+      "step": 1673
+    },
+    {
+      "epoch": 0.1469262817787603,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0029698491150856082,
+      "loss": 1.3234,
+      "step": 1674
+    },
+    {
+      "epoch": 0.1470140513616628,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0029697599159824676,
+      "loss": 1.3224,
+      "step": 1675
+    },
+    {
+      "epoch": 0.14710182094456528,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.0029696705866242455,
+      "loss": 1.3654,
+      "step": 1676
+    },
+    {
+      "epoch": 0.14718959052746777,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0029695811270197578,
+      "loss": 1.2866,
+      "step": 1677
+    },
+    {
+      "epoch": 0.14727736011037024,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.002969491537177834,
+      "loss": 1.3362,
+      "step": 1678
+    },
+    {
+      "epoch": 0.14736512969327273,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.002969401817107317,
+      "loss": 1.337,
+      "step": 1679
+    },
+    {
+      "epoch": 0.14745289927617522,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0029693119668170614,
+      "loss": 1.4001,
+      "step": 1680
+    },
+    {
+      "epoch": 0.1475406688590777,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0029692219863159345,
+      "loss": 1.3351,
+      "step": 1681
+    },
+    {
+      "epoch": 0.1476284384419802,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0029691318756128173,
+      "loss": 1.3084,
+      "step": 1682
+    },
+    {
+      "epoch": 0.14771620802488267,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.002969041634716604,
+      "loss": 1.3714,
+      "step": 1683
+    },
+    {
+      "epoch": 0.14780397760778516,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0029689512636362006,
+      "loss": 1.2869,
+      "step": 1684
+    },
+    {
+      "epoch": 0.14789174719068765,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.002968860762380526,
+      "loss": 1.3356,
+      "step": 1685
+    },
+    {
+      "epoch": 0.14797951677359014,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.0029687701309585126,
+      "loss": 1.3372,
+      "step": 1686
+    },
+    {
+      "epoch": 0.1480672863564926,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0029686793693791055,
+      "loss": 1.3333,
+      "step": 1687
+    },
+    {
+      "epoch": 0.1481550559393951,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.002968588477651262,
+      "loss": 1.3643,
+      "step": 1688
+    },
+    {
+      "epoch": 0.1482428255222976,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0029684974557839536,
+      "loss": 1.3312,
+      "step": 1689
+    },
+    {
+      "epoch": 0.14833059510520008,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002968406303786163,
+      "loss": 1.3926,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14841836468810257,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0029683150216668875,
+      "loss": 1.3173,
+      "step": 1691
+    },
+    {
+      "epoch": 0.14850613427100504,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.002968223609435135,
+      "loss": 1.3306,
+      "step": 1692
+    },
+    {
+      "epoch": 0.14859390385390753,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0029681320670999285,
+      "loss": 1.3116,
+      "step": 1693
+    },
+    {
+      "epoch": 0.14868167343681002,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0029680403946703024,
+      "loss": 1.3639,
+      "step": 1694
+    },
+    {
+      "epoch": 0.1487694430197125,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0029679485921553046,
+      "loss": 1.38,
+      "step": 1695
+    },
+    {
+      "epoch": 0.14885721260261497,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0029678566595639964,
+      "loss": 1.293,
+      "step": 1696
+    },
+    {
+      "epoch": 0.14894498218551747,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0029677645969054495,
+      "loss": 1.3172,
+      "step": 1697
+    },
+    {
+      "epoch": 0.14903275176841996,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0029676724041887512,
+      "loss": 1.3452,
+      "step": 1698
+    },
+    {
+      "epoch": 0.14912052135132245,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0029675800814230003,
+      "loss": 1.3612,
+      "step": 1699
+    },
+    {
+      "epoch": 0.1492082909342249,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0029674876286173092,
+      "loss": 1.3398,
+      "step": 1700
+    },
+    {
+      "epoch": 0.1492960605171274,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0029673950457808017,
+      "loss": 1.3592,
+      "step": 1701
+    },
+    {
+      "epoch": 0.1493838301000299,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.002967302332922616,
+      "loss": 1.375,
+      "step": 1702
+    },
+    {
+      "epoch": 0.1494715996829324,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.002967209490051903,
+      "loss": 1.3833,
+      "step": 1703
+    },
+    {
+      "epoch": 0.14955936926583488,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002967116517177824,
+      "loss": 1.3336,
+      "step": 1704
+    },
+    {
+      "epoch": 0.14964713884873734,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0029670234143095568,
+      "loss": 1.3037,
+      "step": 1705
+    },
+    {
+      "epoch": 0.14973490843163983,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00296693018145629,
+      "loss": 1.3435,
+      "step": 1706
+    },
+    {
+      "epoch": 0.14982267801454233,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.002966836818627225,
+      "loss": 1.4144,
+      "step": 1707
+    },
+    {
+      "epoch": 0.14991044759744482,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.002966743325831577,
+      "loss": 1.3661,
+      "step": 1708
+    },
+    {
+      "epoch": 0.14999821718034728,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0029666497030785724,
+      "loss": 1.3676,
+      "step": 1709
+    },
+    {
+      "epoch": 0.15008598676324977,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029665559503774517,
+      "loss": 1.3522,
+      "step": 1710
+    },
+    {
+      "epoch": 0.15017375634615227,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.002966462067737468,
+      "loss": 1.3386,
+      "step": 1711
+    },
+    {
+      "epoch": 0.15026152592905476,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0029663680551678877,
+      "loss": 1.3055,
+      "step": 1712
+    },
+    {
+      "epoch": 0.15034929551195722,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 0.002966273912677988,
+      "loss": 1.3126,
+      "step": 1713
+    },
+    {
+      "epoch": 0.1504370650948597,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.002966179640277062,
+      "loss": 1.3661,
+      "step": 1714
+    },
+    {
+      "epoch": 0.1505248346777622,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.002966085237974414,
+      "loss": 1.3166,
+      "step": 1715
+    },
+    {
+      "epoch": 0.1506126042606647,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.00296599070577936,
+      "loss": 1.2842,
+      "step": 1716
+    },
+    {
+      "epoch": 0.1507003738435672,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0029658960437012297,
+      "loss": 1.3629,
+      "step": 1717
+    },
+    {
+      "epoch": 0.15078814342646965,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002965801251749367,
+      "loss": 1.2343,
+      "step": 1718
+    },
+    {
+      "epoch": 0.15087591300937214,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0029657063299331275,
+      "loss": 1.3709,
+      "step": 1719
+    },
+    {
+      "epoch": 0.15096368259227463,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.002965611278261879,
+      "loss": 1.3609,
+      "step": 1720
+    },
+    {
+      "epoch": 0.15105145217517713,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0029655160967450036,
+      "loss": 1.3305,
+      "step": 1721
+    },
+    {
+      "epoch": 0.1511392217580796,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0029654207853918937,
+      "loss": 1.3748,
+      "step": 1722
+    },
+    {
+      "epoch": 0.15122699134098208,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.0029653253442119575,
+      "loss": 1.3154,
+      "step": 1723
+    },
+    {
+      "epoch": 0.15131476092388457,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.002965229773214614,
+      "loss": 1.3339,
+      "step": 1724
+    },
+    {
+      "epoch": 0.15140253050678706,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.002965134072409297,
+      "loss": 1.2991,
+      "step": 1725
+    },
+    {
+      "epoch": 0.15149030008968956,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.00296503824180545,
+      "loss": 1.3629,
+      "step": 1726
+    },
+    {
+      "epoch": 0.15157806967259202,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029649422814125317,
+      "loss": 1.3489,
+      "step": 1727
+    },
+    {
+      "epoch": 0.1516658392554945,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.0029648461912400136,
+      "loss": 1.2968,
+      "step": 1728
+    },
+    {
+      "epoch": 0.151753608838397,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002964749971297379,
+      "loss": 1.312,
+      "step": 1729
+    },
+    {
+      "epoch": 0.1518413784212995,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002964653621594124,
+      "loss": 1.3319,
+      "step": 1730
+    },
+    {
+      "epoch": 0.15192914800420196,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.0029645571421397588,
+      "loss": 1.3371,
+      "step": 1731
+    },
+    {
+      "epoch": 0.15201691758710445,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.002964460532943805,
+      "loss": 1.3227,
+      "step": 1732
+    },
+    {
+      "epoch": 0.15210468717000694,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.002964363794015798,
+      "loss": 1.3408,
+      "step": 1733
+    },
+    {
+      "epoch": 0.15219245675290943,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029642669253652845,
+      "loss": 1.285,
+      "step": 1734
+    },
+    {
+      "epoch": 0.1522802263358119,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.002964169927001826,
+      "loss": 1.3042,
+      "step": 1735
+    },
+    {
+      "epoch": 0.1523679959187144,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0029640727989349957,
+      "loss": 1.3264,
+      "step": 1736
+    },
+    {
+      "epoch": 0.15245576550161688,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0029639755411743794,
+      "loss": 1.3388,
+      "step": 1737
+    },
+    {
+      "epoch": 0.15254353508451937,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.002963878153729577,
+      "loss": 1.2758,
+      "step": 1738
+    },
+    {
+      "epoch": 0.15263130466742186,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.002963780636610199,
+      "loss": 1.3148,
+      "step": 1739
+    },
+    {
+      "epoch": 0.15271907425032433,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.002963682989825871,
+      "loss": 1.3109,
+      "step": 1740
+    },
+    {
+      "epoch": 0.15280684383322682,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.002963585213386229,
+      "loss": 1.3591,
+      "step": 1741
+    },
+    {
+      "epoch": 0.1528946134161293,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0029634873073009255,
+      "loss": 1.3326,
+      "step": 1742
+    },
+    {
+      "epoch": 0.1529823829990318,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0029633892715796206,
+      "loss": 1.4084,
+      "step": 1743
+    },
+    {
+      "epoch": 0.15307015258193427,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0029632911062319922,
+      "loss": 1.3302,
+      "step": 1744
+    },
+    {
+      "epoch": 0.15315792216483676,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.002963192811267728,
+      "loss": 1.323,
+      "step": 1745
+    },
+    {
+      "epoch": 0.15324569174773925,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.002963094386696529,
+      "loss": 1.3406,
+      "step": 1746
+    },
+    {
+      "epoch": 0.15333346133064174,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0029629958325281098,
+      "loss": 1.3364,
+      "step": 1747
+    },
+    {
+      "epoch": 0.1534212309135442,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.002962897148772197,
+      "loss": 1.3756,
+      "step": 1748
+    },
+    {
+      "epoch": 0.1535090004964467,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.002962798335438531,
+      "loss": 1.3815,
+      "step": 1749
+    },
+    {
+      "epoch": 0.1535967700793492,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0029626993925368633,
+      "loss": 1.2959,
+      "step": 1750
+    },
+    {
+      "epoch": 0.15368453966225168,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0029626003200769603,
+      "loss": 1.3421,
+      "step": 1751
+    },
+    {
+      "epoch": 0.15377230924515417,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0029625011180685985,
+      "loss": 1.2874,
+      "step": 1752
+    },
+    {
+      "epoch": 0.15386007882805663,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.00296240178652157,
+      "loss": 1.4442,
+      "step": 1753
+    },
+    {
+      "epoch": 0.15394784841095913,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0029623023254456786,
+      "loss": 1.3041,
+      "step": 1754
+    },
+    {
+      "epoch": 0.15403561799386162,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0029622027348507394,
+      "loss": 1.3858,
+      "step": 1755
+    },
+    {
+      "epoch": 0.1541233875767641,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.002962103014746583,
+      "loss": 1.2826,
+      "step": 1756
+    },
+    {
+      "epoch": 0.15421115715966657,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002962003165143051,
+      "loss": 1.3288,
+      "step": 1757
+    },
+    {
+      "epoch": 0.15429892674256906,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0029619031860499976,
+      "loss": 1.3348,
+      "step": 1758
+    },
+    {
+      "epoch": 0.15438669632547156,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00296180307747729,
+      "loss": 1.3248,
+      "step": 1759
+    },
+    {
+      "epoch": 0.15447446590837405,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.00296170283943481,
+      "loss": 1.3364,
+      "step": 1760
+    },
+    {
+      "epoch": 0.15456223549127654,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0029616024719324495,
+      "loss": 1.2944,
+      "step": 1761
+    },
+    {
+      "epoch": 0.154650005074179,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.002961501974980115,
+      "loss": 1.3995,
+      "step": 1762
+    },
+    {
+      "epoch": 0.1547377746570815,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.002961401348587725,
+      "loss": 1.3033,
+      "step": 1763
+    },
+    {
+      "epoch": 0.154825544239984,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0029613005927652107,
+      "loss": 1.3228,
+      "step": 1764
+    },
+    {
+      "epoch": 0.15491331382288648,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0029611997075225166,
+      "loss": 1.3379,
+      "step": 1765
+    },
+    {
+      "epoch": 0.15500108340578894,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002961098692869599,
+      "loss": 1.3414,
+      "step": 1766
+    },
+    {
+      "epoch": 0.15508885298869143,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0029609975488164284,
+      "loss": 1.2882,
+      "step": 1767
+    },
+    {
+      "epoch": 0.15517662257159393,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.002960896275372987,
+      "loss": 1.3297,
+      "step": 1768
+    },
+    {
+      "epoch": 0.15526439215449642,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.002960794872549271,
+      "loss": 1.3558,
+      "step": 1769
+    },
+    {
+      "epoch": 0.15535216173739888,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.002960693340355287,
+      "loss": 1.292,
+      "step": 1770
+    },
+    {
+      "epoch": 0.15543993132030137,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0029605916788010564,
+      "loss": 1.2973,
+      "step": 1771
+    },
+    {
+      "epoch": 0.15552770090320386,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0029604898878966127,
+      "loss": 1.3015,
+      "step": 1772
+    },
+    {
+      "epoch": 0.15561547048610636,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.002960387967652003,
+      "loss": 1.3052,
+      "step": 1773
+    },
+    {
+      "epoch": 0.15570324006900885,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002960285918077285,
+      "loss": 1.3675,
+      "step": 1774
+    },
+    {
+      "epoch": 0.1557910096519113,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0029601837391825324,
+      "loss": 1.3843,
+      "step": 1775
+    },
+    {
+      "epoch": 0.1558787792348138,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0029600814309778288,
+      "loss": 1.3903,
+      "step": 1776
+    },
+    {
+      "epoch": 0.1559665488177163,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.002959978993473271,
+      "loss": 1.3131,
+      "step": 1777
+    },
+    {
+      "epoch": 0.15605431840061879,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.00295987642667897,
+      "loss": 1.368,
+      "step": 1778
+    },
+    {
+      "epoch": 0.15614208798352125,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.002959773730605049,
+      "loss": 1.3273,
+      "step": 1779
+    },
+    {
+      "epoch": 0.15622985756642374,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0029596709052616427,
+      "loss": 1.2855,
+      "step": 1780
+    },
+    {
+      "epoch": 0.15631762714932623,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0029595679506589007,
+      "loss": 1.3126,
+      "step": 1781
+    },
+    {
+      "epoch": 0.15640539673222872,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0029594648668069836,
+      "loss": 1.3176,
+      "step": 1782
+    },
+    {
+      "epoch": 0.1564931663151312,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0029593616537160654,
+      "loss": 1.2997,
+      "step": 1783
+    },
+    {
+      "epoch": 0.15658093589803368,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0029592583113963325,
+      "loss": 1.3273,
+      "step": 1784
+    },
+    {
+      "epoch": 0.15666870548093617,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.002959154839857985,
+      "loss": 1.2225,
+      "step": 1785
+    },
+    {
+      "epoch": 0.15675647506383866,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.0029590512391112355,
+      "loss": 1.3383,
+      "step": 1786
+    },
+    {
+      "epoch": 0.15684424464674115,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.002958947509166307,
+      "loss": 1.2893,
+      "step": 1787
+    },
+    {
+      "epoch": 0.15693201422964362,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.00295884365003344,
+      "loss": 1.3182,
+      "step": 1788
+    },
+    {
+      "epoch": 0.1570197838125461,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002958739661722883,
+      "loss": 1.3575,
+      "step": 1789
+    },
+    {
+      "epoch": 0.1571075533954486,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0029586355442448994,
+      "loss": 1.3134,
+      "step": 1790
+    },
+    {
+      "epoch": 0.1571953229783511,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.002958531297609766,
+      "loss": 1.2641,
+      "step": 1791
+    },
+    {
+      "epoch": 0.15728309256125356,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.002958426921827771,
+      "loss": 1.3329,
+      "step": 1792
+    },
+    {
+      "epoch": 0.15737086214415605,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.002958322416909216,
+      "loss": 1.3176,
+      "step": 1793
+    },
+    {
+      "epoch": 0.15745863172705854,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.002958217782864415,
+      "loss": 1.3579,
+      "step": 1794
+    },
+    {
+      "epoch": 0.15754640130996103,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.002958113019703696,
+      "loss": 1.3458,
+      "step": 1795
+    },
+    {
+      "epoch": 0.15763417089286352,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029580081274373973,
+      "loss": 1.3286,
+      "step": 1796
+    },
+    {
+      "epoch": 0.157721940475766,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.002957903106075872,
+      "loss": 1.2778,
+      "step": 1797
+    },
+    {
+      "epoch": 0.15780971005866848,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0029577979556294856,
+      "loss": 1.4001,
+      "step": 1798
+    },
+    {
+      "epoch": 0.15789747964157097,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002957692676108615,
+      "loss": 1.3583,
+      "step": 1799
+    },
+    {
+      "epoch": 0.15798524922447346,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0029575872675236523,
+      "loss": 1.3638,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15807301880737593,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0029574817298849993,
+      "loss": 1.3447,
+      "step": 1801
+    },
+    {
+      "epoch": 0.15816078839027842,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0029573760632030737,
+      "loss": 1.3347,
+      "step": 1802
+    },
+    {
+      "epoch": 0.1582485579731809,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0029572702674883033,
+      "loss": 1.338,
+      "step": 1803
+    },
+    {
+      "epoch": 0.1583363275560834,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0029571643427511304,
+      "loss": 1.3664,
+      "step": 1804
+    },
+    {
+      "epoch": 0.15842409713898586,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0029570582890020088,
+      "loss": 1.3042,
+      "step": 1805
+    },
+    {
+      "epoch": 0.15851186672188836,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.002956952106251406,
+      "loss": 1.3055,
+      "step": 1806
+    },
+    {
+      "epoch": 0.15859963630479085,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0029568457945098017,
+      "loss": 1.3305,
+      "step": 1807
+    },
+    {
+      "epoch": 0.15868740588769334,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0029567393537876883,
+      "loss": 1.2891,
+      "step": 1808
+    },
+    {
+      "epoch": 0.15877517547059583,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029566327840955713,
+      "loss": 1.3102,
+      "step": 1809
+    },
+    {
+      "epoch": 0.1588629450534983,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.0029565260854439684,
+      "loss": 1.3204,
+      "step": 1810
+    },
+    {
+      "epoch": 0.1589507146364008,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0029564192578434103,
+      "loss": 1.3443,
+      "step": 1811
+    },
+    {
+      "epoch": 0.15903848421930328,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0029563123013044415,
+      "loss": 1.3179,
+      "step": 1812
+    },
+    {
+      "epoch": 0.15912625380220577,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.002956205215837617,
+      "loss": 1.3436,
+      "step": 1813
+    },
+    {
+      "epoch": 0.15921402338510823,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.0029560980014535056,
+      "loss": 1.3426,
+      "step": 1814
+    },
+    {
+      "epoch": 0.15930179296801072,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.00295599065816269,
+      "loss": 1.3132,
+      "step": 1815
+    },
+    {
+      "epoch": 0.15938956255091322,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002955883185975764,
+      "loss": 1.2819,
+      "step": 1816
+    },
+    {
+      "epoch": 0.1594773321338157,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.002955775584903334,
+      "loss": 1.276,
+      "step": 1817
+    },
+    {
+      "epoch": 0.15956510171671817,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0029556678549560213,
+      "loss": 1.2953,
+      "step": 1818
+    },
+    {
+      "epoch": 0.15965287129962066,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0029555599961444577,
+      "loss": 1.3201,
+      "step": 1819
+    },
+    {
+      "epoch": 0.15974064088252315,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0029554520084792875,
+      "loss": 1.3501,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15982841046542565,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.00295534389197117,
+      "loss": 1.3125,
+      "step": 1821
+    },
+    {
+      "epoch": 0.15991618004832814,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.002955235646630775,
+      "loss": 1.3271,
+      "step": 1822
+    },
+    {
+      "epoch": 0.1600039496312306,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0029551272724687863,
+      "loss": 1.3424,
+      "step": 1823
+    },
+    {
+      "epoch": 0.1600917192141331,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0029550187694959,
+      "loss": 1.3016,
+      "step": 1824
+    },
+    {
+      "epoch": 0.16017948879703559,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0029549101377228247,
+      "loss": 1.3332,
+      "step": 1825
+    },
+    {
+      "epoch": 0.16026725837993808,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0029548013771602826,
+      "loss": 1.2631,
+      "step": 1826
+    },
+    {
+      "epoch": 0.16035502796284054,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0029546924878190066,
+      "loss": 1.2696,
+      "step": 1827
+    },
+    {
+      "epoch": 0.16044279754574303,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0029545834697097453,
+      "loss": 1.2918,
+      "step": 1828
+    },
+    {
+      "epoch": 0.16053056712864552,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0029544743228432572,
+      "loss": 1.2892,
+      "step": 1829
+    },
+    {
+      "epoch": 0.16061833671154802,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0029543650472303147,
+      "loss": 1.3321,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1607061062944505,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.002954255642881704,
+      "loss": 1.3393,
+      "step": 1831
+    },
+    {
+      "epoch": 0.16079387587735297,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002954146109808221,
+      "loss": 1.3037,
+      "step": 1832
+    },
+    {
+      "epoch": 0.16088164546025546,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0029540364480206776,
+      "loss": 1.3358,
+      "step": 1833
+    },
+    {
+      "epoch": 0.16096941504315795,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.002953926657529897,
+      "loss": 1.2961,
+      "step": 1834
+    },
+    {
+      "epoch": 0.16105718462606045,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0029538167383467142,
+      "loss": 1.3394,
+      "step": 1835
+    },
+    {
+      "epoch": 0.1611449542089629,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0029537066904819784,
+      "loss": 1.3531,
+      "step": 1836
+    },
+    {
+      "epoch": 0.1612327237918654,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0029535965139465504,
+      "loss": 1.2862,
+      "step": 1837
+    },
+    {
+      "epoch": 0.1613204933747679,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.002953486208751305,
+      "loss": 1.3085,
+      "step": 1838
+    },
+    {
+      "epoch": 0.16140826295767038,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0029533757749071283,
+      "loss": 1.3163,
+      "step": 1839
+    },
+    {
+      "epoch": 0.16149603254057285,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0029532652124249197,
+      "loss": 1.2983,
+      "step": 1840
+    },
+    {
+      "epoch": 0.16158380212347534,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0029531545213155918,
+      "loss": 1.3236,
+      "step": 1841
+    },
+    {
+      "epoch": 0.16167157170637783,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.002953043701590069,
+      "loss": 1.3114,
+      "step": 1842
+    },
+    {
+      "epoch": 0.16175934128928032,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.002952932753259288,
+      "loss": 1.3014,
+      "step": 1843
+    },
+    {
+      "epoch": 0.16184711087218281,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0029528216763342,
+      "loss": 1.3481,
+      "step": 1844
+    },
+    {
+      "epoch": 0.16193488045508528,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0029527104708257676,
+      "loss": 1.2856,
+      "step": 1845
+    },
+    {
+      "epoch": 0.16202265003798777,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0029525991367449667,
+      "loss": 1.3673,
+      "step": 1846
+    },
+    {
+      "epoch": 0.16211041962089026,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.0029524876741027846,
+      "loss": 1.3396,
+      "step": 1847
+    },
+    {
+      "epoch": 0.16219818920379275,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.002952376082910223,
+      "loss": 1.2964,
+      "step": 1848
+    },
+    {
+      "epoch": 0.16228595878669522,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.002952264363178295,
+      "loss": 1.2986,
+      "step": 1849
+    },
+    {
+      "epoch": 0.1623737283695977,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0029521525149180275,
+      "loss": 1.349,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1624614979525002,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.002952040538140459,
+      "loss": 1.3361,
+      "step": 1851
+    },
+    {
+      "epoch": 0.1625492675354027,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.002951928432856641,
+      "loss": 1.2789,
+      "step": 1852
+    },
+    {
+      "epoch": 0.16263703711830516,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.002951816199077639,
+      "loss": 1.3311,
+      "step": 1853
+    },
+    {
+      "epoch": 0.16272480670120765,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0029517038368145287,
+      "loss": 1.3419,
+      "step": 1854
+    },
+    {
+      "epoch": 0.16281257628411014,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0029515913460784,
+      "loss": 1.2982,
+      "step": 1855
+    },
+    {
+      "epoch": 0.16290034586701263,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0029514787268803554,
+      "loss": 1.3098,
+      "step": 1856
+    },
+    {
+      "epoch": 0.16298811544991512,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029513659792315106,
+      "loss": 1.2828,
+      "step": 1857
+    },
+    {
+      "epoch": 0.16307588503281759,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0029512531031429933,
+      "loss": 1.3173,
+      "step": 1858
+    },
+    {
+      "epoch": 0.16316365461572008,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0029511400986259426,
+      "loss": 1.3546,
+      "step": 1859
+    },
+    {
+      "epoch": 0.16325142419862257,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002951026965691513,
+      "loss": 1.3284,
+      "step": 1860
+    },
+    {
+      "epoch": 0.16333919378152506,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.00295091370435087,
+      "loss": 1.3001,
+      "step": 1861
+    },
+    {
+      "epoch": 0.16342696336442752,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0029508003146151915,
+      "loss": 1.3214,
+      "step": 1862
+    },
+    {
+      "epoch": 0.16351473294733002,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0029506867964956694,
+      "loss": 1.2816,
+      "step": 1863
+    },
+    {
+      "epoch": 0.1636025025302325,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0029505731500035063,
+      "loss": 1.3581,
+      "step": 1864
+    },
+    {
+      "epoch": 0.163690272113135,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0029504593751499197,
+      "loss": 1.33,
+      "step": 1865
+    },
+    {
+      "epoch": 0.1637780416960375,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.002950345471946139,
+      "loss": 1.318,
+      "step": 1866
+    },
+    {
+      "epoch": 0.16386581127893995,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0029502314404034045,
+      "loss": 1.2866,
+      "step": 1867
+    },
+    {
+      "epoch": 0.16395358086184245,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.002950117280532972,
+      "loss": 1.3199,
+      "step": 1868
+    },
+    {
+      "epoch": 0.16404135044474494,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0029500029923461087,
+      "loss": 1.307,
+      "step": 1869
+    },
+    {
+      "epoch": 0.16412912002764743,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0029498885758540935,
+      "loss": 1.2498,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1642168896105499,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0029497740310682192,
+      "loss": 1.3835,
+      "step": 1871
+    },
+    {
+      "epoch": 0.16430465919345238,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002949659357999791,
+      "loss": 1.3665,
+      "step": 1872
+    },
+    {
+      "epoch": 0.16439242877635488,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.0029495445566601265,
+      "loss": 1.3354,
+      "step": 1873
+    },
+    {
+      "epoch": 0.16448019835925737,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0029494296270605566,
+      "loss": 1.3732,
+      "step": 1874
+    },
+    {
+      "epoch": 0.16456796794215983,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.002949314569212424,
+      "loss": 1.3693,
+      "step": 1875
+    },
+    {
+      "epoch": 0.16465573752506232,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.0029491993831270847,
+      "loss": 1.3408,
+      "step": 1876
+    },
+    {
+      "epoch": 0.16474350710796482,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.002949084068815907,
+      "loss": 1.3244,
+      "step": 1877
+    },
+    {
+      "epoch": 0.1648312766908673,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.002948968626290271,
+      "loss": 1.32,
+      "step": 1878
+    },
+    {
+      "epoch": 0.1649190462737698,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0029488530555615722,
+      "loss": 1.2796,
+      "step": 1879
+    },
+    {
+      "epoch": 0.16500681585667226,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0029487373566412156,
+      "loss": 1.3104,
+      "step": 1880
+    },
+    {
+      "epoch": 0.16509458543957475,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0029486215295406216,
+      "loss": 1.3216,
+      "step": 1881
+    },
+    {
+      "epoch": 0.16518235502247725,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.0029485055742712206,
+      "loss": 1.3015,
+      "step": 1882
+    },
+    {
+      "epoch": 0.16527012460537974,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.002948389490844457,
+      "loss": 1.3027,
+      "step": 1883
+    },
+    {
+      "epoch": 0.1653578941882822,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002948273279271789,
+      "loss": 1.2722,
+      "step": 1884
+    },
+    {
+      "epoch": 0.1654456637711847,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0029481569395646843,
+      "loss": 1.2734,
+      "step": 1885
+    },
+    {
+      "epoch": 0.16553343335408718,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.002948040471734627,
+      "loss": 1.3116,
+      "step": 1886
+    },
+    {
+      "epoch": 0.16562120293698968,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.00294792387579311,
+      "loss": 1.2624,
+      "step": 1887
+    },
+    {
+      "epoch": 0.16570897251989214,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0029478071517516436,
+      "loss": 1.3011,
+      "step": 1888
+    },
+    {
+      "epoch": 0.16579674210279463,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002947690299621746,
+      "loss": 1.3676,
+      "step": 1889
+    },
+    {
+      "epoch": 0.16588451168569712,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0029475733194149504,
+      "loss": 1.3082,
+      "step": 1890
+    },
+    {
+      "epoch": 0.16597228126859961,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0029474562111428022,
+      "loss": 1.3579,
+      "step": 1891
+    },
+    {
+      "epoch": 0.1660600508515021,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.00294733897481686,
+      "loss": 1.3564,
+      "step": 1892
+    },
+    {
+      "epoch": 0.16614782043440457,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002947221610448694,
+      "loss": 1.3186,
+      "step": 1893
+    },
+    {
+      "epoch": 0.16623559001730706,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.002947104118049888,
+      "loss": 1.3272,
+      "step": 1894
+    },
+    {
+      "epoch": 0.16632335960020955,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0029469864976320384,
+      "loss": 1.2855,
+      "step": 1895
+    },
+    {
+      "epoch": 0.16641112918311204,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.002946868749206753,
+      "loss": 1.3089,
+      "step": 1896
+    },
+    {
+      "epoch": 0.1664988987660145,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.002946750872785654,
+      "loss": 1.2869,
+      "step": 1897
+    },
+    {
+      "epoch": 0.166586668348917,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.002946632868380374,
+      "loss": 1.38,
+      "step": 1898
+    },
+    {
+      "epoch": 0.1666744379318195,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.002946514736002561,
+      "loss": 1.3421,
+      "step": 1899
+    },
+    {
+      "epoch": 0.16676220751472198,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.002946396475663873,
+      "loss": 1.2534,
+      "step": 1900
+    },
+    {
+      "epoch": 0.16684997709762447,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0029462780873759833,
+      "loss": 1.3187,
+      "step": 1901
+    },
+    {
+      "epoch": 0.16693774668052694,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0029461595711505746,
+      "loss": 1.312,
+      "step": 1902
+    },
+    {
+      "epoch": 0.16702551626342943,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0029460409269993454,
+      "loss": 1.2773,
+      "step": 1903
+    },
+    {
+      "epoch": 0.16711328584633192,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002945922154934005,
+      "loss": 1.3131,
+      "step": 1904
+    },
+    {
+      "epoch": 0.1672010554292344,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0029458032549662754,
+      "loss": 1.3198,
+      "step": 1905
+    },
+    {
+      "epoch": 0.16728882501213688,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.0029456842271078916,
+      "loss": 1.3323,
+      "step": 1906
+    },
+    {
+      "epoch": 0.16737659459503937,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0029455650713706015,
+      "loss": 1.3593,
+      "step": 1907
+    },
+    {
+      "epoch": 0.16746436417794186,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.0029454457877661653,
+      "loss": 1.2929,
+      "step": 1908
+    },
+    {
+      "epoch": 0.16755213376084435,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0029453263763063557,
+      "loss": 1.3747,
+      "step": 1909
+    },
+    {
+      "epoch": 0.16763990334374682,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.002945206837002958,
+      "loss": 1.3061,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1677276729266493,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0029450871698677703,
+      "loss": 1.2855,
+      "step": 1911
+    },
+    {
+      "epoch": 0.1678154425095518,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0029449673749126033,
+      "loss": 1.312,
+      "step": 1912
+    },
+    {
+      "epoch": 0.1679032120924543,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0029448474521492803,
+      "loss": 1.3613,
+      "step": 1913
+    },
+    {
+      "epoch": 0.16799098167535678,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0029447274015896373,
+      "loss": 1.345,
+      "step": 1914
+    },
+    {
+      "epoch": 0.16807875125825925,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.002944607223245523,
+      "loss": 1.2973,
+      "step": 1915
+    },
+    {
+      "epoch": 0.16816652084116174,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0029444869171287983,
+      "loss": 1.3704,
+      "step": 1916
+    },
+    {
+      "epoch": 0.16825429042406423,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0029443664832513364,
+      "loss": 1.2957,
+      "step": 1917
+    },
+    {
+      "epoch": 0.16834206000696672,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002944245921625024,
+      "loss": 1.2621,
+      "step": 1918
+    },
+    {
+      "epoch": 0.16842982958986918,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0029441252322617607,
+      "loss": 1.382,
+      "step": 1919
+    },
+    {
+      "epoch": 0.16851759917277168,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0029440044151734573,
+      "loss": 1.2368,
+      "step": 1920
+    },
+    {
+      "epoch": 0.16860536875567417,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0029438834703720386,
+      "loss": 1.3038,
+      "step": 1921
+    },
+    {
+      "epoch": 0.16869313833857666,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002943762397869441,
+      "loss": 1.2904,
+      "step": 1922
+    },
+    {
+      "epoch": 0.16878090792147912,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0029436411976776137,
+      "loss": 1.3509,
+      "step": 1923
+    },
+    {
+      "epoch": 0.16886867750438161,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.0029435198698085185,
+      "loss": 1.3115,
+      "step": 1924
+    },
+    {
+      "epoch": 0.1689564470872841,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002943398414274131,
+      "loss": 1.3038,
+      "step": 1925
+    },
+    {
+      "epoch": 0.1690442166701866,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.002943276831086438,
+      "loss": 1.3011,
+      "step": 1926
+    },
+    {
+      "epoch": 0.1691319862530891,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.002943155120257438,
+      "loss": 1.2891,
+      "step": 1927
+    },
+    {
+      "epoch": 0.16921975583599155,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0029430332817991446,
+      "loss": 1.2734,
+      "step": 1928
+    },
+    {
+      "epoch": 0.16930752541889404,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.002942911315723583,
+      "loss": 1.3928,
+      "step": 1929
+    },
+    {
+      "epoch": 0.16939529500179654,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.00294278922204279,
+      "loss": 1.3,
+      "step": 1930
+    },
+    {
+      "epoch": 0.16948306458469903,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0029426670007688164,
+      "loss": 1.2545,
+      "step": 1931
+    },
+    {
+      "epoch": 0.1695708341676015,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.002942544651913724,
+      "loss": 1.3696,
+      "step": 1932
+    },
+    {
+      "epoch": 0.16965860375050398,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0029424221754895895,
+      "loss": 1.2651,
+      "step": 1933
+    },
+    {
+      "epoch": 0.16974637333340648,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0029422995715085,
+      "loss": 1.3701,
+      "step": 1934
+    },
+    {
+      "epoch": 0.16983414291630897,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.0029421768399825565,
+      "loss": 1.2822,
+      "step": 1935
+    },
+    {
+      "epoch": 0.16992191249921146,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0029420539809238712,
+      "loss": 1.2544,
+      "step": 1936
+    },
+    {
+      "epoch": 0.17000968208211392,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.002941930994344571,
+      "loss": 1.3308,
+      "step": 1937
+    },
+    {
+      "epoch": 0.1700974516650164,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.002941807880256792,
+      "loss": 1.3318,
+      "step": 1938
+    },
+    {
+      "epoch": 0.1701852212479189,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.0029416846386726885,
+      "loss": 1.3578,
+      "step": 1939
+    },
+    {
+      "epoch": 0.1702729908308214,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0029415612696044213,
+      "loss": 1.3668,
+      "step": 1940
+    },
+    {
+      "epoch": 0.17036076041372386,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0029414377730641672,
+      "loss": 1.4032,
+      "step": 1941
+    },
+    {
+      "epoch": 0.17044852999662635,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.0029413141490641148,
+      "loss": 1.3287,
+      "step": 1942
+    },
+    {
+      "epoch": 0.17053629957952884,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.0029411903976164656,
+      "loss": 1.3229,
+      "step": 1943
+    },
+    {
+      "epoch": 0.17062406916243134,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0029410665187334334,
+      "loss": 1.2777,
+      "step": 1944
+    },
+    {
+      "epoch": 0.1707118387453338,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0029409425124272436,
+      "loss": 1.3409,
+      "step": 1945
+    },
+    {
+      "epoch": 0.1707996083282363,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002940818378710136,
+      "loss": 1.3148,
+      "step": 1946
+    },
+    {
+      "epoch": 0.17088737791113878,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0029406941175943624,
+      "loss": 1.2958,
+      "step": 1947
+    },
+    {
+      "epoch": 0.17097514749404127,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.002940569729092186,
+      "loss": 1.37,
+      "step": 1948
+    },
+    {
+      "epoch": 0.17106291707694377,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.002940445213215884,
+      "loss": 1.2833,
+      "step": 1949
+    },
+    {
+      "epoch": 0.17115068665984623,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.0029403205699777453,
+      "loss": 1.3383,
+      "step": 1950
+    },
+    {
+      "epoch": 0.17123845624274872,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0029401957993900724,
+      "loss": 1.3357,
+      "step": 1951
+    },
+    {
+      "epoch": 0.1713262258256512,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002940070901465179,
+      "loss": 1.375,
+      "step": 1952
+    },
+    {
+      "epoch": 0.1714139954085537,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0029399458762153917,
+      "loss": 1.28,
+      "step": 1953
+    },
+    {
+      "epoch": 0.17150176499145617,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.0029398207236530512,
+      "loss": 1.3448,
+      "step": 1954
+    },
+    {
+      "epoch": 0.17158953457435866,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0029396954437905082,
+      "loss": 1.283,
+      "step": 1955
+    },
+    {
+      "epoch": 0.17167730415726115,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0029395700366401283,
+      "loss": 1.2932,
+      "step": 1956
+    },
+    {
+      "epoch": 0.17176507374016364,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002939444502214288,
+      "loss": 1.3213,
+      "step": 1957
+    },
+    {
+      "epoch": 0.17185284332306613,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0029393188405253774,
+      "loss": 1.2877,
+      "step": 1958
+    },
+    {
+      "epoch": 0.1719406129059686,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0029391930515857995,
+      "loss": 1.3419,
+      "step": 1959
+    },
+    {
+      "epoch": 0.1720283824888711,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0029390671354079683,
+      "loss": 1.4361,
+      "step": 1960
+    },
+    {
+      "epoch": 0.17211615207177358,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.002938941092004311,
+      "loss": 1.335,
+      "step": 1961
+    },
+    {
+      "epoch": 0.17220392165467607,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.002938814921387268,
+      "loss": 1.3516,
+      "step": 1962
+    },
+    {
+      "epoch": 0.17229169123757854,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.002938688623569292,
+      "loss": 1.3127,
+      "step": 1963
+    },
+    {
+      "epoch": 0.17237946082048103,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.002938562198562848,
+      "loss": 1.2745,
+      "step": 1964
+    },
+    {
+      "epoch": 0.17246723040338352,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0029384356463804133,
+      "loss": 1.3242,
+      "step": 1965
+    },
+    {
+      "epoch": 0.172554999986286,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0029383089670344787,
+      "loss": 1.3244,
+      "step": 1966
+    },
+    {
+      "epoch": 0.17264276956918848,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0029381821605375465,
+      "loss": 1.3211,
+      "step": 1967
+    },
+    {
+      "epoch": 0.17273053915209097,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0029380552269021322,
+      "loss": 1.3583,
+      "step": 1968
+    },
+    {
+      "epoch": 0.17281830873499346,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.002937928166140764,
+      "loss": 1.3176,
+      "step": 1969
+    },
+    {
+      "epoch": 0.17290607831789595,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0029378009782659816,
+      "loss": 1.4045,
+      "step": 1970
+    },
+    {
+      "epoch": 0.17299384790079844,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.0029376736632903376,
+      "loss": 1.2959,
+      "step": 1971
+    },
+    {
+      "epoch": 0.1730816174837009,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.002937546221226399,
+      "loss": 1.3519,
+      "step": 1972
+    },
+    {
+      "epoch": 0.1731693870666034,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0029374186520867425,
+      "loss": 1.3616,
+      "step": 1973
+    },
+    {
+      "epoch": 0.1732571566495059,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.002937290955883959,
+      "loss": 1.2925,
+      "step": 1974
+    },
+    {
+      "epoch": 0.17334492623240838,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0029371631326306517,
+      "loss": 1.3424,
+      "step": 1975
+    },
+    {
+      "epoch": 0.17343269581531084,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0029370351823394363,
+      "loss": 1.2905,
+      "step": 1976
+    },
+    {
+      "epoch": 0.17352046539821334,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.002936907105022941,
+      "loss": 1.3576,
+      "step": 1977
+    },
+    {
+      "epoch": 0.17360823498111583,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002936778900693807,
+      "loss": 1.3014,
+      "step": 1978
+    },
+    {
+      "epoch": 0.17369600456401832,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0029366505693646863,
+      "loss": 1.3876,
+      "step": 1979
+    },
+    {
+      "epoch": 0.17378377414692078,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0029365221110482457,
+      "loss": 1.3465,
+      "step": 1980
+    },
+    {
+      "epoch": 0.17387154372982327,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.002936393525757163,
+      "loss": 1.2636,
+      "step": 1981
+    },
+    {
+      "epoch": 0.17395931331272577,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.0029362648135041296,
+      "loss": 1.3364,
+      "step": 1982
+    },
+    {
+      "epoch": 0.17404708289562826,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.0029361359743018487,
+      "loss": 1.3762,
+      "step": 1983
+    },
+    {
+      "epoch": 0.17413485247853075,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.002936007008163036,
+      "loss": 1.3732,
+      "step": 1984
+    },
+    {
+      "epoch": 0.1742226220614332,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0029358779151004194,
+      "loss": 1.3531,
+      "step": 1985
+    },
+    {
+      "epoch": 0.1743103916443357,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0029357486951267415,
+      "loss": 1.2781,
+      "step": 1986
+    },
+    {
+      "epoch": 0.1743981612272382,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.002935619348254754,
+      "loss": 1.2857,
+      "step": 1987
+    },
+    {
+      "epoch": 0.1744859308101407,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0029354898744972245,
+      "loss": 1.2691,
+      "step": 1988
+    },
+    {
+      "epoch": 0.17457370039304315,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0029353602738669306,
+      "loss": 1.3292,
+      "step": 1989
+    },
+    {
+      "epoch": 0.17466146997594564,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0029352305463766636,
+      "loss": 1.2833,
+      "step": 1990
+    },
+    {
+      "epoch": 0.17474923955884814,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0029351006920392267,
+      "loss": 1.3149,
+      "step": 1991
+    },
+    {
+      "epoch": 0.17483700914175063,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.002934970710867437,
+      "loss": 1.2937,
+      "step": 1992
+    },
+    {
+      "epoch": 0.17492477872465312,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0029348406028741223,
+      "loss": 1.2783,
+      "step": 1993
+    },
+    {
+      "epoch": 0.17501254830755558,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0029347103680721237,
+      "loss": 1.304,
+      "step": 1994
+    },
+    {
+      "epoch": 0.17510031789045807,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002934580006474295,
+      "loss": 1.3559,
+      "step": 1995
+    },
+    {
+      "epoch": 0.17518808747336057,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0029344495180935034,
+      "loss": 1.347,
+      "step": 1996
+    },
+    {
+      "epoch": 0.17527585705626306,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0029343189029426256,
+      "loss": 1.3305,
+      "step": 1997
+    },
+    {
+      "epoch": 0.17536362663916552,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0029341881610345545,
+      "loss": 1.2606,
+      "step": 1998
+    },
+    {
+      "epoch": 0.175451396222068,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.002934057292382193,
+      "loss": 1.2748,
+      "step": 1999
+    },
+    {
+      "epoch": 0.1755391658049705,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0029339262969984574,
+      "loss": 1.2638,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1755391658049705,
+      "eval_loss": 1.303299069404602,
+      "eval_runtime": 437.5784,
+      "eval_samples_per_second": 33.685,
+      "eval_steps_per_second": 8.421,
+      "step": 2000
+    },
+    {
+      "epoch": 0.175626935387873,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0029337951748962766,
+      "loss": 1.3049,
+      "step": 2001
+    },
+    {
+      "epoch": 0.17571470497077546,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0029336639260885916,
+      "loss": 1.3366,
+      "step": 2002
+    },
+    {
+      "epoch": 0.17580247455367795,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0029335325505883566,
+      "loss": 1.3756,
+      "step": 2003
+    },
+    {
+      "epoch": 0.17589024413658044,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0029334010484085376,
+      "loss": 1.3383,
+      "step": 2004
+    },
+    {
+      "epoch": 0.17597801371948293,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.002933269419562113,
+      "loss": 1.3419,
+      "step": 2005
+    },
+    {
+      "epoch": 0.17606578330238543,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.002933137664062074,
+      "loss": 1.3246,
+      "step": 2006
+    },
+    {
+      "epoch": 0.1761535528852879,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.002933005781921426,
+      "loss": 1.2929,
+      "step": 2007
+    },
+    {
+      "epoch": 0.17624132246819038,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029328737731531824,
+      "loss": 1.3092,
+      "step": 2008
+    },
+    {
+      "epoch": 0.17632909205109287,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0029327416377703743,
+      "loss": 1.3364,
+      "step": 2009
+    },
+    {
+      "epoch": 0.17641686163399536,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.0029326093757860417,
+      "loss": 1.3512,
+      "step": 2010
+    },
+    {
+      "epoch": 0.17650463121689783,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0029324769872132394,
+      "loss": 1.2838,
+      "step": 2011
+    },
+    {
+      "epoch": 0.17659240079980032,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0029323444720650324,
+      "loss": 1.272,
+      "step": 2012
+    },
+    {
+      "epoch": 0.1766801703827028,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0029322118303545,
+      "loss": 1.3825,
+      "step": 2013
+    },
+    {
+      "epoch": 0.1767679399656053,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.0029320790620947336,
+      "loss": 1.2813,
+      "step": 2014
+    },
+    {
+      "epoch": 0.17685570954850777,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0029319461672988366,
+      "loss": 1.3582,
+      "step": 2015
+    },
+    {
+      "epoch": 0.17694347913141026,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0029318131459799253,
+      "loss": 1.2998,
+      "step": 2016
+    },
+    {
+      "epoch": 0.17703124871431275,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0029316799981511286,
+      "loss": 1.3067,
+      "step": 2017
+    },
+    {
+      "epoch": 0.17711901829721524,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.002931546723825587,
+      "loss": 1.2548,
+      "step": 2018
+    },
+    {
+      "epoch": 0.17720678788011773,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0029314133230164546,
+      "loss": 1.3689,
+      "step": 2019
+    },
+    {
+      "epoch": 0.1772945574630202,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0029312797957368976,
+      "loss": 1.3229,
+      "step": 2020
+    },
+    {
+      "epoch": 0.1773823270459227,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0029311461420000948,
+      "loss": 1.3033,
+      "step": 2021
+    },
+    {
+      "epoch": 0.17747009662882518,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002931012361819237,
+      "loss": 1.2997,
+      "step": 2022
+    },
+    {
+      "epoch": 0.17755786621172767,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002930878455207527,
+      "loss": 1.3353,
+      "step": 2023
+    },
+    {
+      "epoch": 0.17764563579463014,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.002930744422178183,
+      "loss": 1.2805,
+      "step": 2024
+    },
+    {
+      "epoch": 0.17773340537753263,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.002930610262744431,
+      "loss": 1.3085,
+      "step": 2025
+    },
+    {
+      "epoch": 0.17782117496043512,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0029304759769195137,
+      "loss": 1.304,
+      "step": 2026
+    },
+    {
+      "epoch": 0.1779089445433376,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0029303415647166844,
+      "loss": 1.2915,
+      "step": 2027
+    },
+    {
+      "epoch": 0.1779967141262401,
+      "grad_norm": 0.125,
+      "learning_rate": 0.002930207026149208,
+      "loss": 1.3411,
+      "step": 2028
+    },
+    {
+      "epoch": 0.17808448370914257,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029300723612303643,
+      "loss": 1.3686,
+      "step": 2029
+    },
+    {
+      "epoch": 0.17817225329204506,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.002929937569973443,
+      "loss": 1.283,
+      "step": 2030
+    },
+    {
+      "epoch": 0.17826002287494755,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0029298026523917483,
+      "loss": 1.3273,
+      "step": 2031
+    },
+    {
+      "epoch": 0.17834779245785004,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002929667608498596,
+      "loss": 1.3028,
+      "step": 2032
+    },
+    {
+      "epoch": 0.1784355620407525,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.002929532438307314,
+      "loss": 1.3427,
+      "step": 2033
+    },
+    {
+      "epoch": 0.178523331623655,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0029293971418312427,
+      "loss": 1.3177,
+      "step": 2034
+    },
+    {
+      "epoch": 0.1786111012065575,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.002929261719083736,
+      "loss": 1.3875,
+      "step": 2035
+    },
+    {
+      "epoch": 0.17869887078945998,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0029291261700781592,
+      "loss": 1.3412,
+      "step": 2036
+    },
+    {
+      "epoch": 0.17878664037236244,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0029289904948278916,
+      "loss": 1.3463,
+      "step": 2037
+    },
+    {
+      "epoch": 0.17887440995526493,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.002928854693346322,
+      "loss": 1.3193,
+      "step": 2038
+    },
+    {
+      "epoch": 0.17896217953816743,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.002928718765646855,
+      "loss": 1.3168,
+      "step": 2039
+    },
+    {
+      "epoch": 0.17904994912106992,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.002928582711742905,
+      "loss": 1.3616,
+      "step": 2040
+    },
+    {
+      "epoch": 0.1791377187039724,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.0029284465316479,
+      "loss": 1.301,
+      "step": 2041
+    },
+    {
+      "epoch": 0.17922548828687487,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0029283102253752817,
+      "loss": 1.299,
+      "step": 2042
+    },
+    {
+      "epoch": 0.17931325786977736,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.002928173792938502,
+      "loss": 1.3139,
+      "step": 2043
+    },
+    {
+      "epoch": 0.17940102745267986,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0029280372343510265,
+      "loss": 1.3261,
+      "step": 2044
+    },
+    {
+      "epoch": 0.17948879703558235,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0029279005496263325,
+      "loss": 1.3821,
+      "step": 2045
+    },
+    {
+      "epoch": 0.1795765666184848,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0029277637387779107,
+      "loss": 1.3539,
+      "step": 2046
+    },
+    {
+      "epoch": 0.1796643362013873,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.002927626801819264,
+      "loss": 1.2577,
+      "step": 2047
+    },
+    {
+      "epoch": 0.1797521057842898,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0029274897387639074,
+      "loss": 1.3158,
+      "step": 2048
+    },
+    {
+      "epoch": 0.1798398753671923,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.002927352549625368,
+      "loss": 1.2814,
+      "step": 2049
+    },
+    {
+      "epoch": 0.17992764495009475,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.002927215234417186,
+      "loss": 1.3311,
+      "step": 2050
+    },
+    {
+      "epoch": 0.18001541453299724,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0029270777931529146,
+      "loss": 1.3424,
+      "step": 2051
+    },
+    {
+      "epoch": 0.18010318411589973,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 0.002926940225846118,
+      "loss": 1.2519,
+      "step": 2052
+    },
+    {
+      "epoch": 0.18019095369880223,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.002926802532510373,
+      "loss": 1.3654,
+      "step": 2053
+    },
+    {
+      "epoch": 0.18027872328170472,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 0.002926664713159271,
+      "loss": 1.3042,
+      "step": 2054
+    },
+    {
+      "epoch": 0.18036649286460718,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.002926526767806413,
+      "loss": 1.2686,
+      "step": 2055
+    },
+    {
+      "epoch": 0.18045426244750967,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0029263886964654142,
+      "loss": 1.2894,
+      "step": 2056
+    },
+    {
+      "epoch": 0.18054203203041216,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.002926250499149901,
+      "loss": 1.3748,
+      "step": 2057
+    },
+    {
+      "epoch": 0.18062980161331466,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002926112175873514,
+      "loss": 1.3336,
+      "step": 2058
+    },
+    {
+      "epoch": 0.18071757119621712,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0029259737266499044,
+      "loss": 1.3166,
+      "step": 2059
+    },
+    {
+      "epoch": 0.1808053407791196,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0029258351514927367,
+      "loss": 1.3143,
+      "step": 2060
+    },
+    {
+      "epoch": 0.1808931103620221,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0029256964504156876,
+      "loss": 1.2732,
+      "step": 2061
+    },
+    {
+      "epoch": 0.1809808799449246,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002925557623432447,
+      "loss": 1.318,
+      "step": 2062
+    },
+    {
+      "epoch": 0.18106864952782709,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002925418670556715,
+      "loss": 1.3617,
+      "step": 2063
+    },
+    {
+      "epoch": 0.18115641911072955,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0029252795918022077,
+      "loss": 1.301,
+      "step": 2064
+    },
+    {
+      "epoch": 0.18124418869363204,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.0029251403871826506,
+      "loss": 1.3337,
+      "step": 2065
+    },
+    {
+      "epoch": 0.18133195827653453,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002925001056711783,
+      "loss": 1.2517,
+      "step": 2066
+    },
+    {
+      "epoch": 0.18141972785943702,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0029248616004033555,
+      "loss": 1.3646,
+      "step": 2067
+    },
+    {
+      "epoch": 0.1815074974423395,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0029247220182711323,
+      "loss": 1.2915,
+      "step": 2068
+    },
+    {
+      "epoch": 0.18159526702524198,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.00292458231032889,
+      "loss": 1.2812,
+      "step": 2069
+    },
+    {
+      "epoch": 0.18168303660814447,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0029244424765904175,
+      "loss": 1.3036,
+      "step": 2070
+    },
+    {
+      "epoch": 0.18177080619104696,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0029243025170695145,
+      "loss": 1.2834,
+      "step": 2071
+    },
+    {
+      "epoch": 0.18185857577394943,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0029241624317799956,
+      "loss": 1.3161,
+      "step": 2072
+    },
+    {
+      "epoch": 0.18194634535685192,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002924022220735686,
+      "loss": 1.2891,
+      "step": 2073
+    },
+    {
+      "epoch": 0.1820341149397544,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0029238818839504247,
+      "loss": 1.3126,
+      "step": 2074
+    },
+    {
+      "epoch": 0.1821218845226569,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.002923741421438061,
+      "loss": 1.3688,
+      "step": 2075
+    },
+    {
+      "epoch": 0.1822096541055594,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0029236008332124597,
+      "loss": 1.3122,
+      "step": 2076
+    },
+    {
+      "epoch": 0.18229742368846186,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.002923460119287496,
+      "loss": 1.3137,
+      "step": 2077
+    },
+    {
+      "epoch": 0.18238519327136435,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.002923319279677057,
+      "loss": 1.2864,
+      "step": 2078
+    },
+    {
+      "epoch": 0.18247296285426684,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.002923178314395043,
+      "loss": 1.2935,
+      "step": 2079
+    },
+    {
+      "epoch": 0.18256073243716933,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0029230372234553673,
+      "loss": 1.2647,
+      "step": 2080
+    },
+    {
+      "epoch": 0.1826485020200718,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0029228960068719547,
+      "loss": 1.3071,
+      "step": 2081
+    },
+    {
+      "epoch": 0.1827362716029743,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0029227546646587426,
+      "loss": 1.3071,
+      "step": 2082
+    },
+    {
+      "epoch": 0.18282404118587678,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.002922613196829681,
+      "loss": 1.3912,
+      "step": 2083
+    },
+    {
+      "epoch": 0.18291181076877927,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.002922471603398733,
+      "loss": 1.3432,
+      "step": 2084
+    },
+    {
+      "epoch": 0.18299958035168173,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0029223298843798727,
+      "loss": 1.3597,
+      "step": 2085
+    },
+    {
+      "epoch": 0.18308734993458423,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0029221880397870873,
+      "loss": 1.2975,
+      "step": 2086
+    },
+    {
+      "epoch": 0.18317511951748672,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0029220460696343753,
+      "loss": 1.3389,
+      "step": 2087
+    },
+    {
+      "epoch": 0.1832628891003892,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00292190397393575,
+      "loss": 1.32,
+      "step": 2088
+    },
+    {
+      "epoch": 0.1833506586832917,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.002921761752705235,
+      "loss": 1.336,
+      "step": 2089
+    },
+    {
+      "epoch": 0.18343842826619416,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.002921619405956867,
+      "loss": 1.3098,
+      "step": 2090
+    },
+    {
+      "epoch": 0.18352619784909666,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.002921476933704696,
+      "loss": 1.2683,
+      "step": 2091
+    },
+    {
+      "epoch": 0.18361396743199915,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0029213343359627817,
+      "loss": 1.2809,
+      "step": 2092
+    },
+    {
+      "epoch": 0.18370173701490164,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0029211916127451993,
+      "loss": 1.2894,
+      "step": 2093
+    },
+    {
+      "epoch": 0.1837895065978041,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0029210487640660343,
+      "loss": 1.2965,
+      "step": 2094
+    },
+    {
+      "epoch": 0.1838772761807066,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.002920905789939386,
+      "loss": 1.3088,
+      "step": 2095
+    },
+    {
+      "epoch": 0.1839650457636091,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0029207626903793643,
+      "loss": 1.2807,
+      "step": 2096
+    },
+    {
+      "epoch": 0.18405281534651158,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0029206194654000936,
+      "loss": 1.2842,
+      "step": 2097
+    },
+    {
+      "epoch": 0.18414058492941407,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.002920476115015709,
+      "loss": 1.3278,
+      "step": 2098
+    },
+    {
+      "epoch": 0.18422835451231653,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0029203326392403597,
+      "loss": 1.3635,
+      "step": 2099
+    },
+    {
+      "epoch": 0.18431612409521902,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0029201890380882046,
+      "loss": 1.321,
+      "step": 2100
+    },
+    {
+      "epoch": 0.18440389367812152,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0029200453115734177,
+      "loss": 1.3196,
+      "step": 2101
+    },
+    {
+      "epoch": 0.184491663261024,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0029199014597101835,
+      "loss": 1.308,
+      "step": 2102
+    },
+    {
+      "epoch": 0.18457943284392647,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.0029197574825126998,
+      "loss": 1.317,
+      "step": 2103
+    },
+    {
+      "epoch": 0.18466720242682896,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002919613379995177,
+      "loss": 1.2931,
+      "step": 2104
+    },
+    {
+      "epoch": 0.18475497200973146,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0029194691521718377,
+      "loss": 1.2302,
+      "step": 2105
+    },
+    {
+      "epoch": 0.18484274159263395,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.0029193247990569153,
+      "loss": 1.3406,
+      "step": 2106
+    },
+    {
+      "epoch": 0.1849305111755364,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0029191803206646584,
+      "loss": 1.276,
+      "step": 2107
+    },
+    {
+      "epoch": 0.1850182807584389,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.002919035717009325,
+      "loss": 1.3149,
+      "step": 2108
+    },
+    {
+      "epoch": 0.1851060503413414,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0029188909881051883,
+      "loss": 1.3338,
+      "step": 2109
+    },
+    {
+      "epoch": 0.18519381992424389,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0029187461339665317,
+      "loss": 1.3204,
+      "step": 2110
+    },
+    {
+      "epoch": 0.18528158950714638,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0029186011546076516,
+      "loss": 1.317,
+      "step": 2111
+    },
+    {
+      "epoch": 0.18536935909004884,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0029184560500428573,
+      "loss": 1.2885,
+      "step": 2112
+    },
+    {
+      "epoch": 0.18545712867295133,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.0029183108202864697,
+      "loss": 1.3115,
+      "step": 2113
+    },
+    {
+      "epoch": 0.18554489825585382,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.002918165465352823,
+      "loss": 1.3109,
+      "step": 2114
+    },
+    {
+      "epoch": 0.18563266783875632,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.002918019985256263,
+      "loss": 1.2992,
+      "step": 2115
+    },
+    {
+      "epoch": 0.18572043742165878,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002917874380011147,
+      "loss": 1.3017,
+      "step": 2116
+    },
+    {
+      "epoch": 0.18580820700456127,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0029177286496318462,
+      "loss": 1.3196,
+      "step": 2117
+    },
+    {
+      "epoch": 0.18589597658746376,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0029175827941327445,
+      "loss": 1.2701,
+      "step": 2118
+    },
+    {
+      "epoch": 0.18598374617036625,
+      "grad_norm": 0.068359375,
+      "learning_rate": 0.0029174368135282368,
+      "loss": 1.2962,
+      "step": 2119
+    },
+    {
+      "epoch": 0.18607151575326872,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.0029172907078327303,
+      "loss": 1.2666,
+      "step": 2120
+    },
+    {
+      "epoch": 0.1861592853361712,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.002917144477060645,
+      "loss": 1.2587,
+      "step": 2121
+    },
+    {
+      "epoch": 0.1862470549190737,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.002916998121226414,
+      "loss": 1.2867,
+      "step": 2122
+    },
+    {
+      "epoch": 0.1863348245019762,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0029168516403444817,
+      "loss": 1.3342,
+      "step": 2123
+    },
+    {
+      "epoch": 0.18642259408487868,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.0029167050344293052,
+      "loss": 1.3325,
+      "step": 2124
+    },
+    {
+      "epoch": 0.18651036366778115,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002916558303495354,
+      "loss": 1.3282,
+      "step": 2125
+    },
+    {
+      "epoch": 0.18659813325068364,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.0029164114475571097,
+      "loss": 1.2693,
+      "step": 2126
+    },
+    {
+      "epoch": 0.18668590283358613,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.002916264466629066,
+      "loss": 1.2907,
+      "step": 2127
+    },
+    {
+      "epoch": 0.18677367241648862,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.00291611736072573,
+      "loss": 1.3408,
+      "step": 2128
+    },
+    {
+      "epoch": 0.1868614419993911,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.00291597012986162,
+      "loss": 1.345,
+      "step": 2129
+    },
+    {
+      "epoch": 0.18694921158229358,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002915822774051268,
+      "loss": 1.3048,
+      "step": 2130
+    },
+    {
+      "epoch": 0.18703698116519607,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0029156752933092167,
+      "loss": 1.3583,
+      "step": 2131
+    },
+    {
+      "epoch": 0.18712475074809856,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.002915527687650022,
+      "loss": 1.3022,
+      "step": 2132
+    },
+    {
+      "epoch": 0.18721252033100105,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0029153799570882516,
+      "loss": 1.38,
+      "step": 2133
+    },
+    {
+      "epoch": 0.18730028991390352,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002915232101638487,
+      "loss": 1.31,
+      "step": 2134
+    },
+    {
+      "epoch": 0.187388059496806,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.00291508412131532,
+      "loss": 1.2774,
+      "step": 2135
+    },
+    {
+      "epoch": 0.1874758290797085,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.002914936016133355,
+      "loss": 1.2977,
+      "step": 2136
+    },
+    {
+      "epoch": 0.187563598662611,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0029147877861072113,
+      "loss": 1.3102,
+      "step": 2137
+    },
+    {
+      "epoch": 0.18765136824551346,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0029146394312515173,
+      "loss": 1.2452,
+      "step": 2138
+    },
+    {
+      "epoch": 0.18773913782841595,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0029144909515809152,
+      "loss": 1.3241,
+      "step": 2139
+    },
+    {
+      "epoch": 0.18782690741131844,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0029143423471100598,
+      "loss": 1.2632,
+      "step": 2140
+    },
+    {
+      "epoch": 0.18791467699422093,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0029141936178536175,
+      "loss": 1.3414,
+      "step": 2141
+    },
+    {
+      "epoch": 0.1880024465771234,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0029140447638262667,
+      "loss": 1.3056,
+      "step": 2142
+    },
+    {
+      "epoch": 0.18809021616002589,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0029138957850427002,
+      "loss": 1.3506,
+      "step": 2143
+    },
+    {
+      "epoch": 0.18817798574292838,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.00291374668151762,
+      "loss": 1.3226,
+      "step": 2144
+    },
+    {
+      "epoch": 0.18826575532583087,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.002913597453265743,
+      "loss": 1.3124,
+      "step": 2145
+    },
+    {
+      "epoch": 0.18835352490873336,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.002913448100301797,
+      "loss": 1.3316,
+      "step": 2146
+    },
+    {
+      "epoch": 0.18844129449163582,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0029132986226405233,
+      "loss": 1.2576,
+      "step": 2147
+    },
+    {
+      "epoch": 0.18852906407453832,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0029131490202966735,
+      "loss": 1.347,
+      "step": 2148
+    },
+    {
+      "epoch": 0.1886168336574408,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0029129992932850138,
+      "loss": 1.3147,
+      "step": 2149
+    },
+    {
+      "epoch": 0.1887046032403433,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0029128494416203214,
+      "loss": 1.3429,
+      "step": 2150
+    },
+    {
+      "epoch": 0.18879237282324576,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0029126994653173852,
+      "loss": 1.3109,
+      "step": 2151
+    },
+    {
+      "epoch": 0.18888014240614825,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0029125493643910086,
+      "loss": 1.2686,
+      "step": 2152
+    },
+    {
+      "epoch": 0.18896791198905075,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0029123991388560052,
+      "loss": 1.3486,
+      "step": 2153
+    },
+    {
+      "epoch": 0.18905568157195324,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.002912248788727202,
+      "loss": 1.2511,
+      "step": 2154
+    },
+    {
+      "epoch": 0.1891434511548557,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0029120983140194377,
+      "loss": 1.3016,
+      "step": 2155
+    },
+    {
+      "epoch": 0.1892312207377582,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.002911947714747564,
+      "loss": 1.259,
+      "step": 2156
+    },
+    {
+      "epoch": 0.18931899032066069,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0029117969909264437,
+      "loss": 1.2719,
+      "step": 2157
+    },
+    {
+      "epoch": 0.18940675990356318,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0029116461425709532,
+      "loss": 1.3267,
+      "step": 2158
+    },
+    {
+      "epoch": 0.18949452948646567,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002911495169695981,
+      "loss": 1.2936,
+      "step": 2159
+    },
+    {
+      "epoch": 0.18958229906936813,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0029113440723164265,
+      "loss": 1.3098,
+      "step": 2160
+    },
+    {
+      "epoch": 0.18967006865227062,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.002911192850447203,
+      "loss": 1.2725,
+      "step": 2161
+    },
+    {
+      "epoch": 0.18975783823517312,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.0029110415041032353,
+      "loss": 1.3344,
+      "step": 2162
+    },
+    {
+      "epoch": 0.1898456078180756,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.002910890033299461,
+      "loss": 1.3558,
+      "step": 2163
+    },
+    {
+      "epoch": 0.18993337740097807,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.00291073843805083,
+      "loss": 1.3024,
+      "step": 2164
+    },
+    {
+      "epoch": 0.19002114698388056,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0029105867183723028,
+      "loss": 1.2924,
+      "step": 2165
+    },
+    {
+      "epoch": 0.19010891656678305,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.002910434874278855,
+      "loss": 1.344,
+      "step": 2166
+    },
+    {
+      "epoch": 0.19019668614968555,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0029102829057854724,
+      "loss": 1.3312,
+      "step": 2167
+    },
+    {
+      "epoch": 0.19028445573258804,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.002910130812907153,
+      "loss": 1.2713,
+      "step": 2168
+    },
+    {
+      "epoch": 0.1903722253154905,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002909978595658909,
+      "loss": 1.293,
+      "step": 2169
+    },
+    {
+      "epoch": 0.190459994898393,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.002909826254055763,
+      "loss": 1.2687,
+      "step": 2170
+    },
+    {
+      "epoch": 0.19054776448129548,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0029096737881127506,
+      "loss": 1.268,
+      "step": 2171
+    },
+    {
+      "epoch": 0.19063553406419798,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0029095211978449197,
+      "loss": 1.3009,
+      "step": 2172
+    },
+    {
+      "epoch": 0.19072330364710044,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.0029093684832673305,
+      "loss": 1.2841,
+      "step": 2173
+    },
+    {
+      "epoch": 0.19081107323000293,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0029092156443950544,
+      "loss": 1.2893,
+      "step": 2174
+    },
+    {
+      "epoch": 0.19089884281290542,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.002909062681243177,
+      "loss": 1.2805,
+      "step": 2175
+    },
+    {
+      "epoch": 0.19098661239580791,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.002908909593826795,
+      "loss": 1.2638,
+      "step": 2176
+    },
+    {
+      "epoch": 0.19107438197871038,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.002908756382161017,
+      "loss": 1.2976,
+      "step": 2177
+    },
+    {
+      "epoch": 0.19116215156161287,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0029086030462609647,
+      "loss": 1.3008,
+      "step": 2178
+    },
+    {
+      "epoch": 0.19124992114451536,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0029084495861417723,
+      "loss": 1.2945,
+      "step": 2179
+    },
+    {
+      "epoch": 0.19133769072741785,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.0029082960018185847,
+      "loss": 1.3352,
+      "step": 2180
+    },
+    {
+      "epoch": 0.19142546031032034,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.002908142293306561,
+      "loss": 1.3138,
+      "step": 2181
+    },
+    {
+      "epoch": 0.1915132298932228,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0029079884606208707,
+      "loss": 1.3149,
+      "step": 2182
+    },
+    {
+      "epoch": 0.1916009994761253,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.002907834503776697,
+      "loss": 1.3493,
+      "step": 2183
+    },
+    {
+      "epoch": 0.1916887690590278,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002907680422789235,
+      "loss": 1.374,
+      "step": 2184
+    },
+    {
+      "epoch": 0.19177653864193028,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0029075262176736917,
+      "loss": 1.3089,
+      "step": 2185
+    },
+    {
+      "epoch": 0.19186430822483275,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0029073718884452863,
+      "loss": 1.3155,
+      "step": 2186
+    },
+    {
+      "epoch": 0.19195207780773524,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0029072174351192507,
+      "loss": 1.2978,
+      "step": 2187
+    },
+    {
+      "epoch": 0.19203984739063773,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.002907062857710829,
+      "loss": 1.2919,
+      "step": 2188
+    },
+    {
+      "epoch": 0.19212761697354022,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002906908156235277,
+      "loss": 1.3122,
+      "step": 2189
+    },
+    {
+      "epoch": 0.19221538655644269,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002906753330707863,
+      "loss": 1.3184,
+      "step": 2190
+    },
+    {
+      "epoch": 0.19230315613934518,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.002906598381143869,
+      "loss": 1.3007,
+      "step": 2191
+    },
+    {
+      "epoch": 0.19239092572224767,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.002906443307558586,
+      "loss": 1.3458,
+      "step": 2192
+    },
+    {
+      "epoch": 0.19247869530515016,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.0029062881099673203,
+      "loss": 1.314,
+      "step": 2193
+    },
+    {
+      "epoch": 0.19256646488805265,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0029061327883853887,
+      "loss": 1.3517,
+      "step": 2194
+    },
+    {
+      "epoch": 0.19265423447095512,
+      "grad_norm": 0.125,
+      "learning_rate": 0.002905977342828122,
+      "loss": 1.3017,
+      "step": 2195
+    },
+    {
+      "epoch": 0.1927420040538576,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.00290582177331086,
+      "loss": 1.2707,
+      "step": 2196
+    },
+    {
+      "epoch": 0.1928297736367601,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002905666079848959,
+      "loss": 1.26,
+      "step": 2197
+    },
+    {
+      "epoch": 0.1929175432196626,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.002905510262457784,
+      "loss": 1.2971,
+      "step": 2198
+    },
+    {
+      "epoch": 0.19300531280256505,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.002905354321152714,
+      "loss": 1.3218,
+      "step": 2199
+    },
+    {
+      "epoch": 0.19309308238546755,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0029051982559491394,
+      "loss": 1.2688,
+      "step": 2200
+    },
+    {
+      "epoch": 0.19318085196837004,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002905042066862464,
+      "loss": 1.3785,
+      "step": 2201
+    },
+    {
+      "epoch": 0.19326862155127253,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002904885753908102,
+      "loss": 1.321,
+      "step": 2202
+    },
+    {
+      "epoch": 0.19335639113417502,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.002904729317101482,
+      "loss": 1.323,
+      "step": 2203
+    },
+    {
+      "epoch": 0.19344416071707748,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0029045727564580427,
+      "loss": 1.2085,
+      "step": 2204
+    },
+    {
+      "epoch": 0.19353193029997998,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0029044160719932364,
+      "loss": 1.3641,
+      "step": 2205
+    },
+    {
+      "epoch": 0.19361969988288247,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.002904259263722527,
+      "loss": 1.3244,
+      "step": 2206
+    },
+    {
+      "epoch": 0.19370746946578496,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.002904102331661392,
+      "loss": 1.3476,
+      "step": 2207
+    },
+    {
+      "epoch": 0.19379523904868742,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0029039452758253185,
+      "loss": 1.343,
+      "step": 2208
+    },
+    {
+      "epoch": 0.19388300863158991,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.002903788096229808,
+      "loss": 1.2731,
+      "step": 2209
+    },
+    {
+      "epoch": 0.1939707782144924,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0029036307928903735,
+      "loss": 1.285,
+      "step": 2210
+    },
+    {
+      "epoch": 0.1940585477973949,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0029034733658225404,
+      "loss": 1.2825,
+      "step": 2211
+    },
+    {
+      "epoch": 0.19414631738029736,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0029033158150418453,
+      "loss": 1.2815,
+      "step": 2212
+    },
+    {
+      "epoch": 0.19423408696319985,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.002903158140563838,
+      "loss": 1.288,
+      "step": 2213
+    },
+    {
+      "epoch": 0.19432185654610235,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.0029030003424040814,
+      "loss": 1.2565,
+      "step": 2214
+    },
+    {
+      "epoch": 0.19440962612900484,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.002902842420578149,
+      "loss": 1.2907,
+      "step": 2215
+    },
+    {
+      "epoch": 0.19449739571190733,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0029026843751016264,
+      "loss": 1.3151,
+      "step": 2216
+    },
+    {
+      "epoch": 0.1945851652948098,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.002902526205990113,
+      "loss": 1.2367,
+      "step": 2217
+    },
+    {
+      "epoch": 0.19467293487771228,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.002902367913259219,
+      "loss": 1.4069,
+      "step": 2218
+    },
+    {
+      "epoch": 0.19476070446061478,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.0029022094969245672,
+      "loss": 1.2718,
+      "step": 2219
+    },
+    {
+      "epoch": 0.19484847404351727,
+      "grad_norm": 0.06787109375,
+      "learning_rate": 0.002902050957001793,
+      "loss": 1.3363,
+      "step": 2220
+    },
+    {
+      "epoch": 0.19493624362641973,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.002901892293506543,
+      "loss": 1.3827,
+      "step": 2221
+    },
+    {
+      "epoch": 0.19502401320932222,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.0029017335064544774,
+      "loss": 1.2886,
+      "step": 2222
+    },
+    {
+      "epoch": 0.19511178279222471,
+      "grad_norm": 0.06591796875,
+      "learning_rate": 0.002901574595861268,
+      "loss": 1.271,
+      "step": 2223
+    },
+    {
+      "epoch": 0.1951995523751272,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.0029014155617425974,
+      "loss": 1.2646,
+      "step": 2224
+    },
+    {
+      "epoch": 0.19528732195802967,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0029012564041141628,
+      "loss": 1.34,
+      "step": 2225
+    },
+    {
+      "epoch": 0.19537509154093216,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.0029010971229916725,
+      "loss": 1.2941,
+      "step": 2226
+    },
+    {
+      "epoch": 0.19546286112383465,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0029009377183908465,
+      "loss": 1.3545,
+      "step": 2227
+    },
+    {
+      "epoch": 0.19555063070673714,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.0029007781903274167,
+      "loss": 1.2947,
+      "step": 2228
+    },
+    {
+      "epoch": 0.19563840028963964,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002900618538817129,
+      "loss": 1.3155,
+      "step": 2229
+    },
+    {
+      "epoch": 0.1957261698725421,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00290045876387574,
+      "loss": 1.3905,
+      "step": 2230
+    },
+    {
+      "epoch": 0.1958139394554446,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0029002988655190184,
+      "loss": 1.339,
+      "step": 2231
+    },
+    {
+      "epoch": 0.19590170903834708,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0029001388437627468,
+      "loss": 1.2732,
+      "step": 2232
+    },
+    {
+      "epoch": 0.19598947862124957,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0028999786986227175,
+      "loss": 1.3036,
+      "step": 2233
+    },
+    {
+      "epoch": 0.19607724820415204,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0028998184301147365,
+      "loss": 1.334,
+      "step": 2234
+    },
+    {
+      "epoch": 0.19616501778705453,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0028996580382546215,
+      "loss": 1.3219,
+      "step": 2235
+    },
+    {
+      "epoch": 0.19625278736995702,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.0028994975230582027,
+      "loss": 1.2653,
+      "step": 2236
+    },
+    {
+      "epoch": 0.1963405569528595,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.002899336884541323,
+      "loss": 1.3116,
+      "step": 2237
+    },
+    {
+      "epoch": 0.196428326535762,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.002899176122719836,
+      "loss": 1.2569,
+      "step": 2238
+    },
+    {
+      "epoch": 0.19651609611866447,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.002899015237609608,
+      "loss": 1.3395,
+      "step": 2239
+    },
+    {
+      "epoch": 0.19660386570156696,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002898854229226518,
+      "loss": 1.264,
+      "step": 2240
+    },
+    {
+      "epoch": 0.19669163528446945,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002898693097586458,
+      "loss": 1.2882,
+      "step": 2241
+    },
+    {
+      "epoch": 0.19677940486737194,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0028985318427053293,
+      "loss": 1.3069,
+      "step": 2242
+    },
+    {
+      "epoch": 0.1968671744502744,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.0028983704645990484,
+      "loss": 1.2961,
+      "step": 2243
+    },
+    {
+      "epoch": 0.1969549440331769,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0028982089632835417,
+      "loss": 1.3366,
+      "step": 2244
+    },
+    {
+      "epoch": 0.1970427136160794,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.00289804733877475,
+      "loss": 1.328,
+      "step": 2245
+    },
+    {
+      "epoch": 0.19713048319898188,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002897885591088623,
+      "loss": 1.2685,
+      "step": 2246
+    },
+    {
+      "epoch": 0.19721825278188435,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0028977237202411268,
+      "loss": 1.3347,
+      "step": 2247
+    },
+    {
+      "epoch": 0.19730602236478684,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.002897561726248236,
+      "loss": 1.2983,
+      "step": 2248
+    },
+    {
+      "epoch": 0.19739379194768933,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0028973996091259395,
+      "loss": 1.2672,
+      "step": 2249
+    },
+    {
+      "epoch": 0.19748156153059182,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0028972373688902374,
+      "loss": 1.2969,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1975693311134943,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.002897075005557142,
+      "loss": 1.2327,
+      "step": 2251
+    },
+    {
+      "epoch": 0.19765710069639678,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0028969125191426774,
+      "loss": 1.3086,
+      "step": 2252
+    },
+    {
+      "epoch": 0.19774487027929927,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002896749909662881,
+      "loss": 1.295,
+      "step": 2253
+    },
+    {
+      "epoch": 0.19783263986220176,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0028965871771338024,
+      "loss": 1.3236,
+      "step": 2254
+    },
+    {
+      "epoch": 0.19792040944510425,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.002896424321571501,
+      "loss": 1.2835,
+      "step": 2255
+    },
+    {
+      "epoch": 0.19800817902800671,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0028962613429920517,
+      "loss": 1.3419,
+      "step": 2256
+    },
+    {
+      "epoch": 0.1980959486109092,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0028960982414115385,
+      "loss": 1.329,
+      "step": 2257
+    },
+    {
+      "epoch": 0.1981837181938117,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.00289593501684606,
+      "loss": 1.3221,
+      "step": 2258
+    },
+    {
+      "epoch": 0.1982714877767142,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.0028957716693117246,
+      "loss": 1.2858,
+      "step": 2259
+    },
+    {
+      "epoch": 0.19835925735961665,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.002895608198824655,
+      "loss": 1.3049,
+      "step": 2260
+    },
+    {
+      "epoch": 0.19844702694251914,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.002895444605400985,
+      "loss": 1.272,
+      "step": 2261
+    },
+    {
+      "epoch": 0.19853479652542164,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.00289528088905686,
+      "loss": 1.3099,
+      "step": 2262
+    },
+    {
+      "epoch": 0.19862256610832413,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0028951170498084386,
+      "loss": 1.2792,
+      "step": 2263
+    },
+    {
+      "epoch": 0.19871033569122662,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.002894953087671892,
+      "loss": 1.3152,
+      "step": 2264
+    },
+    {
+      "epoch": 0.19879810527412908,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.002894789002663401,
+      "loss": 1.3203,
+      "step": 2265
+    },
+    {
+      "epoch": 0.19888587485703157,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.0028946247947991607,
+      "loss": 1.301,
+      "step": 2266
+    },
+    {
+      "epoch": 0.19897364443993407,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0028944604640953778,
+      "loss": 1.3096,
+      "step": 2267
+    },
+    {
+      "epoch": 0.19906141402283656,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0028942960105682713,
+      "loss": 1.33,
+      "step": 2268
+    },
+    {
+      "epoch": 0.19914918360573902,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.002894131434234072,
+      "loss": 1.3559,
+      "step": 2269
+    },
+    {
+      "epoch": 0.1992369531886415,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.002893966735109023,
+      "loss": 1.276,
+      "step": 2270
+    },
+    {
+      "epoch": 0.199324722771544,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0028938019132093793,
+      "loss": 1.2809,
+      "step": 2271
+    },
+    {
+      "epoch": 0.1994124923544465,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.002893636968551408,
+      "loss": 1.2869,
+      "step": 2272
+    },
+    {
+      "epoch": 0.199500261937349,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.002893471901151389,
+      "loss": 1.217,
+      "step": 2273
+    },
+    {
+      "epoch": 0.19958803152025145,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002893306711025614,
+      "loss": 1.3172,
+      "step": 2274
+    },
+    {
+      "epoch": 0.19967580110315394,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.0028931413981903857,
+      "loss": 1.3734,
+      "step": 2275
+    },
+    {
+      "epoch": 0.19976357068605644,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0028929759626620203,
+      "loss": 1.2834,
+      "step": 2276
+    },
+    {
+      "epoch": 0.19985134026895893,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0028928104044568448,
+      "loss": 1.3359,
+      "step": 2277
+    },
+    {
+      "epoch": 0.1999391098518614,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0028926447235912015,
+      "loss": 1.2903,
+      "step": 2278
+    },
+    {
+      "epoch": 0.20002687943476388,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.0028924789200814394,
+      "loss": 1.4082,
+      "step": 2279
+    },
+    {
+      "epoch": 0.20011464901766637,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.002892312993943925,
+      "loss": 1.2767,
+      "step": 2280
+    },
+    {
+      "epoch": 0.20020241860056887,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0028921469451950337,
+      "loss": 1.2858,
+      "step": 2281
+    },
+    {
+      "epoch": 0.20029018818347133,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0028919807738511535,
+      "loss": 1.3065,
+      "step": 2282
+    },
+    {
+      "epoch": 0.20037795776637382,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0028918144799286853,
+      "loss": 1.3137,
+      "step": 2283
+    },
+    {
+      "epoch": 0.2004657273492763,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0028916480634440414,
+      "loss": 1.2735,
+      "step": 2284
+    },
+    {
+      "epoch": 0.2005534969321788,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0028914815244136466,
+      "loss": 1.2725,
+      "step": 2285
+    },
+    {
+      "epoch": 0.2006412665150813,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.002891314862853938,
+      "loss": 1.2455,
+      "step": 2286
+    },
+    {
+      "epoch": 0.20072903609798376,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0028911480787813634,
+      "loss": 1.2959,
+      "step": 2287
+    },
+    {
+      "epoch": 0.20081680568088625,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.002890981172212385,
+      "loss": 1.2491,
+      "step": 2288
+    },
+    {
+      "epoch": 0.20090457526378874,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0028908141431634752,
+      "loss": 1.3692,
+      "step": 2289
+    },
+    {
+      "epoch": 0.20099234484669123,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0028906469916511186,
+      "loss": 1.3203,
+      "step": 2290
+    },
+    {
+      "epoch": 0.2010801144295937,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.002890479717691814,
+      "loss": 1.2715,
+      "step": 2291
+    },
+    {
+      "epoch": 0.2011678840124962,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002890312321302068,
+      "loss": 1.2872,
+      "step": 2292
+    },
+    {
+      "epoch": 0.20125565359539868,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002890144802498405,
+      "loss": 1.3064,
+      "step": 2293
+    },
+    {
+      "epoch": 0.20134342317830117,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0028899771612973563,
+      "loss": 1.2813,
+      "step": 2294
+    },
+    {
+      "epoch": 0.20143119276120364,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.002889809397715468,
+      "loss": 1.309,
+      "step": 2295
+    },
+    {
+      "epoch": 0.20151896234410613,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.0028896415117692985,
+      "loss": 1.3348,
+      "step": 2296
+    },
+    {
+      "epoch": 0.20160673192700862,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0028894735034754163,
+      "loss": 1.2646,
+      "step": 2297
+    },
+    {
+      "epoch": 0.2016945015099111,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0028893053728504038,
+      "loss": 1.2591,
+      "step": 2298
+    },
+    {
+      "epoch": 0.2017822710928136,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.002889137119910854,
+      "loss": 1.3292,
+      "step": 2299
+    },
+    {
+      "epoch": 0.20187004067571607,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0028889687446733746,
+      "loss": 1.2715,
+      "step": 2300
+    },
+    {
+      "epoch": 0.20195781025861856,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.002888800247154582,
+      "loss": 1.3209,
+      "step": 2301
+    },
+    {
+      "epoch": 0.20204557984152105,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0028886316273711065,
+      "loss": 1.3558,
+      "step": 2302
+    },
+    {
+      "epoch": 0.20213334942442354,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0028884628853395908,
+      "loss": 1.2889,
+      "step": 2303
+    },
+    {
+      "epoch": 0.202221119007326,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.0028882940210766886,
+      "loss": 1.319,
+      "step": 2304
+    },
+    {
+      "epoch": 0.2023088885902285,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0028881250345990654,
+      "loss": 1.3286,
+      "step": 2305
+    },
+    {
+      "epoch": 0.202396658173131,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0028879559259234015,
+      "loss": 1.2802,
+      "step": 2306
+    },
+    {
+      "epoch": 0.20248442775603348,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0028877866950663853,
+      "loss": 1.3011,
+      "step": 2307
+    },
+    {
+      "epoch": 0.20257219733893597,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.00288761734204472,
+      "loss": 1.3207,
+      "step": 2308
+    },
+    {
+      "epoch": 0.20265996692183844,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0028874478668751203,
+      "loss": 1.2711,
+      "step": 2309
+    },
+    {
+      "epoch": 0.20274773650474093,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.002887278269574312,
+      "loss": 1.2906,
+      "step": 2310
+    },
+    {
+      "epoch": 0.20283550608764342,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.002887108550159035,
+      "loss": 1.2525,
+      "step": 2311
+    },
+    {
+      "epoch": 0.2029232756705459,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002886938708646039,
+      "loss": 1.3109,
+      "step": 2312
+    },
+    {
+      "epoch": 0.20301104525344837,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.002886768745052086,
+      "loss": 1.3077,
+      "step": 2313
+    },
+    {
+      "epoch": 0.20309881483635087,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.002886598659393952,
+      "loss": 1.3287,
+      "step": 2314
+    },
+    {
+      "epoch": 0.20318658441925336,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0028864284516884227,
+      "loss": 1.3065,
+      "step": 2315
+    },
+    {
+      "epoch": 0.20327435400215585,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.002886258121952298,
+      "loss": 1.2697,
+      "step": 2316
+    },
+    {
+      "epoch": 0.2033621235850583,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0028860876702023882,
+      "loss": 1.2881,
+      "step": 2317
+    },
+    {
+      "epoch": 0.2034498931679608,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0028859170964555167,
+      "loss": 1.3198,
+      "step": 2318
+    },
+    {
+      "epoch": 0.2035376627508633,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.002885746400728518,
+      "loss": 1.3082,
+      "step": 2319
+    },
+    {
+      "epoch": 0.2036254323337658,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.002885575583038239,
+      "loss": 1.3452,
+      "step": 2320
+    },
+    {
+      "epoch": 0.20371320191666828,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0028854046434015385,
+      "loss": 1.3227,
+      "step": 2321
+    },
+    {
+      "epoch": 0.20380097149957074,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0028852335818352884,
+      "loss": 1.3368,
+      "step": 2322
+    },
+    {
+      "epoch": 0.20388874108247323,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.002885062398356371,
+      "loss": 1.3182,
+      "step": 2323
+    },
+    {
+      "epoch": 0.20397651066537573,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0028848910929816824,
+      "loss": 1.2907,
+      "step": 2324
+    },
+    {
+      "epoch": 0.20406428024827822,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0028847196657281284,
+      "loss": 1.3292,
+      "step": 2325
+    },
+    {
+      "epoch": 0.20415204983118068,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0028845481166126295,
+      "loss": 1.3166,
+      "step": 2326
+    },
+    {
+      "epoch": 0.20423981941408317,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0028843764456521156,
+      "loss": 1.2836,
+      "step": 2327
+    },
+    {
+      "epoch": 0.20432758899698567,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0028842046528635315,
+      "loss": 1.3468,
+      "step": 2328
+    },
+    {
+      "epoch": 0.20441535857988816,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0028840327382638315,
+      "loss": 1.3004,
+      "step": 2329
+    },
+    {
+      "epoch": 0.20450312816279062,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.0028838607018699825,
+      "loss": 1.2457,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2045908977456931,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.002883688543698965,
+      "loss": 1.2878,
+      "step": 2331
+    },
+    {
+      "epoch": 0.2046786673285956,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0028835162637677694,
+      "loss": 1.2904,
+      "step": 2332
+    },
+    {
+      "epoch": 0.2047664369114981,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.0028833438620933997,
+      "loss": 1.2577,
+      "step": 2333
+    },
+    {
+      "epoch": 0.2048542064944006,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.00288317133869287,
+      "loss": 1.3141,
+      "step": 2334
+    },
+    {
+      "epoch": 0.20494197607730305,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.0028829986935832094,
+      "loss": 1.2969,
+      "step": 2335
+    },
+    {
+      "epoch": 0.20502974566020554,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.0028828259267814566,
+      "loss": 1.329,
+      "step": 2336
+    },
+    {
+      "epoch": 0.20511751524310803,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.0028826530383046626,
+      "loss": 1.3423,
+      "step": 2337
+    },
+    {
+      "epoch": 0.20520528482601053,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0028824800281698907,
+      "loss": 1.2635,
+      "step": 2338
+    },
+    {
+      "epoch": 0.205293054408913,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0028823068963942174,
+      "loss": 1.2605,
+      "step": 2339
+    },
+    {
+      "epoch": 0.20538082399181548,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002882133642994729,
+      "loss": 1.2924,
+      "step": 2340
+    },
+    {
+      "epoch": 0.20546859357471797,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0028819602679885253,
+      "loss": 1.3222,
+      "step": 2341
+    },
+    {
+      "epoch": 0.20555636315762046,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.002881786771392718,
+      "loss": 1.2976,
+      "step": 2342
+    },
+    {
+      "epoch": 0.20564413274052296,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.00288161315322443,
+      "loss": 1.2954,
+      "step": 2343
+    },
+    {
+      "epoch": 0.20573190232342542,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0028814394135007974,
+      "loss": 1.3379,
+      "step": 2344
+    },
+    {
+      "epoch": 0.2058196719063279,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.002881265552238967,
+      "loss": 1.2876,
+      "step": 2345
+    },
+    {
+      "epoch": 0.2059074414892304,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0028810915694560987,
+      "loss": 1.2907,
+      "step": 2346
+    },
+    {
+      "epoch": 0.2059952110721329,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0028809174651693633,
+      "loss": 1.3451,
+      "step": 2347
+    },
+    {
+      "epoch": 0.20608298065503536,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.0028807432393959448,
+      "loss": 1.3329,
+      "step": 2348
+    },
+    {
+      "epoch": 0.20617075023793785,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.002880568892153038,
+      "loss": 1.3236,
+      "step": 2349
+    },
+    {
+      "epoch": 0.20625851982084034,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0028803944234578513,
+      "loss": 1.3613,
+      "step": 2350
+    },
+    {
+      "epoch": 0.20634628940374283,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.002880219833327603,
+      "loss": 1.2599,
+      "step": 2351
+    },
+    {
+      "epoch": 0.2064340589866453,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.002880045121779525,
+      "loss": 1.3439,
+      "step": 2352
+    },
+    {
+      "epoch": 0.2065218285695478,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.0028798702888308604,
+      "loss": 1.3908,
+      "step": 2353
+    },
+    {
+      "epoch": 0.20660959815245028,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002879695334498865,
+      "loss": 1.3702,
+      "step": 2354
+    },
+    {
+      "epoch": 0.20669736773535277,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0028795202588008054,
+      "loss": 1.2913,
+      "step": 2355
+    },
+    {
+      "epoch": 0.20678513731825526,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002879345061753961,
+      "loss": 1.2786,
+      "step": 2356
+    },
+    {
+      "epoch": 0.20687290690115773,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0028791697433756245,
+      "loss": 1.3243,
+      "step": 2357
+    },
+    {
+      "epoch": 0.20696067648406022,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.0028789943036830966,
+      "loss": 1.3272,
+      "step": 2358
+    },
+    {
+      "epoch": 0.2070484460669627,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0028788187426936943,
+      "loss": 1.3094,
+      "step": 2359
+    },
+    {
+      "epoch": 0.2071362156498652,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0028786430604247444,
+      "loss": 1.2953,
+      "step": 2360
+    },
+    {
+      "epoch": 0.20722398523276767,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.0028784672568935854,
+      "loss": 1.3535,
+      "step": 2361
+    },
+    {
+      "epoch": 0.20731175481567016,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.00287829133211757,
+      "loss": 1.3137,
+      "step": 2362
+    },
+    {
+      "epoch": 0.20739952439857265,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.002878115286114059,
+      "loss": 1.2906,
+      "step": 2363
+    },
+    {
+      "epoch": 0.20748729398147514,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0028779391189004294,
+      "loss": 1.3038,
+      "step": 2364
+    },
+    {
+      "epoch": 0.2075750635643776,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002877762830494067,
+      "loss": 1.296,
+      "step": 2365
+    },
+    {
+      "epoch": 0.2076628331472801,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.0028775864209123715,
+      "loss": 1.2616,
+      "step": 2366
+    },
+    {
+      "epoch": 0.2077506027301826,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0028774098901727537,
+      "loss": 1.2957,
+      "step": 2367
+    },
+    {
+      "epoch": 0.20783837231308508,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.0028772332382926364,
+      "loss": 1.3068,
+      "step": 2368
+    },
+    {
+      "epoch": 0.20792614189598757,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0028770564652894543,
+      "loss": 1.2917,
+      "step": 2369
+    },
+    {
+      "epoch": 0.20801391147889003,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0028768795711806546,
+      "loss": 1.2785,
+      "step": 2370
+    },
+    {
+      "epoch": 0.20810168106179253,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.002876702555983695,
+      "loss": 1.2979,
+      "step": 2371
+    },
+    {
+      "epoch": 0.20818945064469502,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.002876525419716047,
+      "loss": 1.2484,
+      "step": 2372
+    },
+    {
+      "epoch": 0.2082772202275975,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0028763481623951936,
+      "loss": 1.3524,
+      "step": 2373
+    },
+    {
+      "epoch": 0.20836498981049997,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0028761707840386287,
+      "loss": 1.2403,
+      "step": 2374
+    },
+    {
+      "epoch": 0.20845275939340246,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.00287599328466386,
+      "loss": 1.306,
+      "step": 2375
+    },
+    {
+      "epoch": 0.20854052897630496,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002875815664288404,
+      "loss": 1.2855,
+      "step": 2376
+    },
+    {
+      "epoch": 0.20862829855920745,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.002875637922929793,
+      "loss": 1.2874,
+      "step": 2377
+    },
+    {
+      "epoch": 0.20871606814210994,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.002875460060605568,
+      "loss": 1.3003,
+      "step": 2378
+    },
+    {
+      "epoch": 0.2088038377250124,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0028752820773332846,
+      "loss": 1.2508,
+      "step": 2379
+    },
+    {
+      "epoch": 0.2088916073079149,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0028751039731305075,
+      "loss": 1.2825,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2089793768908174,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0028749257480148163,
+      "loss": 1.3611,
+      "step": 2381
+    },
+    {
+      "epoch": 0.20906714647371988,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.002874747402003801,
+      "loss": 1.2617,
+      "step": 2382
+    },
+    {
+      "epoch": 0.20915491605662234,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0028745689351150627,
+      "loss": 1.3146,
+      "step": 2383
+    },
+    {
+      "epoch": 0.20924268563952483,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.0028743903473662158,
+      "loss": 1.3093,
+      "step": 2384
+    },
+    {
+      "epoch": 0.20933045522242733,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0028742116387748865,
+      "loss": 1.348,
+      "step": 2385
+    },
+    {
+      "epoch": 0.20941822480532982,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0028740328093587123,
+      "loss": 1.2962,
+      "step": 2386
+    },
+    {
+      "epoch": 0.20950599438823228,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0028738538591353427,
+      "loss": 1.3539,
+      "step": 2387
+    },
+    {
+      "epoch": 0.20959376397113477,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0028736747881224402,
+      "loss": 1.3226,
+      "step": 2388
+    },
+    {
+      "epoch": 0.20968153355403726,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.0028734955963376782,
+      "loss": 1.2698,
+      "step": 2389
+    },
+    {
+      "epoch": 0.20976930313693976,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.002873316283798741,
+      "loss": 1.2835,
+      "step": 2390
+    },
+    {
+      "epoch": 0.20985707271984225,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.002873136850523328,
+      "loss": 1.2798,
+      "step": 2391
+    },
+    {
+      "epoch": 0.2099448423027447,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0028729572965291474,
+      "loss": 1.3507,
+      "step": 2392
+    },
+    {
+      "epoch": 0.2100326118856472,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0028727776218339203,
+      "loss": 1.2885,
+      "step": 2393
+    },
+    {
+      "epoch": 0.2101203814685497,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.00287259782645538,
+      "loss": 1.314,
+      "step": 2394
+    },
+    {
+      "epoch": 0.21020815105145219,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.0028724179104112724,
+      "loss": 1.3207,
+      "step": 2395
+    },
+    {
+      "epoch": 0.21029592063435465,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0028722378737193535,
+      "loss": 1.2802,
+      "step": 2396
+    },
+    {
+      "epoch": 0.21038369021725714,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0028720577163973927,
+      "loss": 1.2236,
+      "step": 2397
+    },
+    {
+      "epoch": 0.21047145980015963,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.002871877438463171,
+      "loss": 1.2892,
+      "step": 2398
+    },
+    {
+      "epoch": 0.21055922938306212,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0028716970399344805,
+      "loss": 1.3224,
+      "step": 2399
+    },
+    {
+      "epoch": 0.2106469989659646,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0028715165208291265,
+      "loss": 1.2556,
+      "step": 2400
+    },
+    {
+      "epoch": 0.21073476854886708,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002871335881164925,
+      "loss": 1.3055,
+      "step": 2401
+    },
+    {
+      "epoch": 0.21082253813176957,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.0028711551209597043,
+      "loss": 1.299,
+      "step": 2402
+    },
+    {
+      "epoch": 0.21091030771467206,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.0028709742402313054,
+      "loss": 1.3579,
+      "step": 2403
+    },
+    {
+      "epoch": 0.21099807729757455,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.0028707932389975804,
+      "loss": 1.3595,
+      "step": 2404
+    },
+    {
+      "epoch": 0.21108584688047702,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.002870612117276393,
+      "loss": 1.2872,
+      "step": 2405
+    },
+    {
+      "epoch": 0.2111736164633795,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0028704308750856188,
+      "loss": 1.2803,
+      "step": 2406
+    },
+    {
+      "epoch": 0.211261386046282,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.002870249512443147,
+      "loss": 1.3525,
+      "step": 2407
+    },
+    {
+      "epoch": 0.2113491556291845,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.002870068029366876,
+      "loss": 1.2667,
+      "step": 2408
+    },
+    {
+      "epoch": 0.21143692521208696,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.002869886425874718,
+      "loss": 1.3334,
+      "step": 2409
+    },
+    {
+      "epoch": 0.21152469479498945,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002869704701984597,
+      "loss": 1.2823,
+      "step": 2410
+    },
+    {
+      "epoch": 0.21161246437789194,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0028695228577144483,
+      "loss": 1.3074,
+      "step": 2411
+    },
+    {
+      "epoch": 0.21170023396079443,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0028693408930822186,
+      "loss": 1.2667,
+      "step": 2412
+    },
+    {
+      "epoch": 0.21178800354369692,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002869158808105867,
+      "loss": 1.3875,
+      "step": 2413
+    },
+    {
+      "epoch": 0.2118757731265994,
+      "grad_norm": 0.06591796875,
+      "learning_rate": 0.0028689766028033655,
+      "loss": 1.302,
+      "step": 2414
+    },
+    {
+      "epoch": 0.21196354270950188,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0028687942771926968,
+      "loss": 1.334,
+      "step": 2415
+    },
+    {
+      "epoch": 0.21205131229240437,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0028686118312918553,
+      "loss": 1.2637,
+      "step": 2416
+    },
+    {
+      "epoch": 0.21213908187530686,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.0028684292651188475,
+      "loss": 1.3055,
+      "step": 2417
+    },
+    {
+      "epoch": 0.21222685145820933,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0028682465786916927,
+      "loss": 1.2792,
+      "step": 2418
+    },
+    {
+      "epoch": 0.21231462104111182,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0028680637720284205,
+      "loss": 1.3173,
+      "step": 2419
+    },
+    {
+      "epoch": 0.2124023906240143,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.002867880845147074,
+      "loss": 1.2788,
+      "step": 2420
+    },
+    {
+      "epoch": 0.2124901602069168,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0028676977980657066,
+      "loss": 1.2482,
+      "step": 2421
+    },
+    {
+      "epoch": 0.21257792978981926,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0028675146308023854,
+      "loss": 1.2852,
+      "step": 2422
+    },
+    {
+      "epoch": 0.21266569937272176,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002867331343375187,
+      "loss": 1.21,
+      "step": 2423
+    },
+    {
+      "epoch": 0.21275346895562425,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.002867147935802202,
+      "loss": 1.2382,
+      "step": 2424
+    },
+    {
+      "epoch": 0.21284123853852674,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.002866964408101532,
+      "loss": 1.3486,
+      "step": 2425
+    },
+    {
+      "epoch": 0.21292900812142923,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0028667807602912898,
+      "loss": 1.272,
+      "step": 2426
+    },
+    {
+      "epoch": 0.2130167777043317,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.002866596992389601,
+      "loss": 1.2788,
+      "step": 2427
+    },
+    {
+      "epoch": 0.2131045472872342,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.002866413104414603,
+      "loss": 1.2698,
+      "step": 2428
+    },
+    {
+      "epoch": 0.21319231687013668,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.002866229096384445,
+      "loss": 1.329,
+      "step": 2429
+    },
+    {
+      "epoch": 0.21328008645303917,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0028660449683172873,
+      "loss": 1.2794,
+      "step": 2430
+    },
+    {
+      "epoch": 0.21336785603594163,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.002865860720231303,
+      "loss": 1.2735,
+      "step": 2431
+    },
+    {
+      "epoch": 0.21345562561884412,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.002865676352144677,
+      "loss": 1.3595,
+      "step": 2432
+    },
+    {
+      "epoch": 0.21354339520174662,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0028654918640756044,
+      "loss": 1.2883,
+      "step": 2433
+    },
+    {
+      "epoch": 0.2136311647846491,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.0028653072560422947,
+      "loss": 1.2778,
+      "step": 2434
+    },
+    {
+      "epoch": 0.21371893436755157,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0028651225280629674,
+      "loss": 1.35,
+      "step": 2435
+    },
+    {
+      "epoch": 0.21380670395045406,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.0028649376801558545,
+      "loss": 1.2904,
+      "step": 2436
+    },
+    {
+      "epoch": 0.21389447353335655,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0028647527123391997,
+      "loss": 1.2926,
+      "step": 2437
+    },
+    {
+      "epoch": 0.21398224311625905,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.002864567624631259,
+      "loss": 1.322,
+      "step": 2438
+    },
+    {
+      "epoch": 0.21407001269916154,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0028643824170502996,
+      "loss": 1.2998,
+      "step": 2439
+    },
+    {
+      "epoch": 0.214157782282064,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.002864197089614601,
+      "loss": 1.2738,
+      "step": 2440
+    },
+    {
+      "epoch": 0.2142455518649665,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0028640116423424536,
+      "loss": 1.3009,
+      "step": 2441
+    },
+    {
+      "epoch": 0.21433332144786899,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.0028638260752521607,
+      "loss": 1.2854,
+      "step": 2442
+    },
+    {
+      "epoch": 0.21442109103077148,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.0028636403883620367,
+      "loss": 1.2842,
+      "step": 2443
+    },
+    {
+      "epoch": 0.21450886061367394,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.002863454581690409,
+      "loss": 1.3257,
+      "step": 2444
+    },
+    {
+      "epoch": 0.21459663019657643,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0028632686552556162,
+      "loss": 1.2528,
+      "step": 2445
+    },
+    {
+      "epoch": 0.21468439977947892,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0028630826090760065,
+      "loss": 1.2651,
+      "step": 2446
+    },
+    {
+      "epoch": 0.21477216936238142,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0028628964431699436,
+      "loss": 1.3215,
+      "step": 2447
+    },
+    {
+      "epoch": 0.2148599389452839,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.002862710157555801,
+      "loss": 1.2939,
+      "step": 2448
+    },
+    {
+      "epoch": 0.21494770852818637,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.0028625237522519643,
+      "loss": 1.2623,
+      "step": 2449
+    },
+    {
+      "epoch": 0.21503547811108886,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.002862337227276831,
+      "loss": 1.314,
+      "step": 2450
+    },
+    {
+      "epoch": 0.21512324769399135,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0028621505826488106,
+      "loss": 1.2892,
+      "step": 2451
+    },
+    {
+      "epoch": 0.21521101727689385,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.0028619638183863235,
+      "loss": 1.2987,
+      "step": 2452
+    },
+    {
+      "epoch": 0.2152987868597963,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.002861776934507803,
+      "loss": 1.2992,
+      "step": 2453
+    },
+    {
+      "epoch": 0.2153865564426988,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0028615899310316943,
+      "loss": 1.31,
+      "step": 2454
+    },
+    {
+      "epoch": 0.2154743260256013,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0028614028079764534,
+      "loss": 1.2462,
+      "step": 2455
+    },
+    {
+      "epoch": 0.21556209560850378,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.002861215565360548,
+      "loss": 1.31,
+      "step": 2456
+    },
+    {
+      "epoch": 0.21564986519140625,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0028610282032024595,
+      "loss": 1.3042,
+      "step": 2457
+    },
+    {
+      "epoch": 0.21573763477430874,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.0028608407215206786,
+      "loss": 1.2576,
+      "step": 2458
+    },
+    {
+      "epoch": 0.21582540435721123,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.00286065312033371,
+      "loss": 1.3492,
+      "step": 2459
+    },
+    {
+      "epoch": 0.21591317394011372,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0028604653996600685,
+      "loss": 1.3167,
+      "step": 2460
+    },
+    {
+      "epoch": 0.21600094352301621,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0028602775595182816,
+      "loss": 1.2599,
+      "step": 2461
+    },
+    {
+      "epoch": 0.21608871310591868,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0028600895999268882,
+      "loss": 1.305,
+      "step": 2462
+    },
+    {
+      "epoch": 0.21617648268882117,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.0028599015209044404,
+      "loss": 1.2478,
+      "step": 2463
+    },
+    {
+      "epoch": 0.21626425227172366,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0028597133224694984,
+      "loss": 1.3179,
+      "step": 2464
+    },
+    {
+      "epoch": 0.21635202185462615,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.002859525004640639,
+      "loss": 1.3279,
+      "step": 2465
+    },
+    {
+      "epoch": 0.21643979143752862,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0028593365674364473,
+      "loss": 1.291,
+      "step": 2466
+    },
+    {
+      "epoch": 0.2165275610204311,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.002859148010875521,
+      "loss": 1.2743,
+      "step": 2467
+    },
+    {
+      "epoch": 0.2166153306033336,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0028589593349764714,
+      "loss": 1.2508,
+      "step": 2468
+    },
+    {
+      "epoch": 0.2167031001862361,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.002858770539757918,
+      "loss": 1.3308,
+      "step": 2469
+    },
+    {
+      "epoch": 0.21679086976913856,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0028585816252384958,
+      "loss": 1.2618,
+      "step": 2470
+    },
+    {
+      "epoch": 0.21687863935204105,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.002858392591436849,
+      "loss": 1.3408,
+      "step": 2471
+    },
+    {
+      "epoch": 0.21696640893494354,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.002858203438371635,
+      "loss": 1.2779,
+      "step": 2472
+    },
+    {
+      "epoch": 0.21705417851784603,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0028580141660615225,
+      "loss": 1.2347,
+      "step": 2473
+    },
+    {
+      "epoch": 0.21714194810074852,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.002857824774525191,
+      "loss": 1.2368,
+      "step": 2474
+    },
+    {
+      "epoch": 0.21722971768365099,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.002857635263781334,
+      "loss": 1.2678,
+      "step": 2475
+    },
+    {
+      "epoch": 0.21731748726655348,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0028574456338486547,
+      "loss": 1.2988,
+      "step": 2476
+    },
+    {
+      "epoch": 0.21740525684945597,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0028572558847458685,
+      "loss": 1.2508,
+      "step": 2477
+    },
+    {
+      "epoch": 0.21749302643235846,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.002857066016491704,
+      "loss": 1.3486,
+      "step": 2478
+    },
+    {
+      "epoch": 0.21758079601526092,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0028568760291048994,
+      "loss": 1.252,
+      "step": 2479
+    },
+    {
+      "epoch": 0.21766856559816342,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.002856685922604206,
+      "loss": 1.2619,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2177563351810659,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.002856495697008387,
+      "loss": 1.3153,
+      "step": 2481
+    },
+    {
+      "epoch": 0.2178441047639684,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.0028563053523362166,
+      "loss": 1.3575,
+      "step": 2482
+    },
+    {
+      "epoch": 0.2179318743468709,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002856114888606481,
+      "loss": 1.285,
+      "step": 2483
+    },
+    {
+      "epoch": 0.21801964392977335,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.002855924305837978,
+      "loss": 1.269,
+      "step": 2484
+    },
+    {
+      "epoch": 0.21810741351267585,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0028557336040495175,
+      "loss": 1.3152,
+      "step": 2485
+    },
+    {
+      "epoch": 0.21819518309557834,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.0028555427832599213,
+      "loss": 1.2733,
+      "step": 2486
+    },
+    {
+      "epoch": 0.21828295267848083,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0028553518434880224,
+      "loss": 1.3076,
+      "step": 2487
+    },
+    {
+      "epoch": 0.2183707222613833,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.0028551607847526663,
+      "loss": 1.2777,
+      "step": 2488
+    },
+    {
+      "epoch": 0.21845849184428578,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0028549696070727087,
+      "loss": 1.2325,
+      "step": 2489
+    },
+    {
+      "epoch": 0.21854626142718828,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002854778310467019,
+      "loss": 1.2831,
+      "step": 2490
+    },
+    {
+      "epoch": 0.21863403101009077,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.002854586894954478,
+      "loss": 1.2836,
+      "step": 2491
+    },
+    {
+      "epoch": 0.21872180059299323,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.0028543953605539756,
+      "loss": 1.3051,
+      "step": 2492
+    },
+    {
+      "epoch": 0.21880957017589572,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.002854203707284417,
+      "loss": 1.2443,
+      "step": 2493
+    },
+    {
+      "epoch": 0.21889733975879822,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.002854011935164717,
+      "loss": 1.3067,
+      "step": 2494
+    },
+    {
+      "epoch": 0.2189851093417007,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.002853820044213803,
+      "loss": 1.3076,
+      "step": 2495
+    },
+    {
+      "epoch": 0.2190728789246032,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0028536280344506147,
+      "loss": 1.3224,
+      "step": 2496
+    },
+    {
+      "epoch": 0.21916064850750566,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.002853435905894101,
+      "loss": 1.278,
+      "step": 2497
+    },
+    {
+      "epoch": 0.21924841809040815,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.002853243658563226,
+      "loss": 1.3356,
+      "step": 2498
+    },
+    {
+      "epoch": 0.21933618767331065,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.002853051292476963,
+      "loss": 1.3006,
+      "step": 2499
+    },
+    {
+      "epoch": 0.21942395725621314,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0028528588076542965,
+      "loss": 1.2974,
+      "step": 2500
+    },
+    {
+      "epoch": 0.21942395725621314,
+      "eval_loss": 1.2801631689071655,
+      "eval_runtime": 437.6333,
+      "eval_samples_per_second": 33.681,
+      "eval_steps_per_second": 8.42,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2195117268391156,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002852666204114226,
+      "loss": 1.305,
+      "step": 2501
+    },
+    {
+      "epoch": 0.2195994964220181,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0028524734818757597,
+      "loss": 1.3353,
+      "step": 2502
+    },
+    {
+      "epoch": 0.21968726600492058,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0028522806409579187,
+      "loss": 1.3063,
+      "step": 2503
+    },
+    {
+      "epoch": 0.21977503558782308,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0028520876813797353,
+      "loss": 1.3428,
+      "step": 2504
+    },
+    {
+      "epoch": 0.21986280517072554,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0028518946031602546,
+      "loss": 1.3381,
+      "step": 2505
+    },
+    {
+      "epoch": 0.21995057475362803,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002851701406318532,
+      "loss": 1.2715,
+      "step": 2506
+    },
+    {
+      "epoch": 0.22003834433653052,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.002851508090873636,
+      "loss": 1.251,
+      "step": 2507
+    },
+    {
+      "epoch": 0.22012611391943301,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0028513146568446453,
+      "loss": 1.272,
+      "step": 2508
+    },
+    {
+      "epoch": 0.2202138835023355,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002851121104250651,
+      "loss": 1.3294,
+      "step": 2509
+    },
+    {
+      "epoch": 0.22030165308523797,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0028509274331107565,
+      "loss": 1.2951,
+      "step": 2510
+    },
+    {
+      "epoch": 0.22038942266814046,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002850733643444076,
+      "loss": 1.2978,
+      "step": 2511
+    },
+    {
+      "epoch": 0.22047719225104295,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.002850539735269736,
+      "loss": 1.2558,
+      "step": 2512
+    },
+    {
+      "epoch": 0.22056496183394544,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.002850345708606875,
+      "loss": 1.3334,
+      "step": 2513
+    },
+    {
+      "epoch": 0.2206527314168479,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002850151563474641,
+      "loss": 1.2744,
+      "step": 2514
+    },
+    {
+      "epoch": 0.2207405009997504,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.002849957299892197,
+      "loss": 1.3208,
+      "step": 2515
+    },
+    {
+      "epoch": 0.2208282705826529,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.002849762917878716,
+      "loss": 1.2707,
+      "step": 2516
+    },
+    {
+      "epoch": 0.22091604016555538,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0028495684174533816,
+      "loss": 1.3309,
+      "step": 2517
+    },
+    {
+      "epoch": 0.22100380974845787,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002849373798635391,
+      "loss": 1.2875,
+      "step": 2518
+    },
+    {
+      "epoch": 0.22109157933136034,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0028491790614439522,
+      "loss": 1.304,
+      "step": 2519
+    },
+    {
+      "epoch": 0.22117934891426283,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.002848984205898285,
+      "loss": 1.2817,
+      "step": 2520
+    },
+    {
+      "epoch": 0.22126711849716532,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.002848789232017621,
+      "loss": 1.3154,
+      "step": 2521
+    },
+    {
+      "epoch": 0.2213548880800678,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0028485941398212027,
+      "loss": 1.2477,
+      "step": 2522
+    },
+    {
+      "epoch": 0.22144265766297028,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002848398929328286,
+      "loss": 1.3148,
+      "step": 2523
+    },
+    {
+      "epoch": 0.22153042724587277,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0028482036005581363,
+      "loss": 1.3134,
+      "step": 2524
+    },
+    {
+      "epoch": 0.22161819682877526,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002848008153530033,
+      "loss": 1.2918,
+      "step": 2525
+    },
+    {
+      "epoch": 0.22170596641167775,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002847812588263264,
+      "loss": 1.2822,
+      "step": 2526
+    },
+    {
+      "epoch": 0.22179373599458022,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.002847616904777133,
+      "loss": 1.2444,
+      "step": 2527
+    },
+    {
+      "epoch": 0.2218815055774827,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0028474211030909517,
+      "loss": 1.3081,
+      "step": 2528
+    },
+    {
+      "epoch": 0.2219692751603852,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.002847225183224046,
+      "loss": 1.2849,
+      "step": 2529
+    },
+    {
+      "epoch": 0.2220570447432877,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0028470291451957513,
+      "loss": 1.2453,
+      "step": 2530
+    },
+    {
+      "epoch": 0.22214481432619018,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0028468329890254168,
+      "loss": 1.2833,
+      "step": 2531
+    },
+    {
+      "epoch": 0.22223258390909265,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.0028466367147324018,
+      "loss": 1.2666,
+      "step": 2532
+    },
+    {
+      "epoch": 0.22232035349199514,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.002846440322336078,
+      "loss": 1.3726,
+      "step": 2533
+    },
+    {
+      "epoch": 0.22240812307489763,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0028462438118558283,
+      "loss": 1.2715,
+      "step": 2534
+    },
+    {
+      "epoch": 0.22249589265780012,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0028460471833110473,
+      "loss": 1.2975,
+      "step": 2535
+    },
+    {
+      "epoch": 0.22258366224070258,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002845850436721143,
+      "loss": 1.263,
+      "step": 2536
+    },
+    {
+      "epoch": 0.22267143182360508,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002845653572105531,
+      "loss": 1.2012,
+      "step": 2537
+    },
+    {
+      "epoch": 0.22275920140650757,
+      "grad_norm": 0.068359375,
+      "learning_rate": 0.002845456589483643,
+      "loss": 1.2756,
+      "step": 2538
+    },
+    {
+      "epoch": 0.22284697098941006,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.00284525948887492,
+      "loss": 1.2775,
+      "step": 2539
+    },
+    {
+      "epoch": 0.22293474057231252,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.0028450622702988143,
+      "loss": 1.2989,
+      "step": 2540
+    },
+    {
+      "epoch": 0.22302251015521501,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0028448649337747918,
+      "loss": 1.2722,
+      "step": 2541
+    },
+    {
+      "epoch": 0.2231102797381175,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.002844667479322328,
+      "loss": 1.2688,
+      "step": 2542
+    },
+    {
+      "epoch": 0.22319804932102,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.002844469906960911,
+      "loss": 1.3229,
+      "step": 2543
+    },
+    {
+      "epoch": 0.2232858189039225,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0028442722167100402,
+      "loss": 1.323,
+      "step": 2544
+    },
+    {
+      "epoch": 0.22337358848682495,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.0028440744085892274,
+      "loss": 1.2572,
+      "step": 2545
+    },
+    {
+      "epoch": 0.22346135806972744,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.002843876482617995,
+      "loss": 1.262,
+      "step": 2546
+    },
+    {
+      "epoch": 0.22354912765262994,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0028436784388158774,
+      "loss": 1.2959,
+      "step": 2547
+    },
+    {
+      "epoch": 0.22363689723553243,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.002843480277202422,
+      "loss": 1.3413,
+      "step": 2548
+    },
+    {
+      "epoch": 0.2237246668184349,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0028432819977971847,
+      "loss": 1.318,
+      "step": 2549
+    },
+    {
+      "epoch": 0.22381243640133738,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0028430836006197364,
+      "loss": 1.2518,
+      "step": 2550
+    },
+    {
+      "epoch": 0.22390020598423988,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.002842885085689657,
+      "loss": 1.3169,
+      "step": 2551
+    },
+    {
+      "epoch": 0.22398797556714237,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.00284268645302654,
+      "loss": 1.3088,
+      "step": 2552
+    },
+    {
+      "epoch": 0.22407574515004486,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0028424877026499893,
+      "loss": 1.2508,
+      "step": 2553
+    },
+    {
+      "epoch": 0.22416351473294732,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0028422888345796204,
+      "loss": 1.2534,
+      "step": 2554
+    },
+    {
+      "epoch": 0.2242512843158498,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0028420898488350616,
+      "loss": 1.2965,
+      "step": 2555
+    },
+    {
+      "epoch": 0.2243390538987523,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002841890745435952,
+      "loss": 1.3284,
+      "step": 2556
+    },
+    {
+      "epoch": 0.2244268234816548,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0028416915244019406,
+      "loss": 1.2437,
+      "step": 2557
+    },
+    {
+      "epoch": 0.22451459306455726,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.002841492185752692,
+      "loss": 1.3139,
+      "step": 2558
+    },
+    {
+      "epoch": 0.22460236264745975,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0028412927295078786,
+      "loss": 1.2503,
+      "step": 2559
+    },
+    {
+      "epoch": 0.22469013223036224,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002841093155687187,
+      "loss": 1.2863,
+      "step": 2560
+    },
+    {
+      "epoch": 0.22477790181326474,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.002840893464310314,
+      "loss": 1.2805,
+      "step": 2561
+    },
+    {
+      "epoch": 0.2248656713961672,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.002840693655396967,
+      "loss": 1.3579,
+      "step": 2562
+    },
+    {
+      "epoch": 0.2249534409790697,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002840493728966869,
+      "loss": 1.2618,
+      "step": 2563
+    },
+    {
+      "epoch": 0.22504121056197218,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.0028402936850397496,
+      "loss": 1.3318,
+      "step": 2564
+    },
+    {
+      "epoch": 0.22512898014487467,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.002840093523635354,
+      "loss": 1.2483,
+      "step": 2565
+    },
+    {
+      "epoch": 0.22521674972777717,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.002839893244773436,
+      "loss": 1.2382,
+      "step": 2566
+    },
+    {
+      "epoch": 0.22530451931067963,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002839692848473763,
+      "loss": 1.2556,
+      "step": 2567
+    },
+    {
+      "epoch": 0.22539228889358212,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0028394923347561133,
+      "loss": 1.298,
+      "step": 2568
+    },
+    {
+      "epoch": 0.2254800584764846,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002839291703640277,
+      "loss": 1.3292,
+      "step": 2569
+    },
+    {
+      "epoch": 0.2255678280593871,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.002839090955146055,
+      "loss": 1.3303,
+      "step": 2570
+    },
+    {
+      "epoch": 0.22565559764228957,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.002838890089293261,
+      "loss": 1.2944,
+      "step": 2571
+    },
+    {
+      "epoch": 0.22574336722519206,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.0028386891061017187,
+      "loss": 1.3799,
+      "step": 2572
+    },
+    {
+      "epoch": 0.22583113680809455,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0028384880055912654,
+      "loss": 1.2495,
+      "step": 2573
+    },
+    {
+      "epoch": 0.22591890639099704,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.002838286787781749,
+      "loss": 1.3019,
+      "step": 2574
+    },
+    {
+      "epoch": 0.2260066759738995,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.002838085452693028,
+      "loss": 1.3039,
+      "step": 2575
+    },
+    {
+      "epoch": 0.226094445556802,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.002837884000344974,
+      "loss": 1.2842,
+      "step": 2576
+    },
+    {
+      "epoch": 0.2261822151397045,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0028376824307574694,
+      "loss": 1.312,
+      "step": 2577
+    },
+    {
+      "epoch": 0.22626998472260698,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0028374807439504085,
+      "loss": 1.3599,
+      "step": 2578
+    },
+    {
+      "epoch": 0.22635775430550947,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0028372789399436966,
+      "loss": 1.2641,
+      "step": 2579
+    },
+    {
+      "epoch": 0.22644552388841194,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.002837077018757251,
+      "loss": 1.2717,
+      "step": 2580
+    },
+    {
+      "epoch": 0.22653329347131443,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0028368749804110007,
+      "loss": 1.2641,
+      "step": 2581
+    },
+    {
+      "epoch": 0.22662106305421692,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0028366728249248864,
+      "loss": 1.328,
+      "step": 2582
+    },
+    {
+      "epoch": 0.2267088326371194,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0028364705523188593,
+      "loss": 1.3176,
+      "step": 2583
+    },
+    {
+      "epoch": 0.22679660222002188,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.0028362681626128835,
+      "loss": 1.2509,
+      "step": 2584
+    },
+    {
+      "epoch": 0.22688437180292437,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002836065655826934,
+      "loss": 1.2126,
+      "step": 2585
+    },
+    {
+      "epoch": 0.22697214138582686,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.002835863031980997,
+      "loss": 1.3478,
+      "step": 2586
+    },
+    {
+      "epoch": 0.22705991096872935,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.002835660291095071,
+      "loss": 1.294,
+      "step": 2587
+    },
+    {
+      "epoch": 0.22714768055163184,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.002835457433189166,
+      "loss": 1.2759,
+      "step": 2588
+    },
+    {
+      "epoch": 0.2272354501345343,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002835254458283303,
+      "loss": 1.2792,
+      "step": 2589
+    },
+    {
+      "epoch": 0.2273232197174368,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.002835051366397514,
+      "loss": 1.317,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2274109893003393,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0028348481575518453,
+      "loss": 1.2758,
+      "step": 2591
+    },
+    {
+      "epoch": 0.22749875888324178,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.0028346448317663506,
+      "loss": 1.3313,
+      "step": 2592
+    },
+    {
+      "epoch": 0.22758652846614424,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0028344413890610994,
+      "loss": 1.2772,
+      "step": 2593
+    },
+    {
+      "epoch": 0.22767429804904674,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.002834237829456169,
+      "loss": 1.334,
+      "step": 2594
+    },
+    {
+      "epoch": 0.22776206763194923,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.002834034152971651,
+      "loss": 1.2577,
+      "step": 2595
+    },
+    {
+      "epoch": 0.22784983721485172,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0028338303596276465,
+      "loss": 1.2432,
+      "step": 2596
+    },
+    {
+      "epoch": 0.22793760679775418,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.00283362644944427,
+      "loss": 1.2904,
+      "step": 2597
+    },
+    {
+      "epoch": 0.22802537638065667,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 0.0028334224224416467,
+      "loss": 1.2992,
+      "step": 2598
+    },
+    {
+      "epoch": 0.22811314596355917,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0028332182786399126,
+      "loss": 1.2994,
+      "step": 2599
+    },
+    {
+      "epoch": 0.22820091554646166,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.002833014018059216,
+      "loss": 1.2872,
+      "step": 2600
+    },
+    {
+      "epoch": 0.22828868512936415,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.002832809640719717,
+      "loss": 1.3066,
+      "step": 2601
+    },
+    {
+      "epoch": 0.2283764547122666,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0028326051466415862,
+      "loss": 1.2929,
+      "step": 2602
+    },
+    {
+      "epoch": 0.2284642242951691,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.002832400535845007,
+      "loss": 1.2974,
+      "step": 2603
+    },
+    {
+      "epoch": 0.2285519938780716,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.002832195808350174,
+      "loss": 1.281,
+      "step": 2604
+    },
+    {
+      "epoch": 0.2286397634609741,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.0028319909641772913,
+      "loss": 1.2723,
+      "step": 2605
+    },
+    {
+      "epoch": 0.22872753304387655,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0028317860033465784,
+      "loss": 1.2695,
+      "step": 2606
+    },
+    {
+      "epoch": 0.22881530262677904,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0028315809258782623,
+      "loss": 1.3227,
+      "step": 2607
+    },
+    {
+      "epoch": 0.22890307220968154,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0028313757317925846,
+      "loss": 1.2679,
+      "step": 2608
+    },
+    {
+      "epoch": 0.22899084179258403,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0028311704211097964,
+      "loss": 1.2505,
+      "step": 2609
+    },
+    {
+      "epoch": 0.2290786113754865,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0028309649938501615,
+      "loss": 1.2932,
+      "step": 2610
+    },
+    {
+      "epoch": 0.22916638095838898,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0028307594500339545,
+      "loss": 1.3171,
+      "step": 2611
+    },
+    {
+      "epoch": 0.22925415054129147,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0028305537896814613,
+      "loss": 1.2893,
+      "step": 2612
+    },
+    {
+      "epoch": 0.22934192012419397,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.002830348012812981,
+      "loss": 1.2851,
+      "step": 2613
+    },
+    {
+      "epoch": 0.22942968970709646,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0028301421194488214,
+      "loss": 1.2582,
+      "step": 2614
+    },
+    {
+      "epoch": 0.22951745928999892,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002829936109609305,
+      "loss": 1.2551,
+      "step": 2615
+    },
+    {
+      "epoch": 0.2296052288729014,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.002829729983314763,
+      "loss": 1.2668,
+      "step": 2616
+    },
+    {
+      "epoch": 0.2296929984558039,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00282952374058554,
+      "loss": 1.3086,
+      "step": 2617
+    },
+    {
+      "epoch": 0.2297807680387064,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0028293173814419904,
+      "loss": 1.2954,
+      "step": 2618
+    },
+    {
+      "epoch": 0.22986853762160886,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0028291109059044818,
+      "loss": 1.3079,
+      "step": 2619
+    },
+    {
+      "epoch": 0.22995630720451135,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.002828904313993392,
+      "loss": 1.3284,
+      "step": 2620
+    },
+    {
+      "epoch": 0.23004407678741384,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0028286976057291116,
+      "loss": 1.2443,
+      "step": 2621
+    },
+    {
+      "epoch": 0.23013184637031633,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0028284907811320413,
+      "loss": 1.2909,
+      "step": 2622
+    },
+    {
+      "epoch": 0.23021961595321883,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0028282838402225938,
+      "loss": 1.2969,
+      "step": 2623
+    },
+    {
+      "epoch": 0.2303073855361213,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.002828076783021193,
+      "loss": 1.3221,
+      "step": 2624
+    },
+    {
+      "epoch": 0.23039515511902378,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.0028278696095482762,
+      "loss": 1.2783,
+      "step": 2625
+    },
+    {
+      "epoch": 0.23048292470192627,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0028276623198242893,
+      "loss": 1.3408,
+      "step": 2626
+    },
+    {
+      "epoch": 0.23057069428482876,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0028274549138696905,
+      "loss": 1.2444,
+      "step": 2627
+    },
+    {
+      "epoch": 0.23065846386773123,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002827247391704951,
+      "loss": 1.3747,
+      "step": 2628
+    },
+    {
+      "epoch": 0.23074623345063372,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.002827039753350552,
+      "loss": 1.2593,
+      "step": 2629
+    },
+    {
+      "epoch": 0.2308340030335362,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0028268319988269874,
+      "loss": 1.219,
+      "step": 2630
+    },
+    {
+      "epoch": 0.2309217726164387,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.00282662412815476,
+      "loss": 1.2553,
+      "step": 2631
+    },
+    {
+      "epoch": 0.23100954219934117,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0028264161413543874,
+      "loss": 1.2944,
+      "step": 2632
+    },
+    {
+      "epoch": 0.23109731178224366,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002826208038446396,
+      "loss": 1.2736,
+      "step": 2633
+    },
+    {
+      "epoch": 0.23118508136514615,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0028259998194513254,
+      "loss": 1.3153,
+      "step": 2634
+    },
+    {
+      "epoch": 0.23127285094804864,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0028257914843897265,
+      "loss": 1.2136,
+      "step": 2635
+    },
+    {
+      "epoch": 0.23136062053095113,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0028255830332821596,
+      "loss": 1.2581,
+      "step": 2636
+    },
+    {
+      "epoch": 0.2314483901138536,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0028253744661491994,
+      "loss": 1.2819,
+      "step": 2637
+    },
+    {
+      "epoch": 0.2315361596967561,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0028251657830114297,
+      "loss": 1.2644,
+      "step": 2638
+    },
+    {
+      "epoch": 0.23162392927965858,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.0028249569838894473,
+      "loss": 1.3334,
+      "step": 2639
+    },
+    {
+      "epoch": 0.23171169886256107,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.0028247480688038595,
+      "loss": 1.252,
+      "step": 2640
+    },
+    {
+      "epoch": 0.23179946844546354,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.002824539037775286,
+      "loss": 1.3316,
+      "step": 2641
+    },
+    {
+      "epoch": 0.23188723802836603,
+      "grad_norm": 0.061767578125,
+      "learning_rate": 0.0028243298908243565,
+      "loss": 1.2814,
+      "step": 2642
+    },
+    {
+      "epoch": 0.23197500761126852,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.002824120627971713,
+      "loss": 1.2514,
+      "step": 2643
+    },
+    {
+      "epoch": 0.232062777194171,
+      "grad_norm": 0.0615234375,
+      "learning_rate": 0.00282391124923801,
+      "loss": 1.2649,
+      "step": 2644
+    },
+    {
+      "epoch": 0.23215054677707347,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0028237017546439117,
+      "loss": 1.2262,
+      "step": 2645
+    },
+    {
+      "epoch": 0.23223831635997597,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.002823492144210094,
+      "loss": 1.2959,
+      "step": 2646
+    },
+    {
+      "epoch": 0.23232608594287846,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002823282417957245,
+      "loss": 1.3092,
+      "step": 2647
+    },
+    {
+      "epoch": 0.23241385552578095,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.002823072575906064,
+      "loss": 1.3097,
+      "step": 2648
+    },
+    {
+      "epoch": 0.23250162510868344,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0028228626180772606,
+      "loss": 1.3089,
+      "step": 2649
+    },
+    {
+      "epoch": 0.2325893946915859,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.0028226525444915583,
+      "loss": 1.2939,
+      "step": 2650
+    },
+    {
+      "epoch": 0.2326771642744884,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0028224423551696893,
+      "loss": 1.3127,
+      "step": 2651
+    },
+    {
+      "epoch": 0.2327649338573909,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.0028222320501323996,
+      "loss": 1.2854,
+      "step": 2652
+    },
+    {
+      "epoch": 0.23285270344029338,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.002822021629400444,
+      "loss": 1.3329,
+      "step": 2653
+    },
+    {
+      "epoch": 0.23294047302319584,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002821811092994592,
+      "loss": 1.3233,
+      "step": 2654
+    },
+    {
+      "epoch": 0.23302824260609833,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0028216004409356214,
+      "loss": 1.3286,
+      "step": 2655
+    },
+    {
+      "epoch": 0.23311601218900083,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.002821389673244323,
+      "loss": 1.2255,
+      "step": 2656
+    },
+    {
+      "epoch": 0.23320378177190332,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0028211787899414986,
+      "loss": 1.2183,
+      "step": 2657
+    },
+    {
+      "epoch": 0.2332915513548058,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.002820967791047962,
+      "loss": 1.2769,
+      "step": 2658
+    },
+    {
+      "epoch": 0.23337932093770827,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.0028207566765845377,
+      "loss": 1.206,
+      "step": 2659
+    },
+    {
+      "epoch": 0.23346709052061076,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0028205454465720613,
+      "loss": 1.2981,
+      "step": 2660
+    },
+    {
+      "epoch": 0.23355486010351326,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.002820334101031382,
+      "loss": 1.2803,
+      "step": 2661
+    },
+    {
+      "epoch": 0.23364262968641575,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 0.002820122639983357,
+      "loss": 1.2925,
+      "step": 2662
+    },
+    {
+      "epoch": 0.2337303992693182,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.0028199110634488573,
+      "loss": 1.3133,
+      "step": 2663
+    },
+    {
+      "epoch": 0.2338181688522207,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.002819699371448765,
+      "loss": 1.2847,
+      "step": 2664
+    },
+    {
+      "epoch": 0.2339059384351232,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.002819487564003973,
+      "loss": 1.2373,
+      "step": 2665
+    },
+    {
+      "epoch": 0.2339937080180257,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.002819275641135386,
+      "loss": 1.2995,
+      "step": 2666
+    },
+    {
+      "epoch": 0.23408147760092815,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.0028190636028639195,
+      "loss": 1.1893,
+      "step": 2667
+    },
+    {
+      "epoch": 0.23416924718383064,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0028188514492105015,
+      "loss": 1.2701,
+      "step": 2668
+    },
+    {
+      "epoch": 0.23425701676673313,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0028186391801960703,
+      "loss": 1.3307,
+      "step": 2669
+    },
+    {
+      "epoch": 0.23434478634963563,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0028184267958415755,
+      "loss": 1.3358,
+      "step": 2670
+    },
+    {
+      "epoch": 0.23443255593253812,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.0028182142961679798,
+      "loss": 1.3149,
+      "step": 2671
+    },
+    {
+      "epoch": 0.23452032551544058,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0028180016811962558,
+      "loss": 1.3222,
+      "step": 2672
+    },
+    {
+      "epoch": 0.23460809509834307,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.0028177889509473863,
+      "loss": 1.3285,
+      "step": 2673
+    },
+    {
+      "epoch": 0.23469586468124556,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002817576105442369,
+      "loss": 1.2687,
+      "step": 2674
+    },
+    {
+      "epoch": 0.23478363426414806,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0028173631447022097,
+      "loss": 1.2757,
+      "step": 2675
+    },
+    {
+      "epoch": 0.23487140384705052,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0028171500687479267,
+      "loss": 1.2429,
+      "step": 2676
+    },
+    {
+      "epoch": 0.234959173429953,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0028169368776005503,
+      "loss": 1.2815,
+      "step": 2677
+    },
+    {
+      "epoch": 0.2350469430128555,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0028167235712811215,
+      "loss": 1.2862,
+      "step": 2678
+    },
+    {
+      "epoch": 0.235134712595758,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.002816510149810693,
+      "loss": 1.2643,
+      "step": 2679
+    },
+    {
+      "epoch": 0.23522248217866046,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.0028162966132103274,
+      "loss": 1.343,
+      "step": 2680
+    },
+    {
+      "epoch": 0.23531025176156295,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.002816082961501102,
+      "loss": 1.342,
+      "step": 2681
+    },
+    {
+      "epoch": 0.23539802134446544,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0028158691947041017,
+      "loss": 1.1977,
+      "step": 2682
+    },
+    {
+      "epoch": 0.23548579092736793,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002815655312840425,
+      "loss": 1.3142,
+      "step": 2683
+    },
+    {
+      "epoch": 0.23557356051027042,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.0028154413159311815,
+      "loss": 1.2546,
+      "step": 2684
+    },
+    {
+      "epoch": 0.2356613300931729,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002815227203997491,
+      "loss": 1.2795,
+      "step": 2685
+    },
+    {
+      "epoch": 0.23574909967607538,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0028150129770604868,
+      "loss": 1.2788,
+      "step": 2686
+    },
+    {
+      "epoch": 0.23583686925897787,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0028147986351413105,
+      "loss": 1.2825,
+      "step": 2687
+    },
+    {
+      "epoch": 0.23592463884188036,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0028145841782611183,
+      "loss": 1.316,
+      "step": 2688
+    },
+    {
+      "epoch": 0.23601240842478283,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.002814369606441076,
+      "loss": 1.2949,
+      "step": 2689
+    },
+    {
+      "epoch": 0.23610017800768532,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0028141549197023604,
+      "loss": 1.3241,
+      "step": 2690
+    },
+    {
+      "epoch": 0.2361879475905878,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0028139401180661606,
+      "loss": 1.3208,
+      "step": 2691
+    },
+    {
+      "epoch": 0.2362757171734903,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0028137252015536765,
+      "loss": 1.2905,
+      "step": 2692
+    },
+    {
+      "epoch": 0.2363634867563928,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.00281351017018612,
+      "loss": 1.2847,
+      "step": 2693
+    },
+    {
+      "epoch": 0.23645125633929526,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.002813295023984713,
+      "loss": 1.2918,
+      "step": 2694
+    },
+    {
+      "epoch": 0.23653902592219775,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0028130797629706905,
+      "loss": 1.2487,
+      "step": 2695
+    },
+    {
+      "epoch": 0.23662679550510024,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0028128643871652976,
+      "loss": 1.2858,
+      "step": 2696
+    },
+    {
+      "epoch": 0.23671456508800273,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.0028126488965897903,
+      "loss": 1.2701,
+      "step": 2697
+    },
+    {
+      "epoch": 0.2368023346709052,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0028124332912654376,
+      "loss": 1.2544,
+      "step": 2698
+    },
+    {
+      "epoch": 0.2368901042538077,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0028122175712135184,
+      "loss": 1.2927,
+      "step": 2699
+    },
+    {
+      "epoch": 0.23697787383671018,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002812001736455324,
+      "loss": 1.2842,
+      "step": 2700
+    },
+    {
+      "epoch": 0.23706564341961267,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0028117857870121558,
+      "loss": 1.3013,
+      "step": 2701
+    },
+    {
+      "epoch": 0.23715341300251513,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.002811569722905327,
+      "loss": 1.2799,
+      "step": 2702
+    },
+    {
+      "epoch": 0.23724118258541763,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.002811353544156163,
+      "loss": 1.3579,
+      "step": 2703
+    },
+    {
+      "epoch": 0.23732895216832012,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0028111372507859996,
+      "loss": 1.2623,
+      "step": 2704
+    },
+    {
+      "epoch": 0.2374167217512226,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0028109208428161838,
+      "loss": 1.3012,
+      "step": 2705
+    },
+    {
+      "epoch": 0.2375044913341251,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.002810704320268074,
+      "loss": 1.2589,
+      "step": 2706
+    },
+    {
+      "epoch": 0.23759226091702756,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.002810487683163041,
+      "loss": 1.2362,
+      "step": 2707
+    },
+    {
+      "epoch": 0.23768003049993006,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0028102709315224656,
+      "loss": 1.2701,
+      "step": 2708
+    },
+    {
+      "epoch": 0.23776780008283255,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0028100540653677396,
+      "loss": 1.316,
+      "step": 2709
+    },
+    {
+      "epoch": 0.23785556966573504,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002809837084720268,
+      "loss": 1.2557,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2379433392486375,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0028096199896014648,
+      "loss": 1.2801,
+      "step": 2711
+    },
+    {
+      "epoch": 0.23803110883154,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0028094027800327575,
+      "loss": 1.3095,
+      "step": 2712
+    },
+    {
+      "epoch": 0.2381188784144425,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0028091854560355835,
+      "loss": 1.2497,
+      "step": 2713
+    },
+    {
+      "epoch": 0.23820664799734498,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002808968017631391,
+      "loss": 1.3031,
+      "step": 2714
+    },
+    {
+      "epoch": 0.23829441758024744,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.002808750464841641,
+      "loss": 1.279,
+      "step": 2715
+    },
+    {
+      "epoch": 0.23838218716314993,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002808532797687806,
+      "loss": 1.3033,
+      "step": 2716
+    },
+    {
+      "epoch": 0.23846995674605242,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.002808315016191367,
+      "loss": 1.2536,
+      "step": 2717
+    },
+    {
+      "epoch": 0.23855772632895492,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0028080971203738193,
+      "loss": 1.238,
+      "step": 2718
+    },
+    {
+      "epoch": 0.2386454959118574,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.0028078791102566685,
+      "loss": 1.2741,
+      "step": 2719
+    },
+    {
+      "epoch": 0.23873326549475987,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0028076609858614307,
+      "loss": 1.3111,
+      "step": 2720
+    },
+    {
+      "epoch": 0.23882103507766236,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0028074427472096343,
+      "loss": 1.2899,
+      "step": 2721
+    },
+    {
+      "epoch": 0.23890880466056486,
+      "grad_norm": 0.06787109375,
+      "learning_rate": 0.002807224394322819,
+      "loss": 1.2425,
+      "step": 2722
+    },
+    {
+      "epoch": 0.23899657424346735,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002807005927222534,
+      "loss": 1.3177,
+      "step": 2723
+    },
+    {
+      "epoch": 0.2390843438263698,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.002806787345930343,
+      "loss": 1.2674,
+      "step": 2724
+    },
+    {
+      "epoch": 0.2391721134092723,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.0028065686504678174,
+      "loss": 1.2979,
+      "step": 2725
+    },
+    {
+      "epoch": 0.2392598829921748,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.002806349840856542,
+      "loss": 1.3585,
+      "step": 2726
+    },
+    {
+      "epoch": 0.23934765257507729,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0028061309171181132,
+      "loss": 1.297,
+      "step": 2727
+    },
+    {
+      "epoch": 0.23943542215797978,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0028059118792741373,
+      "loss": 1.2898,
+      "step": 2728
+    },
+    {
+      "epoch": 0.23952319174088224,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.0028056927273462334,
+      "loss": 1.2496,
+      "step": 2729
+    },
+    {
+      "epoch": 0.23961096132378473,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.00280547346135603,
+      "loss": 1.2823,
+      "step": 2730
+    },
+    {
+      "epoch": 0.23969873090668722,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.002805254081325167,
+      "loss": 1.2302,
+      "step": 2731
+    },
+    {
+      "epoch": 0.23978650048958972,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.002805034587275298,
+      "loss": 1.2376,
+      "step": 2732
+    },
+    {
+      "epoch": 0.23987427007249218,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0028048149792280854,
+      "loss": 1.3058,
+      "step": 2733
+    },
+    {
+      "epoch": 0.23996203965539467,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.0028045952572052033,
+      "loss": 1.3163,
+      "step": 2734
+    },
+    {
+      "epoch": 0.24004980923829716,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0028043754212283382,
+      "loss": 1.2277,
+      "step": 2735
+    },
+    {
+      "epoch": 0.24013757882119965,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.002804155471319187,
+      "loss": 1.2385,
+      "step": 2736
+    },
+    {
+      "epoch": 0.24022534840410212,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.002803935407499457,
+      "loss": 1.2289,
+      "step": 2737
+    },
+    {
+      "epoch": 0.2403131179870046,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.0028037152297908687,
+      "loss": 1.2196,
+      "step": 2738
+    },
+    {
+      "epoch": 0.2404008875699071,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0028034949382151515,
+      "loss": 1.2046,
+      "step": 2739
+    },
+    {
+      "epoch": 0.2404886571528096,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.0028032745327940486,
+      "loss": 1.2787,
+      "step": 2740
+    },
+    {
+      "epoch": 0.24057642673571208,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0028030540135493124,
+      "loss": 1.2919,
+      "step": 2741
+    },
+    {
+      "epoch": 0.24066419631861455,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.0028028333805027073,
+      "loss": 1.2882,
+      "step": 2742
+    },
+    {
+      "epoch": 0.24075196590151704,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0028026126336760096,
+      "loss": 1.3052,
+      "step": 2743
+    },
+    {
+      "epoch": 0.24083973548441953,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.0028023917730910053,
+      "loss": 1.3639,
+      "step": 2744
+    },
+    {
+      "epoch": 0.24092750506732202,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0028021707987694925,
+      "loss": 1.2574,
+      "step": 2745
+    },
+    {
+      "epoch": 0.2410152746502245,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.0028019497107332814,
+      "loss": 1.2762,
+      "step": 2746
+    },
+    {
+      "epoch": 0.24110304423312698,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002801728509004191,
+      "loss": 1.3217,
+      "step": 2747
+    },
+    {
+      "epoch": 0.24119081381602947,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002801507193604054,
+      "loss": 1.337,
+      "step": 2748
+    },
+    {
+      "epoch": 0.24127858339893196,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.0028012857645547144,
+      "loss": 1.249,
+      "step": 2749
+    },
+    {
+      "epoch": 0.24136635298183443,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0028010642218780244,
+      "loss": 1.289,
+      "step": 2750
+    },
+    {
+      "epoch": 0.24145412256473692,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0028008425655958505,
+      "loss": 1.3001,
+      "step": 2751
+    },
+    {
+      "epoch": 0.2415418921476394,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.002800620795730069,
+      "loss": 1.2688,
+      "step": 2752
+    },
+    {
+      "epoch": 0.2416296617305419,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0028003989123025676,
+      "loss": 1.3052,
+      "step": 2753
+    },
+    {
+      "epoch": 0.2417174313134444,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.0028001769153352457,
+      "loss": 1.345,
+      "step": 2754
+    },
+    {
+      "epoch": 0.24180520089634686,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.002799954804850013,
+      "loss": 1.2633,
+      "step": 2755
+    },
+    {
+      "epoch": 0.24189297047924935,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.0027997325808687915,
+      "loss": 1.2989,
+      "step": 2756
+    },
+    {
+      "epoch": 0.24198074006215184,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.002799510243413513,
+      "loss": 1.3136,
+      "step": 2757
+    },
+    {
+      "epoch": 0.24206850964505433,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.0027992877925061225,
+      "loss": 1.2612,
+      "step": 2758
+    },
+    {
+      "epoch": 0.2421562792279568,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.002799065228168574,
+      "loss": 1.3486,
+      "step": 2759
+    },
+    {
+      "epoch": 0.24224404881085929,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0027988425504228342,
+      "loss": 1.2895,
+      "step": 2760
+    },
+    {
+      "epoch": 0.24233181839376178,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0027986197592908806,
+      "loss": 1.3472,
+      "step": 2761
+    },
+    {
+      "epoch": 0.24241958797666427,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.002798396854794701,
+      "loss": 1.2866,
+      "step": 2762
+    },
+    {
+      "epoch": 0.24250735755956676,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.002798173836956296,
+      "loss": 1.291,
+      "step": 2763
+    },
+    {
+      "epoch": 0.24259512714246922,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.002797950705797677,
+      "loss": 1.2936,
+      "step": 2764
+    },
+    {
+      "epoch": 0.24268289672537172,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002797727461340865,
+      "loss": 1.2728,
+      "step": 2765
+    },
+    {
+      "epoch": 0.2427706663082742,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.002797504103607894,
+      "loss": 1.2896,
+      "step": 2766
+    },
+    {
+      "epoch": 0.2428584358911767,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0027972806326208087,
+      "loss": 1.2608,
+      "step": 2767
+    },
+    {
+      "epoch": 0.24294620547407916,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0027970570484016645,
+      "loss": 1.2772,
+      "step": 2768
+    },
+    {
+      "epoch": 0.24303397505698165,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.002796833350972528,
+      "loss": 1.2883,
+      "step": 2769
+    },
+    {
+      "epoch": 0.24312174463988415,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0027966095403554776,
+      "loss": 1.3528,
+      "step": 2770
+    },
+    {
+      "epoch": 0.24320951422278664,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.002796385616572602,
+      "loss": 1.228,
+      "step": 2771
+    },
+    {
+      "epoch": 0.2432972838056891,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0027961615796460028,
+      "loss": 1.3179,
+      "step": 2772
+    },
+    {
+      "epoch": 0.2433850533885916,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0027959374295977906,
+      "loss": 1.2832,
+      "step": 2773
+    },
+    {
+      "epoch": 0.24347282297149409,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0027957131664500885,
+      "loss": 1.2588,
+      "step": 2774
+    },
+    {
+      "epoch": 0.24356059255439658,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0027954887902250304,
+      "loss": 1.2755,
+      "step": 2775
+    },
+    {
+      "epoch": 0.24364836213729907,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.0027952643009447604,
+      "loss": 1.2313,
+      "step": 2776
+    },
+    {
+      "epoch": 0.24373613172020153,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.002795039698631436,
+      "loss": 1.331,
+      "step": 2777
+    },
+    {
+      "epoch": 0.24382390130310402,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.002794814983307224,
+      "loss": 1.2873,
+      "step": 2778
+    },
+    {
+      "epoch": 0.24391167088600652,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.0027945901549943026,
+      "loss": 1.2511,
+      "step": 2779
+    },
+    {
+      "epoch": 0.243999440468909,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0027943652137148623,
+      "loss": 1.2692,
+      "step": 2780
+    },
+    {
+      "epoch": 0.24408721005181147,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.002794140159491103,
+      "loss": 1.2691,
+      "step": 2781
+    },
+    {
+      "epoch": 0.24417497963471396,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.002793914992345238,
+      "loss": 1.2996,
+      "step": 2782
+    },
+    {
+      "epoch": 0.24426274921761645,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0027936897122994887,
+      "loss": 1.2517,
+      "step": 2783
+    },
+    {
+      "epoch": 0.24435051880051895,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.00279346431937609,
+      "loss": 1.2767,
+      "step": 2784
+    },
+    {
+      "epoch": 0.2444382883834214,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.002793238813597288,
+      "loss": 1.2524,
+      "step": 2785
+    },
+    {
+      "epoch": 0.2445260579663239,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.002793013194985338,
+      "loss": 1.294,
+      "step": 2786
+    },
+    {
+      "epoch": 0.2446138275492264,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.0027927874635625093,
+      "loss": 1.2917,
+      "step": 2787
+    },
+    {
+      "epoch": 0.24470159713212888,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.002792561619351079,
+      "loss": 1.2642,
+      "step": 2788
+    },
+    {
+      "epoch": 0.24478936671503138,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.002792335662373338,
+      "loss": 1.3205,
+      "step": 2789
+    },
+    {
+      "epoch": 0.24487713629793384,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.002792109592651587,
+      "loss": 1.2848,
+      "step": 2790
+    },
+    {
+      "epoch": 0.24496490588083633,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.002791883410208138,
+      "loss": 1.3269,
+      "step": 2791
+    },
+    {
+      "epoch": 0.24505267546373882,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.0027916571150653157,
+      "loss": 1.2572,
+      "step": 2792
+    },
+    {
+      "epoch": 0.24514044504664131,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.0027914307072454524,
+      "loss": 1.2437,
+      "step": 2793
+    },
+    {
+      "epoch": 0.24522821462954378,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002791204186770895,
+      "loss": 1.2579,
+      "step": 2794
+    },
+    {
+      "epoch": 0.24531598421244627,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0027909775536639997,
+      "loss": 1.2755,
+      "step": 2795
+    },
+    {
+      "epoch": 0.24540375379534876,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.002790750807947135,
+      "loss": 1.3189,
+      "step": 2796
+    },
+    {
+      "epoch": 0.24549152337825125,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0027905239496426788,
+      "loss": 1.3053,
+      "step": 2797
+    },
+    {
+      "epoch": 0.24557929296115374,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.002790296978773022,
+      "loss": 1.3057,
+      "step": 2798
+    },
+    {
+      "epoch": 0.2456670625440562,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0027900698953605653,
+      "loss": 1.3154,
+      "step": 2799
+    },
+    {
+      "epoch": 0.2457548321269587,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0027898426994277204,
+      "loss": 1.3037,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2458426017098612,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0027896153909969116,
+      "loss": 1.2648,
+      "step": 2801
+    },
+    {
+      "epoch": 0.24593037129276368,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.002789387970090573,
+      "loss": 1.2515,
+      "step": 2802
+    },
+    {
+      "epoch": 0.24601814087566615,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.00278916043673115,
+      "loss": 1.2333,
+      "step": 2803
+    },
+    {
+      "epoch": 0.24610591045856864,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002788932790941099,
+      "loss": 1.2861,
+      "step": 2804
+    },
+    {
+      "epoch": 0.24619368004147113,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.0027887050327428885,
+      "loss": 1.2821,
+      "step": 2805
+    },
+    {
+      "epoch": 0.24628144962437362,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.002788477162158997,
+      "loss": 1.2608,
+      "step": 2806
+    },
+    {
+      "epoch": 0.24636921920727609,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.002788249179211914,
+      "loss": 1.2737,
+      "step": 2807
+    },
+    {
+      "epoch": 0.24645698879017858,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0027880210839241405,
+      "loss": 1.2698,
+      "step": 2808
+    },
+    {
+      "epoch": 0.24654475837308107,
+      "grad_norm": 0.060546875,
+      "learning_rate": 0.0027877928763181898,
+      "loss": 1.2469,
+      "step": 2809
+    },
+    {
+      "epoch": 0.24663252795598356,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0027875645564165836,
+      "loss": 1.2851,
+      "step": 2810
+    },
+    {
+      "epoch": 0.24672029753888605,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.002787336124241857,
+      "loss": 1.3066,
+      "step": 2811
+    },
+    {
+      "epoch": 0.24680806712178852,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.002787107579816555,
+      "loss": 1.2742,
+      "step": 2812
+    },
+    {
+      "epoch": 0.246895836704691,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.0027868789231632342,
+      "loss": 1.2275,
+      "step": 2813
+    },
+    {
+      "epoch": 0.2469836062875935,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0027866501543044623,
+      "loss": 1.2392,
+      "step": 2814
+    },
+    {
+      "epoch": 0.247071375870496,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0027864212732628176,
+      "loss": 1.2716,
+      "step": 2815
+    },
+    {
+      "epoch": 0.24715914545339845,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.00278619228006089,
+      "loss": 1.3828,
+      "step": 2816
+    },
+    {
+      "epoch": 0.24724691503630095,
+      "grad_norm": 0.125,
+      "learning_rate": 0.00278596317472128,
+      "loss": 1.288,
+      "step": 2817
+    },
+    {
+      "epoch": 0.24733468461920344,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.0027857339572665988,
+      "loss": 1.3191,
+      "step": 2818
+    },
+    {
+      "epoch": 0.24742245420210593,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.002785504627719471,
+      "loss": 1.2204,
+      "step": 2819
+    },
+    {
+      "epoch": 0.2475102237850084,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.002785275186102528,
+      "loss": 1.2903,
+      "step": 2820
+    },
+    {
+      "epoch": 0.24759799336791088,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0027850456324384175,
+      "loss": 1.2806,
+      "step": 2821
+    },
+    {
+      "epoch": 0.24768576295081338,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0027848159667497934,
+      "loss": 1.2507,
+      "step": 2822
+    },
+    {
+      "epoch": 0.24777353253371587,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.002784586189059324,
+      "loss": 1.2892,
+      "step": 2823
+    },
+    {
+      "epoch": 0.24786130211661836,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002784356299389687,
+      "loss": 1.3613,
+      "step": 2824
+    },
+    {
+      "epoch": 0.24794907169952082,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.0027841262977635717,
+      "loss": 1.295,
+      "step": 2825
+    },
+    {
+      "epoch": 0.24803684128242331,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.002783896184203678,
+      "loss": 1.2566,
+      "step": 2826
+    },
+    {
+      "epoch": 0.2481246108653258,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.002783665958732718,
+      "loss": 1.2489,
+      "step": 2827
+    },
+    {
+      "epoch": 0.2482123804482283,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.002783435621373413,
+      "loss": 1.3062,
+      "step": 2828
+    },
+    {
+      "epoch": 0.24830015003113076,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0027832051721484972,
+      "loss": 1.2788,
+      "step": 2829
+    },
+    {
+      "epoch": 0.24838791961403325,
+      "grad_norm": 0.058837890625,
+      "learning_rate": 0.0027829746110807152,
+      "loss": 1.2482,
+      "step": 2830
+    },
+    {
+      "epoch": 0.24847568919693575,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.002782743938192822,
+      "loss": 1.3907,
+      "step": 2831
+    },
+    {
+      "epoch": 0.24856345877983824,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0027825131535075837,
+      "loss": 1.3046,
+      "step": 2832
+    },
+    {
+      "epoch": 0.24865122836274073,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0027822822570477785,
+      "loss": 1.3073,
+      "step": 2833
+    },
+    {
+      "epoch": 0.2487389979456432,
+      "grad_norm": 0.06591796875,
+      "learning_rate": 0.002782051248836195,
+      "loss": 1.2559,
+      "step": 2834
+    },
+    {
+      "epoch": 0.24882676752854568,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0027818201288956315,
+      "loss": 1.3289,
+      "step": 2835
+    },
+    {
+      "epoch": 0.24891453711144818,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.002781588897248901,
+      "loss": 1.2602,
+      "step": 2836
+    },
+    {
+      "epoch": 0.24900230669435067,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.002781357553918823,
+      "loss": 1.2663,
+      "step": 2837
+    },
+    {
+      "epoch": 0.24909007627725313,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0027811260989282316,
+      "loss": 1.3229,
+      "step": 2838
+    },
+    {
+      "epoch": 0.24917784586015562,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.002780894532299969,
+      "loss": 1.2647,
+      "step": 2839
+    },
+    {
+      "epoch": 0.24926561544305811,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.0027806628540568913,
+      "loss": 1.3106,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2493533850259606,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.002780431064221864,
+      "loss": 1.3421,
+      "step": 2841
+    },
+    {
+      "epoch": 0.24944115460886307,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.002780199162817763,
+      "loss": 1.3106,
+      "step": 2842
+    },
+    {
+      "epoch": 0.24952892419176556,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.002779967149867477,
+      "loss": 1.2301,
+      "step": 2843
+    },
+    {
+      "epoch": 0.24961669377466805,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0027797350253939038,
+      "loss": 1.307,
+      "step": 2844
+    },
+    {
+      "epoch": 0.24970446335757054,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.002779502789419954,
+      "loss": 1.3101,
+      "step": 2845
+    },
+    {
+      "epoch": 0.24979223294047304,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0027792704419685485,
+      "loss": 1.2597,
+      "step": 2846
+    },
+    {
+      "epoch": 0.2498800025233755,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.002779037983062618,
+      "loss": 1.2363,
+      "step": 2847
+    },
+    {
+      "epoch": 0.249967772106278,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0027788054127251068,
+      "loss": 1.2775,
+      "step": 2848
+    },
+    {
+      "epoch": 0.2500555416891805,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0027785727309789674,
+      "loss": 1.2657,
+      "step": 2849
+    },
+    {
+      "epoch": 0.25014331127208295,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0027783399378471653,
+      "loss": 1.3093,
+      "step": 2850
+    },
+    {
+      "epoch": 0.25023108085498547,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.002778107033352675,
+      "loss": 1.2647,
+      "step": 2851
+    },
+    {
+      "epoch": 0.25031885043788793,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.0027778740175184854,
+      "loss": 1.2843,
+      "step": 2852
+    },
+    {
+      "epoch": 0.2504066200207904,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.002777640890367592,
+      "loss": 1.3074,
+      "step": 2853
+    },
+    {
+      "epoch": 0.2504943896036929,
+      "grad_norm": 0.06201171875,
+      "learning_rate": 0.002777407651923005,
+      "loss": 1.3022,
+      "step": 2854
+    },
+    {
+      "epoch": 0.2505821591865954,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.002777174302207744,
+      "loss": 1.3172,
+      "step": 2855
+    },
+    {
+      "epoch": 0.2506699287694979,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.002776940841244839,
+      "loss": 1.3158,
+      "step": 2856
+    },
+    {
+      "epoch": 0.25075769835240036,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002776707269057332,
+      "loss": 1.257,
+      "step": 2857
+    },
+    {
+      "epoch": 0.2508454679353028,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.0027764735856682763,
+      "loss": 1.3238,
+      "step": 2858
+    },
+    {
+      "epoch": 0.25093323751820534,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0027762397911007338,
+      "loss": 1.2507,
+      "step": 2859
+    },
+    {
+      "epoch": 0.2510210071011078,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0027760058853777813,
+      "loss": 1.3555,
+      "step": 2860
+    },
+    {
+      "epoch": 0.2511087766840103,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.002775771868522503,
+      "loss": 1.3293,
+      "step": 2861
+    },
+    {
+      "epoch": 0.2511965462669128,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0027755377405579954,
+      "loss": 1.2397,
+      "step": 2862
+    },
+    {
+      "epoch": 0.25128431584981525,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.0027753035015073667,
+      "loss": 1.3477,
+      "step": 2863
+    },
+    {
+      "epoch": 0.2513720854327178,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0027750691513937346,
+      "loss": 1.2326,
+      "step": 2864
+    },
+    {
+      "epoch": 0.25145985501562024,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0027748346902402295,
+      "loss": 1.2899,
+      "step": 2865
+    },
+    {
+      "epoch": 0.2515476245985227,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0027746001180699907,
+      "loss": 1.3252,
+      "step": 2866
+    },
+    {
+      "epoch": 0.2516353941814252,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00277436543490617,
+      "loss": 1.2278,
+      "step": 2867
+    },
+    {
+      "epoch": 0.2517231637643277,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0027741306407719304,
+      "loss": 1.2682,
+      "step": 2868
+    },
+    {
+      "epoch": 0.2518109333472302,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.002773895735690444,
+      "loss": 1.2905,
+      "step": 2869
+    },
+    {
+      "epoch": 0.25189870293013267,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0027736607196848957,
+      "loss": 1.3347,
+      "step": 2870
+    },
+    {
+      "epoch": 0.25198647251303513,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0027734255927784803,
+      "loss": 1.2919,
+      "step": 2871
+    },
+    {
+      "epoch": 0.25207424209593765,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0027731903549944046,
+      "loss": 1.2876,
+      "step": 2872
+    },
+    {
+      "epoch": 0.2521620116788401,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0027729550063558843,
+      "loss": 1.2204,
+      "step": 2873
+    },
+    {
+      "epoch": 0.25224978126174263,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0027727195468861487,
+      "loss": 1.3083,
+      "step": 2874
+    },
+    {
+      "epoch": 0.2523375508446451,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.002772483976608436,
+      "loss": 1.3388,
+      "step": 2875
+    },
+    {
+      "epoch": 0.25242532042754756,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.0027722482955459967,
+      "loss": 1.3027,
+      "step": 2876
+    },
+    {
+      "epoch": 0.2525130900104501,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0027720125037220904,
+      "loss": 1.3228,
+      "step": 2877
+    },
+    {
+      "epoch": 0.25260085959335254,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0027717766011599904,
+      "loss": 1.3671,
+      "step": 2878
+    },
+    {
+      "epoch": 0.25268862917625506,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.002771540587882978,
+      "loss": 1.3272,
+      "step": 2879
+    },
+    {
+      "epoch": 0.25277639875915753,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002771304463914347,
+      "loss": 1.2853,
+      "step": 2880
+    },
+    {
+      "epoch": 0.25286416834206,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0027710682292774024,
+      "loss": 1.3132,
+      "step": 2881
+    },
+    {
+      "epoch": 0.2529519379249625,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.00277083188399546,
+      "loss": 1.3118,
+      "step": 2882
+    },
+    {
+      "epoch": 0.253039707507865,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.002770595428091845,
+      "loss": 1.2235,
+      "step": 2883
+    },
+    {
+      "epoch": 0.25312747709076744,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0027703588615898955,
+      "loss": 1.2472,
+      "step": 2884
+    },
+    {
+      "epoch": 0.25321524667366996,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.002770122184512959,
+      "loss": 1.2688,
+      "step": 2885
+    },
+    {
+      "epoch": 0.2533030162565724,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0027698853968843953,
+      "loss": 1.3174,
+      "step": 2886
+    },
+    {
+      "epoch": 0.25339078583947494,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0027696484987275736,
+      "loss": 1.321,
+      "step": 2887
+    },
+    {
+      "epoch": 0.2534785554223774,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.0027694114900658757,
+      "loss": 1.2987,
+      "step": 2888
+    },
+    {
+      "epoch": 0.25356632500527987,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.0027691743709226933,
+      "loss": 1.2478,
+      "step": 2889
+    },
+    {
+      "epoch": 0.2536540945881824,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.0027689371413214285,
+      "loss": 1.2137,
+      "step": 2890
+    },
+    {
+      "epoch": 0.25374186417108485,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.0027686998012854953,
+      "loss": 1.3501,
+      "step": 2891
+    },
+    {
+      "epoch": 0.25382963375398737,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0027684623508383186,
+      "loss": 1.2621,
+      "step": 2892
+    },
+    {
+      "epoch": 0.25391740333688984,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.002768224790003333,
+      "loss": 1.3532,
+      "step": 2893
+    },
+    {
+      "epoch": 0.2540051729197923,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.0027679871188039853,
+      "loss": 1.2514,
+      "step": 2894
+    },
+    {
+      "epoch": 0.2540929425026948,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.0027677493372637325,
+      "loss": 1.2835,
+      "step": 2895
+    },
+    {
+      "epoch": 0.2541807120855973,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.0027675114454060426,
+      "loss": 1.2727,
+      "step": 2896
+    },
+    {
+      "epoch": 0.25426848166849975,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.002767273443254395,
+      "loss": 1.2444,
+      "step": 2897
+    },
+    {
+      "epoch": 0.25435625125140227,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0027670353308322797,
+      "loss": 1.3285,
+      "step": 2898
+    },
+    {
+      "epoch": 0.25444402083430473,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0027667971081631972,
+      "loss": 1.21,
+      "step": 2899
+    },
+    {
+      "epoch": 0.25453179041720725,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002766558775270658,
+      "loss": 1.2441,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2546195600001097,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.002766320332178187,
+      "loss": 1.2806,
+      "step": 2901
+    },
+    {
+      "epoch": 0.2547073295830122,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0027660817789093156,
+      "loss": 1.276,
+      "step": 2902
+    },
+    {
+      "epoch": 0.2547950991659147,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.002765843115487589,
+      "loss": 1.2851,
+      "step": 2903
+    },
+    {
+      "epoch": 0.25488286874881716,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002765604341936562,
+      "loss": 1.3001,
+      "step": 2904
+    },
+    {
+      "epoch": 0.2549706383317197,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0027653654582798005,
+      "loss": 1.2329,
+      "step": 2905
+    },
+    {
+      "epoch": 0.25505840791462214,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0027651264645408816,
+      "loss": 1.2928,
+      "step": 2906
+    },
+    {
+      "epoch": 0.2551461774975246,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.002764887360743393,
+      "loss": 1.3153,
+      "step": 2907
+    },
+    {
+      "epoch": 0.2552339470804271,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0027646481469109333,
+      "loss": 1.2155,
+      "step": 2908
+    },
+    {
+      "epoch": 0.2553217166633296,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002764408823067112,
+      "loss": 1.3383,
+      "step": 2909
+    },
+    {
+      "epoch": 0.25540948624623205,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0027641693892355495,
+      "loss": 1.3011,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2554972558291346,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0027639298454398766,
+      "loss": 1.2554,
+      "step": 2911
+    },
+    {
+      "epoch": 0.25558502541203704,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.002763690191703736,
+      "loss": 1.2817,
+      "step": 2912
+    },
+    {
+      "epoch": 0.25567279499493956,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0027634504280507795,
+      "loss": 1.2574,
+      "step": 2913
+    },
+    {
+      "epoch": 0.255760564577842,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0027632105545046712,
+      "loss": 1.3318,
+      "step": 2914
+    },
+    {
+      "epoch": 0.2558483341607445,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.0027629705710890866,
+      "loss": 1.3059,
+      "step": 2915
+    },
+    {
+      "epoch": 0.255936103743647,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.0027627304778277107,
+      "loss": 1.3332,
+      "step": 2916
+    },
+    {
+      "epoch": 0.25602387332654947,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.002762490274744239,
+      "loss": 1.2467,
+      "step": 2917
+    },
+    {
+      "epoch": 0.256111642909452,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 0.002762249961862379,
+      "loss": 1.2977,
+      "step": 2918
+    },
+    {
+      "epoch": 0.25619941249235445,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002762009539205849,
+      "loss": 1.2804,
+      "step": 2919
+    },
+    {
+      "epoch": 0.2562871820752569,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.002761769006798377,
+      "loss": 1.2428,
+      "step": 2920
+    },
+    {
+      "epoch": 0.25637495165815943,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0027615283646637034,
+      "loss": 1.2527,
+      "step": 2921
+    },
+    {
+      "epoch": 0.2564627212410619,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0027612876128255783,
+      "loss": 1.2966,
+      "step": 2922
+    },
+    {
+      "epoch": 0.25655049082396436,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.002761046751307763,
+      "loss": 1.2712,
+      "step": 2923
+    },
+    {
+      "epoch": 0.2566382604068669,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0027608057801340297,
+      "loss": 1.2714,
+      "step": 2924
+    },
+    {
+      "epoch": 0.25672602998976934,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.002760564699328161,
+      "loss": 1.3367,
+      "step": 2925
+    },
+    {
+      "epoch": 0.25681379957267186,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0027603235089139503,
+      "loss": 1.2798,
+      "step": 2926
+    },
+    {
+      "epoch": 0.2569015691555743,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0027600822089152034,
+      "loss": 1.3074,
+      "step": 2927
+    },
+    {
+      "epoch": 0.2569893387384768,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.0027598407993557343,
+      "loss": 1.267,
+      "step": 2928
+    },
+    {
+      "epoch": 0.2570771083213793,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0027595992802593697,
+      "loss": 1.2576,
+      "step": 2929
+    },
+    {
+      "epoch": 0.2571648779042818,
+      "grad_norm": 0.0634765625,
+      "learning_rate": 0.0027593576516499467,
+      "loss": 1.2105,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2572526474871843,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0027591159135513135,
+      "loss": 1.245,
+      "step": 2931
+    },
+    {
+      "epoch": 0.25734041707008676,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.0027588740659873277,
+      "loss": 1.2673,
+      "step": 2932
+    },
+    {
+      "epoch": 0.2574281866529892,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002758632108981859,
+      "loss": 1.2934,
+      "step": 2933
+    },
+    {
+      "epoch": 0.25751595623589174,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.002758390042558788,
+      "loss": 1.2124,
+      "step": 2934
+    },
+    {
+      "epoch": 0.2576037258187942,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0027581478667420053,
+      "loss": 1.2355,
+      "step": 2935
+    },
+    {
+      "epoch": 0.25769149540169667,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0027579055815554134,
+      "loss": 1.2205,
+      "step": 2936
+    },
+    {
+      "epoch": 0.2577792649845992,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0027576631870229237,
+      "loss": 1.2628,
+      "step": 2937
+    },
+    {
+      "epoch": 0.25786703456750165,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0027574206831684606,
+      "loss": 1.2896,
+      "step": 2938
+    },
+    {
+      "epoch": 0.25795480415040417,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.002757178070015958,
+      "loss": 1.2801,
+      "step": 2939
+    },
+    {
+      "epoch": 0.25804257373330663,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.002756935347589361,
+      "loss": 1.2961,
+      "step": 2940
+    },
+    {
+      "epoch": 0.2581303433162091,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.0027566925159126245,
+      "loss": 1.2878,
+      "step": 2941
+    },
+    {
+      "epoch": 0.2582181128991116,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.002756449575009716,
+      "loss": 1.2064,
+      "step": 2942
+    },
+    {
+      "epoch": 0.2583058824820141,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.0027562065249046126,
+      "loss": 1.2946,
+      "step": 2943
+    },
+    {
+      "epoch": 0.2583936520649166,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.002755963365621301,
+      "loss": 1.3275,
+      "step": 2944
+    },
+    {
+      "epoch": 0.25848142164781907,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0027557200971837833,
+      "loss": 1.3336,
+      "step": 2945
+    },
+    {
+      "epoch": 0.25856919123072153,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0027554767196160657,
+      "loss": 1.332,
+      "step": 2946
+    },
+    {
+      "epoch": 0.25865696081362405,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0027552332329421705,
+      "loss": 1.2901,
+      "step": 2947
+    },
+    {
+      "epoch": 0.2587447303965265,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.002754989637186128,
+      "loss": 1.3124,
+      "step": 2948
+    },
+    {
+      "epoch": 0.25883249997942903,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0027547459323719816,
+      "loss": 1.3086,
+      "step": 2949
+    },
+    {
+      "epoch": 0.2589202695623315,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0027545021185237813,
+      "loss": 1.3149,
+      "step": 2950
+    },
+    {
+      "epoch": 0.25900803914523396,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0027542581956655934,
+      "loss": 1.2337,
+      "step": 2951
+    },
+    {
+      "epoch": 0.2590958087281365,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.002754014163821491,
+      "loss": 1.2668,
+      "step": 2952
+    },
+    {
+      "epoch": 0.25918357831103894,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0027537700230155586,
+      "loss": 1.3238,
+      "step": 2953
+    },
+    {
+      "epoch": 0.2592713478939414,
+      "grad_norm": 0.060791015625,
+      "learning_rate": 0.002753525773271892,
+      "loss": 1.299,
+      "step": 2954
+    },
+    {
+      "epoch": 0.2593591174768439,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0027532814146145985,
+      "loss": 1.2783,
+      "step": 2955
+    },
+    {
+      "epoch": 0.2594468870597464,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0027530369470677944,
+      "loss": 1.299,
+      "step": 2956
+    },
+    {
+      "epoch": 0.2595346566426489,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.002752792370655608,
+      "loss": 1.306,
+      "step": 2957
+    },
+    {
+      "epoch": 0.2596224262255514,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0027525476854021787,
+      "loss": 1.2516,
+      "step": 2958
+    },
+    {
+      "epoch": 0.25971019580845384,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0027523028913316555,
+      "loss": 1.2497,
+      "step": 2959
+    },
+    {
+      "epoch": 0.25979796539135636,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.002752057988468199,
+      "loss": 1.2862,
+      "step": 2960
+    },
+    {
+      "epoch": 0.2598857349742588,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.002751812976835978,
+      "loss": 1.2235,
+      "step": 2961
+    },
+    {
+      "epoch": 0.25997350455716134,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.002751567856459177,
+      "loss": 1.2284,
+      "step": 2962
+    },
+    {
+      "epoch": 0.2600612741400638,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.002751322627361987,
+      "loss": 1.2561,
+      "step": 2963
+    },
+    {
+      "epoch": 0.26014904372296627,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0027510772895686115,
+      "loss": 1.2054,
+      "step": 2964
+    },
+    {
+      "epoch": 0.2602368133058688,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.002750831843103264,
+      "loss": 1.2635,
+      "step": 2965
+    },
+    {
+      "epoch": 0.26032458288877125,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.00275058628799017,
+      "loss": 1.2572,
+      "step": 2966
+    },
+    {
+      "epoch": 0.2604123524716737,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0027503406242535638,
+      "loss": 1.2678,
+      "step": 2967
+    },
+    {
+      "epoch": 0.26050012205457623,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0027500948519176916,
+      "loss": 1.3122,
+      "step": 2968
+    },
+    {
+      "epoch": 0.2605878916374787,
+      "grad_norm": 0.061279296875,
+      "learning_rate": 0.00274984897100681,
+      "loss": 1.3071,
+      "step": 2969
+    },
+    {
+      "epoch": 0.2606756612203812,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002749602981545188,
+      "loss": 1.3317,
+      "step": 2970
+    },
+    {
+      "epoch": 0.2607634308032837,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0027493568835571017,
+      "loss": 1.2548,
+      "step": 2971
+    },
+    {
+      "epoch": 0.26085120038618614,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.002749110677066841,
+      "loss": 1.2289,
+      "step": 2972
+    },
+    {
+      "epoch": 0.26093896996908866,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.002748864362098706,
+      "loss": 1.2861,
+      "step": 2973
+    },
+    {
+      "epoch": 0.2610267395519911,
+      "grad_norm": 0.068359375,
+      "learning_rate": 0.002748617938677006,
+      "loss": 1.3302,
+      "step": 2974
+    },
+    {
+      "epoch": 0.26111450913489365,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0027483714068260627,
+      "loss": 1.3258,
+      "step": 2975
+    },
+    {
+      "epoch": 0.2612022787177961,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0027481247665702073,
+      "loss": 1.2715,
+      "step": 2976
+    },
+    {
+      "epoch": 0.2612900483006986,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0027478780179337825,
+      "loss": 1.2598,
+      "step": 2977
+    },
+    {
+      "epoch": 0.2613778178836011,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.002747631160941142,
+      "loss": 1.3123,
+      "step": 2978
+    },
+    {
+      "epoch": 0.26146558746650356,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0027473841956166486,
+      "loss": 1.2426,
+      "step": 2979
+    },
+    {
+      "epoch": 0.261553357049406,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002747137121984677,
+      "loss": 1.2666,
+      "step": 2980
+    },
+    {
+      "epoch": 0.26164112663230854,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.0027468899400696126,
+      "loss": 1.2684,
+      "step": 2981
+    },
+    {
+      "epoch": 0.261728896215211,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.002746642649895851,
+      "loss": 1.2418,
+      "step": 2982
+    },
+    {
+      "epoch": 0.2618166657981135,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.0027463952514878,
+      "loss": 1.229,
+      "step": 2983
+    },
+    {
+      "epoch": 0.261904435381016,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.002746147744869876,
+      "loss": 1.3024,
+      "step": 2984
+    },
+    {
+      "epoch": 0.26199220496391845,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.002745900130066506,
+      "loss": 1.2839,
+      "step": 2985
+    },
+    {
+      "epoch": 0.26207997454682097,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0027456524071021293,
+      "loss": 1.2519,
+      "step": 2986
+    },
+    {
+      "epoch": 0.26216774412972343,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.0027454045760011963,
+      "loss": 1.2891,
+      "step": 2987
+    },
+    {
+      "epoch": 0.26225551371262595,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002745156636788165,
+      "loss": 1.3284,
+      "step": 2988
+    },
+    {
+      "epoch": 0.2623432832955284,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0027449085894875073,
+      "loss": 1.1907,
+      "step": 2989
+    },
+    {
+      "epoch": 0.2624310528784309,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0027446604341237043,
+      "loss": 1.2706,
+      "step": 2990
+    },
+    {
+      "epoch": 0.2625188224613334,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.0027444121707212474,
+      "loss": 1.2473,
+      "step": 2991
+    },
+    {
+      "epoch": 0.26260659204423586,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0027441637993046403,
+      "loss": 1.2635,
+      "step": 2992
+    },
+    {
+      "epoch": 0.26269436162713833,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.002743915319898395,
+      "loss": 1.33,
+      "step": 2993
+    },
+    {
+      "epoch": 0.26278213121004085,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.002743666732527036,
+      "loss": 1.237,
+      "step": 2994
+    },
+    {
+      "epoch": 0.2628699007929433,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.0027434180372150983,
+      "loss": 1.261,
+      "step": 2995
+    },
+    {
+      "epoch": 0.26295767037584583,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.0027431692339871267,
+      "loss": 1.2457,
+      "step": 2996
+    },
+    {
+      "epoch": 0.2630454399587483,
+      "grad_norm": 0.06201171875,
+      "learning_rate": 0.0027429203228676773,
+      "loss": 1.2615,
+      "step": 2997
+    },
+    {
+      "epoch": 0.26313320954165076,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0027426713038813163,
+      "loss": 1.2716,
+      "step": 2998
+    },
+    {
+      "epoch": 0.2632209791245533,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002742422177052621,
+      "loss": 1.3092,
+      "step": 2999
+    },
+    {
+      "epoch": 0.26330874870745574,
+      "grad_norm": 0.061767578125,
+      "learning_rate": 0.002742172942406179,
+      "loss": 1.3333,
+      "step": 3000
+    },
+    {
+      "epoch": 0.26330874870745574,
+      "eval_loss": 1.2623319625854492,
+      "eval_runtime": 437.7402,
+      "eval_samples_per_second": 33.673,
+      "eval_steps_per_second": 8.418,
+      "step": 3000
+    },
+    {
+      "epoch": 0.26339651829035826,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.0027419235999665897,
+      "loss": 1.3216,
+      "step": 3001
+    },
+    {
+      "epoch": 0.2634842878732607,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.002741674149758461,
+      "loss": 1.2587,
+      "step": 3002
+    },
+    {
+      "epoch": 0.2635720574561632,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.002741424591806414,
+      "loss": 1.2656,
+      "step": 3003
+    },
+    {
+      "epoch": 0.2636598270390657,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.002741174926135078,
+      "loss": 1.2652,
+      "step": 3004
+    },
+    {
+      "epoch": 0.26374759662196817,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002740925152769094,
+      "loss": 1.2813,
+      "step": 3005
+    },
+    {
+      "epoch": 0.26383536620487064,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0027406752717331145,
+      "loss": 1.1798,
+      "step": 3006
+    },
+    {
+      "epoch": 0.26392313578777316,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002740425283051801,
+      "loss": 1.1798,
+      "step": 3007
+    },
+    {
+      "epoch": 0.2640109053706756,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0027401751867498266,
+      "loss": 1.3219,
+      "step": 3008
+    },
+    {
+      "epoch": 0.26409867495357814,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0027399249828518753,
+      "loss": 1.2289,
+      "step": 3009
+    },
+    {
+      "epoch": 0.2641864445364806,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.00273967467138264,
+      "loss": 1.2492,
+      "step": 3010
+    },
+    {
+      "epoch": 0.26427421411938307,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.002739424252366827,
+      "loss": 1.2659,
+      "step": 3011
+    },
+    {
+      "epoch": 0.2643619837022856,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0027391737258291513,
+      "loss": 1.2757,
+      "step": 3012
+    },
+    {
+      "epoch": 0.26444975328518805,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0027389230917943376,
+      "loss": 1.2943,
+      "step": 3013
+    },
+    {
+      "epoch": 0.26453752286809057,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0027386723502871237,
+      "loss": 1.2918,
+      "step": 3014
+    },
+    {
+      "epoch": 0.26462529245099303,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.0027384215013322565,
+      "loss": 1.3467,
+      "step": 3015
+    },
+    {
+      "epoch": 0.2647130620338955,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0027381705449544937,
+      "loss": 1.1855,
+      "step": 3016
+    },
+    {
+      "epoch": 0.264800831616798,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.002737919481178604,
+      "loss": 1.2759,
+      "step": 3017
+    },
+    {
+      "epoch": 0.2648886011997005,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.002737668310029366,
+      "loss": 1.3151,
+      "step": 3018
+    },
+    {
+      "epoch": 0.264976370782603,
+      "grad_norm": 0.061279296875,
+      "learning_rate": 0.0027374170315315697,
+      "loss": 1.2375,
+      "step": 3019
+    },
+    {
+      "epoch": 0.26506414036550546,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0027371656457100147,
+      "loss": 1.3008,
+      "step": 3020
+    },
+    {
+      "epoch": 0.2651519099484079,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.0027369141525895122,
+      "loss": 1.2934,
+      "step": 3021
+    },
+    {
+      "epoch": 0.26523967953131045,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.0027366625521948844,
+      "loss": 1.2363,
+      "step": 3022
+    },
+    {
+      "epoch": 0.2653274491142129,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.002736410844550961,
+      "loss": 1.2215,
+      "step": 3023
+    },
+    {
+      "epoch": 0.2654152186971154,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0027361590296825873,
+      "loss": 1.2465,
+      "step": 3024
+    },
+    {
+      "epoch": 0.2655029882800179,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.002735907107614614,
+      "loss": 1.2531,
+      "step": 3025
+    },
+    {
+      "epoch": 0.26559075786292036,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0027356550783719067,
+      "loss": 1.2992,
+      "step": 3026
+    },
+    {
+      "epoch": 0.2656785274458229,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0027354029419793388,
+      "loss": 1.2623,
+      "step": 3027
+    },
+    {
+      "epoch": 0.26576629702872534,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0027351506984617945,
+      "loss": 1.2807,
+      "step": 3028
+    },
+    {
+      "epoch": 0.2658540666116278,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.0027348983478441705,
+      "loss": 1.2909,
+      "step": 3029
+    },
+    {
+      "epoch": 0.2659418361945303,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0027346458901513727,
+      "loss": 1.2481,
+      "step": 3030
+    },
+    {
+      "epoch": 0.2660296057774328,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0027343933254083165,
+      "loss": 1.3352,
+      "step": 3031
+    },
+    {
+      "epoch": 0.2661173753603353,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0027341406536399303,
+      "loss": 1.3,
+      "step": 3032
+    },
+    {
+      "epoch": 0.26620514494323777,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.002733887874871151,
+      "loss": 1.2858,
+      "step": 3033
+    },
+    {
+      "epoch": 0.26629291452614023,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.0027336349891269275,
+      "loss": 1.2806,
+      "step": 3034
+    },
+    {
+      "epoch": 0.26638068410904275,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.002733381996432218,
+      "loss": 1.3209,
+      "step": 3035
+    },
+    {
+      "epoch": 0.2664684536919452,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0027331288968119928,
+      "loss": 1.2558,
+      "step": 3036
+    },
+    {
+      "epoch": 0.2665562232748477,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0027328756902912305,
+      "loss": 1.2399,
+      "step": 3037
+    },
+    {
+      "epoch": 0.2666439928577502,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.002732622376894922,
+      "loss": 1.2427,
+      "step": 3038
+    },
+    {
+      "epoch": 0.26673176244065266,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0027323689566480696,
+      "loss": 1.2434,
+      "step": 3039
+    },
+    {
+      "epoch": 0.2668195320235552,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0027321154295756834,
+      "loss": 1.2715,
+      "step": 3040
+    },
+    {
+      "epoch": 0.26690730160645765,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0027318617957027857,
+      "loss": 1.2983,
+      "step": 3041
+    },
+    {
+      "epoch": 0.2669950711893601,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00273160805505441,
+      "loss": 1.306,
+      "step": 3042
+    },
+    {
+      "epoch": 0.26708284077226263,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.002731354207655599,
+      "loss": 1.2572,
+      "step": 3043
+    },
+    {
+      "epoch": 0.2671706103551651,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002731100253531406,
+      "loss": 1.3614,
+      "step": 3044
+    },
+    {
+      "epoch": 0.2672583799380676,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.0027308461927068963,
+      "loss": 1.2965,
+      "step": 3045
+    },
+    {
+      "epoch": 0.2673461495209701,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.0027305920252071437,
+      "loss": 1.2385,
+      "step": 3046
+    },
+    {
+      "epoch": 0.26743391910387254,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.0027303377510572335,
+      "loss": 1.3075,
+      "step": 3047
+    },
+    {
+      "epoch": 0.26752168868677506,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0027300833702822627,
+      "loss": 1.299,
+      "step": 3048
+    },
+    {
+      "epoch": 0.2676094582696775,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0027298288829073368,
+      "loss": 1.2657,
+      "step": 3049
+    },
+    {
+      "epoch": 0.26769722785258,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0027295742889575725,
+      "loss": 1.2794,
+      "step": 3050
+    },
+    {
+      "epoch": 0.2677849974354825,
+      "grad_norm": 0.0595703125,
+      "learning_rate": 0.0027293195884580974,
+      "loss": 1.2349,
+      "step": 3051
+    },
+    {
+      "epoch": 0.26787276701838497,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00272906478143405,
+      "loss": 1.288,
+      "step": 3052
+    },
+    {
+      "epoch": 0.2679605366012875,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.002728809867910578,
+      "loss": 1.2928,
+      "step": 3053
+    },
+    {
+      "epoch": 0.26804830618418996,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002728554847912841,
+      "loss": 1.2692,
+      "step": 3054
+    },
+    {
+      "epoch": 0.2681360757670924,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0027282997214660076,
+      "loss": 1.2758,
+      "step": 3055
+    },
+    {
+      "epoch": 0.26822384534999494,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0027280444885952587,
+      "loss": 1.303,
+      "step": 3056
+    },
+    {
+      "epoch": 0.2683116149328974,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0027277891493257844,
+      "loss": 1.2385,
+      "step": 3057
+    },
+    {
+      "epoch": 0.2683993845157999,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0027275337036827855,
+      "loss": 1.2298,
+      "step": 3058
+    },
+    {
+      "epoch": 0.2684871540987024,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0027272781516914733,
+      "loss": 1.3027,
+      "step": 3059
+    },
+    {
+      "epoch": 0.26857492368160485,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.002727022493377071,
+      "loss": 1.3254,
+      "step": 3060
+    },
+    {
+      "epoch": 0.26866269326450737,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0027267667287648096,
+      "loss": 1.2489,
+      "step": 3061
+    },
+    {
+      "epoch": 0.26875046284740983,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0027265108578799326,
+      "loss": 1.2781,
+      "step": 3062
+    },
+    {
+      "epoch": 0.2688382324303123,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.002726254880747694,
+      "loss": 1.2501,
+      "step": 3063
+    },
+    {
+      "epoch": 0.2689260020132148,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0027259987973933565,
+      "loss": 1.2445,
+      "step": 3064
+    },
+    {
+      "epoch": 0.2690137715961173,
+      "grad_norm": 0.068359375,
+      "learning_rate": 0.0027257426078421956,
+      "loss": 1.3197,
+      "step": 3065
+    },
+    {
+      "epoch": 0.2691015411790198,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0027254863121194957,
+      "loss": 1.2904,
+      "step": 3066
+    },
+    {
+      "epoch": 0.26918931076192226,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.0027252299102505525,
+      "loss": 1.2786,
+      "step": 3067
+    },
+    {
+      "epoch": 0.2692770803448247,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002724973402260671,
+      "loss": 1.2729,
+      "step": 3068
+    },
+    {
+      "epoch": 0.26936484992772725,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.0027247167881751687,
+      "loss": 1.2518,
+      "step": 3069
+    },
+    {
+      "epoch": 0.2694526195106297,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.002724460068019371,
+      "loss": 1.3055,
+      "step": 3070
+    },
+    {
+      "epoch": 0.26954038909353223,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0027242032418186173,
+      "loss": 1.2591,
+      "step": 3071
+    },
+    {
+      "epoch": 0.2696281586764347,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0027239463095982534,
+      "loss": 1.2607,
+      "step": 3072
+    },
+    {
+      "epoch": 0.26971592825933716,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.002723689271383638,
+      "loss": 1.3159,
+      "step": 3073
+    },
+    {
+      "epoch": 0.2698036978422397,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.00272343212720014,
+      "loss": 1.2002,
+      "step": 3074
+    },
+    {
+      "epoch": 0.26989146742514214,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0027231748770731386,
+      "loss": 1.316,
+      "step": 3075
+    },
+    {
+      "epoch": 0.2699792370080446,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.002722917521028023,
+      "loss": 1.2967,
+      "step": 3076
+    },
+    {
+      "epoch": 0.2700670065909471,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0027226600590901933,
+      "loss": 1.3347,
+      "step": 3077
+    },
+    {
+      "epoch": 0.2701547761738496,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0027224024912850604,
+      "loss": 1.3171,
+      "step": 3078
+    },
+    {
+      "epoch": 0.2702425457567521,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0027221448176380446,
+      "loss": 1.299,
+      "step": 3079
+    },
+    {
+      "epoch": 0.27033031533965457,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.0027218870381745777,
+      "loss": 1.2303,
+      "step": 3080
+    },
+    {
+      "epoch": 0.27041808492255703,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.002721629152920101,
+      "loss": 1.332,
+      "step": 3081
+    },
+    {
+      "epoch": 0.27050585450545955,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0027213711619000678,
+      "loss": 1.221,
+      "step": 3082
+    },
+    {
+      "epoch": 0.270593624088362,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0027211130651399397,
+      "loss": 1.2766,
+      "step": 3083
+    },
+    {
+      "epoch": 0.27068139367126454,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0027208548626651906,
+      "loss": 1.3172,
+      "step": 3084
+    },
+    {
+      "epoch": 0.270769163254167,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.002720596554501303,
+      "loss": 1.2946,
+      "step": 3085
+    },
+    {
+      "epoch": 0.27085693283706946,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0027203381406737726,
+      "loss": 1.2846,
+      "step": 3086
+    },
+    {
+      "epoch": 0.270944702419972,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.002720079621208102,
+      "loss": 1.3227,
+      "step": 3087
+    },
+    {
+      "epoch": 0.27103247200287445,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0027198209961298075,
+      "loss": 1.2751,
+      "step": 3088
+    },
+    {
+      "epoch": 0.27112024158577697,
+      "grad_norm": 0.06201171875,
+      "learning_rate": 0.002719562265464413,
+      "loss": 1.2337,
+      "step": 3089
+    },
+    {
+      "epoch": 0.27120801116867943,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0027193034292374556,
+      "loss": 1.3,
+      "step": 3090
+    },
+    {
+      "epoch": 0.2712957807515819,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0027190444874744804,
+      "loss": 1.2456,
+      "step": 3091
+    },
+    {
+      "epoch": 0.2713835503344844,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.002718785440201044,
+      "loss": 1.2532,
+      "step": 3092
+    },
+    {
+      "epoch": 0.2714713199173869,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.002718526287442715,
+      "loss": 1.249,
+      "step": 3093
+    },
+    {
+      "epoch": 0.27155908950028934,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.002718267029225068,
+      "loss": 1.2369,
+      "step": 3094
+    },
+    {
+      "epoch": 0.27164685908319186,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0027180076655736927,
+      "loss": 1.3229,
+      "step": 3095
+    },
+    {
+      "epoch": 0.2717346286660943,
+      "grad_norm": 0.060302734375,
+      "learning_rate": 0.0027177481965141863,
+      "loss": 1.2477,
+      "step": 3096
+    },
+    {
+      "epoch": 0.27182239824899684,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002717488622072158,
+      "loss": 1.2562,
+      "step": 3097
+    },
+    {
+      "epoch": 0.2719101678318993,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.0027172289422732264,
+      "loss": 1.2525,
+      "step": 3098
+    },
+    {
+      "epoch": 0.27199793741480177,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0027169691571430213,
+      "loss": 1.3141,
+      "step": 3099
+    },
+    {
+      "epoch": 0.2720857069977043,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.0027167092667071823,
+      "loss": 1.2962,
+      "step": 3100
+    },
+    {
+      "epoch": 0.27217347658060675,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0027164492709913592,
+      "loss": 1.2762,
+      "step": 3101
+    },
+    {
+      "epoch": 0.2722612461635093,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.0027161891700212123,
+      "loss": 1.243,
+      "step": 3102
+    },
+    {
+      "epoch": 0.27234901574641174,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0027159289638224136,
+      "loss": 1.2123,
+      "step": 3103
+    },
+    {
+      "epoch": 0.2724367853293142,
+      "grad_norm": 0.06787109375,
+      "learning_rate": 0.002715668652420644,
+      "loss": 1.3245,
+      "step": 3104
+    },
+    {
+      "epoch": 0.2725245549122167,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0027154082358415944,
+      "loss": 1.2756,
+      "step": 3105
+    },
+    {
+      "epoch": 0.2726123244951192,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.002715147714110968,
+      "loss": 1.3757,
+      "step": 3106
+    },
+    {
+      "epoch": 0.27270009407802165,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0027148870872544767,
+      "loss": 1.2723,
+      "step": 3107
+    },
+    {
+      "epoch": 0.27278786366092417,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.002714626355297843,
+      "loss": 1.2623,
+      "step": 3108
+    },
+    {
+      "epoch": 0.27287563324382663,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0027143655182668013,
+      "loss": 1.3115,
+      "step": 3109
+    },
+    {
+      "epoch": 0.27296340282672915,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0027141045761870943,
+      "loss": 1.2493,
+      "step": 3110
+    },
+    {
+      "epoch": 0.2730511724096316,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0027138435290844755,
+      "loss": 1.2701,
+      "step": 3111
+    },
+    {
+      "epoch": 0.2731389419925341,
+      "grad_norm": 0.060302734375,
+      "learning_rate": 0.0027135823769847103,
+      "loss": 1.274,
+      "step": 3112
+    },
+    {
+      "epoch": 0.2732267115754366,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.002713321119913573,
+      "loss": 1.2382,
+      "step": 3113
+    },
+    {
+      "epoch": 0.27331448115833906,
+      "grad_norm": 0.0634765625,
+      "learning_rate": 0.002713059757896848,
+      "loss": 1.3094,
+      "step": 3114
+    },
+    {
+      "epoch": 0.2734022507412416,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0027127982909603316,
+      "loss": 1.2698,
+      "step": 3115
+    },
+    {
+      "epoch": 0.27349002032414405,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.002712536719129829,
+      "loss": 1.2649,
+      "step": 3116
+    },
+    {
+      "epoch": 0.2735777899070465,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0027122750424311563,
+      "loss": 1.2478,
+      "step": 3117
+    },
+    {
+      "epoch": 0.27366555948994903,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0027120132608901403,
+      "loss": 1.2676,
+      "step": 3118
+    },
+    {
+      "epoch": 0.2737533290728515,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.0027117513745326177,
+      "loss": 1.2763,
+      "step": 3119
+    },
+    {
+      "epoch": 0.27384109865575396,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0027114893833844354,
+      "loss": 1.2725,
+      "step": 3120
+    },
+    {
+      "epoch": 0.2739288682386565,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0027112272874714512,
+      "loss": 1.2625,
+      "step": 3121
+    },
+    {
+      "epoch": 0.27401663782155894,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.002710965086819533,
+      "loss": 1.2014,
+      "step": 3122
+    },
+    {
+      "epoch": 0.27410440740446146,
+      "grad_norm": 0.0634765625,
+      "learning_rate": 0.0027107027814545588,
+      "loss": 1.2619,
+      "step": 3123
+    },
+    {
+      "epoch": 0.2741921769873639,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.002710440371402416,
+      "loss": 1.2834,
+      "step": 3124
+    },
+    {
+      "epoch": 0.2742799465702664,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.0027101778566890056,
+      "loss": 1.2103,
+      "step": 3125
+    },
+    {
+      "epoch": 0.2743677161531689,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0027099152373402354,
+      "loss": 1.2863,
+      "step": 3126
+    },
+    {
+      "epoch": 0.27445548573607137,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002709652513382025,
+      "loss": 1.2799,
+      "step": 3127
+    },
+    {
+      "epoch": 0.2745432553189739,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0027093896848403047,
+      "loss": 1.3309,
+      "step": 3128
+    },
+    {
+      "epoch": 0.27463102490187635,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0027091267517410137,
+      "loss": 1.2526,
+      "step": 3129
+    },
+    {
+      "epoch": 0.2747187944847788,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0027088637141101037,
+      "loss": 1.2656,
+      "step": 3130
+    },
+    {
+      "epoch": 0.27480656406768134,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0027086005719735348,
+      "loss": 1.2612,
+      "step": 3131
+    },
+    {
+      "epoch": 0.2748943336505838,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0027083373253572775,
+      "loss": 1.318,
+      "step": 3132
+    },
+    {
+      "epoch": 0.27498210323348626,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002708073974287314,
+      "loss": 1.2395,
+      "step": 3133
+    },
+    {
+      "epoch": 0.2750698728163888,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.002707810518789636,
+      "loss": 1.2935,
+      "step": 3134
+    },
+    {
+      "epoch": 0.27515764239929125,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.0027075469588902458,
+      "loss": 1.2433,
+      "step": 3135
+    },
+    {
+      "epoch": 0.27524541198219377,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.0027072832946151548,
+      "loss": 1.3055,
+      "step": 3136
+    },
+    {
+      "epoch": 0.27533318156509623,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.0027070195259903855,
+      "loss": 1.2839,
+      "step": 3137
+    },
+    {
+      "epoch": 0.2754209511479987,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0027067556530419724,
+      "loss": 1.2903,
+      "step": 3138
+    },
+    {
+      "epoch": 0.2755087207309012,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.002706491675795957,
+      "loss": 1.2727,
+      "step": 3139
+    },
+    {
+      "epoch": 0.2755964903138037,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 0.0027062275942783937,
+      "loss": 1.3098,
+      "step": 3140
+    },
+    {
+      "epoch": 0.2756842598967062,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.002705963408515347,
+      "loss": 1.2358,
+      "step": 3141
+    },
+    {
+      "epoch": 0.27577202947960866,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 0.002705699118532889,
+      "loss": 1.2854,
+      "step": 3142
+    },
+    {
+      "epoch": 0.2758597990625111,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0027054347243571056,
+      "loss": 1.3004,
+      "step": 3143
+    },
+    {
+      "epoch": 0.27594756864541364,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.002705170226014091,
+      "loss": 1.2557,
+      "step": 3144
+    },
+    {
+      "epoch": 0.2760353382283161,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.00270490562352995,
+      "loss": 1.343,
+      "step": 3145
+    },
+    {
+      "epoch": 0.27612310781121857,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.002704640916930798,
+      "loss": 1.332,
+      "step": 3146
+    },
+    {
+      "epoch": 0.2762108773941211,
+      "grad_norm": 0.057373046875,
+      "learning_rate": 0.0027043761062427608,
+      "loss": 1.2694,
+      "step": 3147
+    },
+    {
+      "epoch": 0.27629864697702355,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.002704111191491974,
+      "loss": 1.225,
+      "step": 3148
+    },
+    {
+      "epoch": 0.2763864165599261,
+      "grad_norm": 0.058349609375,
+      "learning_rate": 0.002703846172704583,
+      "loss": 1.2331,
+      "step": 3149
+    },
+    {
+      "epoch": 0.27647418614282854,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.002703581049906745,
+      "loss": 1.2785,
+      "step": 3150
+    },
+    {
+      "epoch": 0.276561955725731,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.002703315823124626,
+      "loss": 1.3095,
+      "step": 3151
+    },
+    {
+      "epoch": 0.2766497253086335,
+      "grad_norm": 0.06591796875,
+      "learning_rate": 0.002703050492384404,
+      "loss": 1.2248,
+      "step": 3152
+    },
+    {
+      "epoch": 0.276737494891536,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.002702785057712264,
+      "loss": 1.3222,
+      "step": 3153
+    },
+    {
+      "epoch": 0.2768252644744385,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.002702519519134404,
+      "loss": 1.3349,
+      "step": 3154
+    },
+    {
+      "epoch": 0.27691303405734097,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.0027022538766770334,
+      "loss": 1.2534,
+      "step": 3155
+    },
+    {
+      "epoch": 0.27700080364024343,
+      "grad_norm": 0.06787109375,
+      "learning_rate": 0.002701988130366368,
+      "loss": 1.2315,
+      "step": 3156
+    },
+    {
+      "epoch": 0.27708857322314595,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0027017222802286365,
+      "loss": 1.2524,
+      "step": 3157
+    },
+    {
+      "epoch": 0.2771763428060484,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.002701456326290078,
+      "loss": 1.2928,
+      "step": 3158
+    },
+    {
+      "epoch": 0.27726411238895093,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0027011902685769398,
+      "loss": 1.2641,
+      "step": 3159
+    },
+    {
+      "epoch": 0.2773518819718534,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002700924107115482,
+      "loss": 1.2903,
+      "step": 3160
+    },
+    {
+      "epoch": 0.27743965155475586,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.0027006578419319725,
+      "loss": 1.3015,
+      "step": 3161
+    },
+    {
+      "epoch": 0.2775274211376584,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0027003914730526915,
+      "loss": 1.272,
+      "step": 3162
+    },
+    {
+      "epoch": 0.27761519072056084,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0027001250005039288,
+      "loss": 1.2366,
+      "step": 3163
+    },
+    {
+      "epoch": 0.2777029603034633,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.002699858424311983,
+      "loss": 1.2801,
+      "step": 3164
+    },
+    {
+      "epoch": 0.27779072988636583,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.002699591744503165,
+      "loss": 1.2869,
+      "step": 3165
+    },
+    {
+      "epoch": 0.2778784994692683,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.0026993249611037954,
+      "loss": 1.2579,
+      "step": 3166
+    },
+    {
+      "epoch": 0.2779662690521708,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.002699058074140204,
+      "loss": 1.2847,
+      "step": 3167
+    },
+    {
+      "epoch": 0.2780540386350733,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.002698791083638731,
+      "loss": 1.2227,
+      "step": 3168
+    },
+    {
+      "epoch": 0.27814180821797574,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0026985239896257277,
+      "loss": 1.3268,
+      "step": 3169
+    },
+    {
+      "epoch": 0.27822957780087826,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.002698256792127556,
+      "loss": 1.2644,
+      "step": 3170
+    },
+    {
+      "epoch": 0.2783173473837807,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002697989491170587,
+      "loss": 1.2894,
+      "step": 3171
+    },
+    {
+      "epoch": 0.27840511696668324,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.002697722086781201,
+      "loss": 1.2656,
+      "step": 3172
+    },
+    {
+      "epoch": 0.2784928865495857,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0026974545789857913,
+      "loss": 1.2645,
+      "step": 3173
+    },
+    {
+      "epoch": 0.27858065613248817,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0026971869678107585,
+      "loss": 1.2479,
+      "step": 3174
+    },
+    {
+      "epoch": 0.2786684257153907,
+      "grad_norm": 0.05859375,
+      "learning_rate": 0.0026969192532825164,
+      "loss": 1.2105,
+      "step": 3175
+    },
+    {
+      "epoch": 0.27875619529829315,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002696651435427486,
+      "loss": 1.2568,
+      "step": 3176
+    },
+    {
+      "epoch": 0.2788439648811956,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0026963835142721007,
+      "loss": 1.2287,
+      "step": 3177
+    },
+    {
+      "epoch": 0.27893173446409814,
+      "grad_norm": 0.059326171875,
+      "learning_rate": 0.002696115489842803,
+      "loss": 1.2107,
+      "step": 3178
+    },
+    {
+      "epoch": 0.2790195040470006,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0026958473621660445,
+      "loss": 1.2313,
+      "step": 3179
+    },
+    {
+      "epoch": 0.2791072736299031,
+      "grad_norm": 0.058349609375,
+      "learning_rate": 0.0026955791312682905,
+      "loss": 1.228,
+      "step": 3180
+    },
+    {
+      "epoch": 0.2791950432128056,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0026953107971760134,
+      "loss": 1.2692,
+      "step": 3181
+    },
+    {
+      "epoch": 0.27928281279570805,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.0026950423599156966,
+      "loss": 1.3474,
+      "step": 3182
+    },
+    {
+      "epoch": 0.27937058237861057,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0026947738195138337,
+      "loss": 1.2893,
+      "step": 3183
+    },
+    {
+      "epoch": 0.27945835196151303,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.0026945051759969295,
+      "loss": 1.299,
+      "step": 3184
+    },
+    {
+      "epoch": 0.27954612154441555,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0026942364293914963,
+      "loss": 1.2903,
+      "step": 3185
+    },
+    {
+      "epoch": 0.279633891127318,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.00269396757972406,
+      "loss": 1.2591,
+      "step": 3186
+    },
+    {
+      "epoch": 0.2797216607102205,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.002693698627021154,
+      "loss": 1.2394,
+      "step": 3187
+    },
+    {
+      "epoch": 0.279809430293123,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.002693429571309324,
+      "loss": 1.2616,
+      "step": 3188
+    },
+    {
+      "epoch": 0.27989719987602546,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.002693160412615123,
+      "loss": 1.2913,
+      "step": 3189
+    },
+    {
+      "epoch": 0.2799849694589279,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002692891150965117,
+      "loss": 1.2628,
+      "step": 3190
+    },
+    {
+      "epoch": 0.28007273904183044,
+      "grad_norm": 0.058837890625,
+      "learning_rate": 0.0026926217863858814,
+      "loss": 1.2134,
+      "step": 3191
+    },
+    {
+      "epoch": 0.2801605086247329,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0026923523189040003,
+      "loss": 1.2175,
+      "step": 3192
+    },
+    {
+      "epoch": 0.2802482782076354,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.0026920827485460703,
+      "loss": 1.2723,
+      "step": 3193
+    },
+    {
+      "epoch": 0.2803360477905379,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0026918130753386957,
+      "loss": 1.3215,
+      "step": 3194
+    },
+    {
+      "epoch": 0.28042381737344035,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.0026915432993084936,
+      "loss": 1.1965,
+      "step": 3195
+    },
+    {
+      "epoch": 0.2805115869563429,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0026912734204820885,
+      "loss": 1.2367,
+      "step": 3196
+    },
+    {
+      "epoch": 0.28059935653924534,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.002691003438886117,
+      "loss": 1.2584,
+      "step": 3197
+    },
+    {
+      "epoch": 0.28068712612214786,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.002690733354547225,
+      "loss": 1.2969,
+      "step": 3198
+    },
+    {
+      "epoch": 0.2807748957050503,
+      "grad_norm": 0.060302734375,
+      "learning_rate": 0.0026904631674920694,
+      "loss": 1.2951,
+      "step": 3199
+    },
+    {
+      "epoch": 0.2808626652879528,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0026901928777473155,
+      "loss": 1.2894,
+      "step": 3200
+    },
+    {
+      "epoch": 0.2809504348708553,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.002689922485339641,
+      "loss": 1.2851,
+      "step": 3201
+    },
+    {
+      "epoch": 0.28103820445375777,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.002689651990295731,
+      "loss": 1.3346,
+      "step": 3202
+    },
+    {
+      "epoch": 0.28112597403666023,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.0026893813926422844,
+      "loss": 1.3104,
+      "step": 3203
+    },
+    {
+      "epoch": 0.28121374361956275,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0026891106924060064,
+      "loss": 1.2237,
+      "step": 3204
+    },
+    {
+      "epoch": 0.2813015132024652,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.0026888398896136143,
+      "loss": 1.2676,
+      "step": 3205
+    },
+    {
+      "epoch": 0.28138928278536773,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002688568984291836,
+      "loss": 1.2077,
+      "step": 3206
+    },
+    {
+      "epoch": 0.2814770523682702,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0026882979764674083,
+      "loss": 1.3397,
+      "step": 3207
+    },
+    {
+      "epoch": 0.28156482195117266,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0026880268661670786,
+      "loss": 1.2755,
+      "step": 3208
+    },
+    {
+      "epoch": 0.2816525915340752,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0026877556534176043,
+      "loss": 1.2845,
+      "step": 3209
+    },
+    {
+      "epoch": 0.28174036111697764,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0026874843382457534,
+      "loss": 1.3066,
+      "step": 3210
+    },
+    {
+      "epoch": 0.28182813069988016,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0026872129206783036,
+      "loss": 1.2582,
+      "step": 3211
+    },
+    {
+      "epoch": 0.2819159002827826,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0026869414007420424,
+      "loss": 1.2557,
+      "step": 3212
+    },
+    {
+      "epoch": 0.2820036698656851,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002686669778463768,
+      "loss": 1.3134,
+      "step": 3213
+    },
+    {
+      "epoch": 0.2820914394485876,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0026863980538702884,
+      "loss": 1.2484,
+      "step": 3214
+    },
+    {
+      "epoch": 0.2821792090314901,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.002686126226988421,
+      "loss": 1.2878,
+      "step": 3215
+    },
+    {
+      "epoch": 0.28226697861439254,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0026858542978449957,
+      "loss": 1.3048,
+      "step": 3216
+    },
+    {
+      "epoch": 0.28235474819729506,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0026855822664668496,
+      "loss": 1.2652,
+      "step": 3217
+    },
+    {
+      "epoch": 0.2824425177801975,
+      "grad_norm": 0.06787109375,
+      "learning_rate": 0.0026853101328808316,
+      "loss": 1.2367,
+      "step": 3218
+    },
+    {
+      "epoch": 0.28253028736310004,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0026850378971137994,
+      "loss": 1.2976,
+      "step": 3219
+    },
+    {
+      "epoch": 0.2826180569460025,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0026847655591926223,
+      "loss": 1.2665,
+      "step": 3220
+    },
+    {
+      "epoch": 0.28270582652890497,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.002684493119144179,
+      "loss": 1.2501,
+      "step": 3221
+    },
+    {
+      "epoch": 0.2827935961118075,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0026842205769953584,
+      "loss": 1.3064,
+      "step": 3222
+    },
+    {
+      "epoch": 0.28288136569470995,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.0026839479327730588,
+      "loss": 1.2508,
+      "step": 3223
+    },
+    {
+      "epoch": 0.28296913527761247,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.0026836751865041896,
+      "loss": 1.3441,
+      "step": 3224
+    },
+    {
+      "epoch": 0.28305690486051494,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.002683402338215669,
+      "loss": 1.2478,
+      "step": 3225
+    },
+    {
+      "epoch": 0.2831446744434174,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.002683129387934427,
+      "loss": 1.2065,
+      "step": 3226
+    },
+    {
+      "epoch": 0.2832324440263199,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.002682856335687402,
+      "loss": 1.2661,
+      "step": 3227
+    },
+    {
+      "epoch": 0.2833202136092224,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.002682583181501544,
+      "loss": 1.2668,
+      "step": 3228
+    },
+    {
+      "epoch": 0.2834079831921249,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002682309925403811,
+      "loss": 1.2687,
+      "step": 3229
+    },
+    {
+      "epoch": 0.28349575277502737,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.002682036567421174,
+      "loss": 1.2489,
+      "step": 3230
+    },
+    {
+      "epoch": 0.28358352235792983,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.002681763107580611,
+      "loss": 1.2499,
+      "step": 3231
+    },
+    {
+      "epoch": 0.28367129194083235,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.002681489545909111,
+      "loss": 1.2641,
+      "step": 3232
+    },
+    {
+      "epoch": 0.2837590615237348,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0026812158824336746,
+      "loss": 1.3063,
+      "step": 3233
+    },
+    {
+      "epoch": 0.2838468311066373,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0026809421171813115,
+      "loss": 1.318,
+      "step": 3234
+    },
+    {
+      "epoch": 0.2839346006895398,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.0026806682501790404,
+      "loss": 1.3138,
+      "step": 3235
+    },
+    {
+      "epoch": 0.28402237027244226,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0026803942814538906,
+      "loss": 1.3088,
+      "step": 3236
+    },
+    {
+      "epoch": 0.2841101398553448,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.0026801202110329028,
+      "loss": 1.2563,
+      "step": 3237
+    },
+    {
+      "epoch": 0.28419790943824724,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0026798460389431257,
+      "loss": 1.2921,
+      "step": 3238
+    },
+    {
+      "epoch": 0.2842856790211497,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.00267957176521162,
+      "loss": 1.2426,
+      "step": 3239
+    },
+    {
+      "epoch": 0.2843734486040522,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0026792973898654548,
+      "loss": 1.306,
+      "step": 3240
+    },
+    {
+      "epoch": 0.2844612181869547,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.00267902291293171,
+      "loss": 1.2449,
+      "step": 3241
+    },
+    {
+      "epoch": 0.2845489877698572,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0026787483344374746,
+      "loss": 1.2781,
+      "step": 3242
+    },
+    {
+      "epoch": 0.2846367573527597,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.00267847365440985,
+      "loss": 1.2482,
+      "step": 3243
+    },
+    {
+      "epoch": 0.28472452693566214,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002678198872875945,
+      "loss": 1.3168,
+      "step": 3244
+    },
+    {
+      "epoch": 0.28481229651856466,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0026779239898628796,
+      "loss": 1.2142,
+      "step": 3245
+    },
+    {
+      "epoch": 0.2849000661014671,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.002677649005397783,
+      "loss": 1.2748,
+      "step": 3246
+    },
+    {
+      "epoch": 0.2849878356843696,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0026773739195077965,
+      "loss": 1.2776,
+      "step": 3247
+    },
+    {
+      "epoch": 0.2850756052672721,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0026770987322200695,
+      "loss": 1.2221,
+      "step": 3248
+    },
+    {
+      "epoch": 0.28516337485017457,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.002676823443561761,
+      "loss": 1.2471,
+      "step": 3249
+    },
+    {
+      "epoch": 0.2852511444330771,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0026765480535600418,
+      "loss": 1.2988,
+      "step": 3250
+    },
+    {
+      "epoch": 0.28533891401597955,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.002676272562242092,
+      "loss": 1.2773,
+      "step": 3251
+    },
+    {
+      "epoch": 0.285426683598882,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0026759969696351004,
+      "loss": 1.2824,
+      "step": 3252
+    },
+    {
+      "epoch": 0.28551445318178453,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.0026757212757662677,
+      "loss": 1.2609,
+      "step": 3253
+    },
+    {
+      "epoch": 0.285602222764687,
+      "grad_norm": 0.068359375,
+      "learning_rate": 0.002675445480662804,
+      "loss": 1.2466,
+      "step": 3254
+    },
+    {
+      "epoch": 0.2856899923475895,
+      "grad_norm": 0.05712890625,
+      "learning_rate": 0.002675169584351928,
+      "loss": 1.2627,
+      "step": 3255
+    },
+    {
+      "epoch": 0.285777761930492,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.0026748935868608713,
+      "loss": 1.2983,
+      "step": 3256
+    },
+    {
+      "epoch": 0.28586553151339444,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.002674617488216873,
+      "loss": 1.2704,
+      "step": 3257
+    },
+    {
+      "epoch": 0.28595330109629696,
+      "grad_norm": 0.05615234375,
+      "learning_rate": 0.002674341288447182,
+      "loss": 1.2435,
+      "step": 3258
+    },
+    {
+      "epoch": 0.2860410706791994,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0026740649875790596,
+      "loss": 1.2771,
+      "step": 3259
+    },
+    {
+      "epoch": 0.2861288402621019,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.002673788585639775,
+      "loss": 1.2363,
+      "step": 3260
+    },
+    {
+      "epoch": 0.2862166098450044,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.002673512082656607,
+      "loss": 1.2083,
+      "step": 3261
+    },
+    {
+      "epoch": 0.2863043794279069,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.002673235478656847,
+      "loss": 1.3354,
+      "step": 3262
+    },
+    {
+      "epoch": 0.2863921490108094,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.002672958773667794,
+      "loss": 1.278,
+      "step": 3263
+    },
+    {
+      "epoch": 0.28647991859371186,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.002672681967716758,
+      "loss": 1.2476,
+      "step": 3264
+    },
+    {
+      "epoch": 0.2865676881766143,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0026724050608310573,
+      "loss": 1.3382,
+      "step": 3265
+    },
+    {
+      "epoch": 0.28665545775951684,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0026721280530380234,
+      "loss": 1.308,
+      "step": 3266
+    },
+    {
+      "epoch": 0.2867432273424193,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.002671850944364995,
+      "loss": 1.3281,
+      "step": 3267
+    },
+    {
+      "epoch": 0.2868309969253218,
+      "grad_norm": 0.05908203125,
+      "learning_rate": 0.0026715737348393213,
+      "loss": 1.2219,
+      "step": 3268
+    },
+    {
+      "epoch": 0.2869187665082243,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.002671296424488362,
+      "loss": 1.2588,
+      "step": 3269
+    },
+    {
+      "epoch": 0.28700653609112675,
+      "grad_norm": 0.059326171875,
+      "learning_rate": 0.0026710190133394872,
+      "loss": 1.1849,
+      "step": 3270
+    },
+    {
+      "epoch": 0.28709430567402927,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0026707415014200755,
+      "loss": 1.292,
+      "step": 3271
+    },
+    {
+      "epoch": 0.28718207525693173,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002670463888757517,
+      "loss": 1.2453,
+      "step": 3272
+    },
+    {
+      "epoch": 0.2872698448398342,
+      "grad_norm": 0.0595703125,
+      "learning_rate": 0.002670186175379209,
+      "loss": 1.2285,
+      "step": 3273
+    },
+    {
+      "epoch": 0.2873576144227367,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0026699083613125636,
+      "loss": 1.2668,
+      "step": 3274
+    },
+    {
+      "epoch": 0.2874453840056392,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002669630446584998,
+      "loss": 1.2596,
+      "step": 3275
+    },
+    {
+      "epoch": 0.2875331535885417,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002669352431223942,
+      "loss": 1.272,
+      "step": 3276
+    },
+    {
+      "epoch": 0.28762092317144416,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0026690743152568337,
+      "loss": 1.1951,
+      "step": 3277
+    },
+    {
+      "epoch": 0.28770869275434663,
+      "grad_norm": 0.0615234375,
+      "learning_rate": 0.0026687960987111237,
+      "loss": 1.2812,
+      "step": 3278
+    },
+    {
+      "epoch": 0.28779646233724915,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0026685177816142694,
+      "loss": 1.2518,
+      "step": 3279
+    },
+    {
+      "epoch": 0.2878842319201516,
+      "grad_norm": 0.060302734375,
+      "learning_rate": 0.0026682393639937403,
+      "loss": 1.2975,
+      "step": 3280
+    },
+    {
+      "epoch": 0.28797200150305413,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0026679608458770147,
+      "loss": 1.1963,
+      "step": 3281
+    },
+    {
+      "epoch": 0.2880597710859566,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.0026676822272915816,
+      "loss": 1.2776,
+      "step": 3282
+    },
+    {
+      "epoch": 0.28814754066885906,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0026674035082649397,
+      "loss": 1.206,
+      "step": 3283
+    },
+    {
+      "epoch": 0.2882353102517616,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0026671246888245966,
+      "loss": 1.2309,
+      "step": 3284
+    },
+    {
+      "epoch": 0.28832307983466404,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.002666845768998072,
+      "loss": 1.2657,
+      "step": 3285
+    },
+    {
+      "epoch": 0.2884108494175665,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.002666566748812893,
+      "loss": 1.2762,
+      "step": 3286
+    },
+    {
+      "epoch": 0.288498619000469,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.0026662876282965986,
+      "loss": 1.1997,
+      "step": 3287
+    },
+    {
+      "epoch": 0.2885863885833715,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.0026660084074767365,
+      "loss": 1.2974,
+      "step": 3288
+    },
+    {
+      "epoch": 0.288674158166274,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.0026657290863808646,
+      "loss": 1.2317,
+      "step": 3289
+    },
+    {
+      "epoch": 0.2887619277491765,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.0026654496650365504,
+      "loss": 1.2428,
+      "step": 3290
+    },
+    {
+      "epoch": 0.28884969733207894,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.002665170143471373,
+      "loss": 1.2552,
+      "step": 3291
+    },
+    {
+      "epoch": 0.28893746691498146,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.0026648905217129186,
+      "loss": 1.2861,
+      "step": 3292
+    },
+    {
+      "epoch": 0.2890252364978839,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.0026646107997887855,
+      "loss": 1.2968,
+      "step": 3293
+    },
+    {
+      "epoch": 0.28911300608078644,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.0026643309777265815,
+      "loss": 1.3059,
+      "step": 3294
+    },
+    {
+      "epoch": 0.2892007756636889,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.002664051055553923,
+      "loss": 1.3011,
+      "step": 3295
+    },
+    {
+      "epoch": 0.28928854524659137,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.0026637710332984388,
+      "loss": 1.2771,
+      "step": 3296
+    },
+    {
+      "epoch": 0.2893763148294939,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.0026634909109877637,
+      "loss": 1.3054,
+      "step": 3297
+    },
+    {
+      "epoch": 0.28946408441239635,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002663210688649546,
+      "loss": 1.2648,
+      "step": 3298
+    },
+    {
+      "epoch": 0.28955185399529887,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.002662930366311443,
+      "loss": 1.2607,
+      "step": 3299
+    },
+    {
+      "epoch": 0.28963962357820133,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0026626499440011203,
+      "loss": 1.2232,
+      "step": 3300
+    },
+    {
+      "epoch": 0.2897273931611038,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.002662369421746255,
+      "loss": 1.2023,
+      "step": 3301
+    },
+    {
+      "epoch": 0.2898151627440063,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.002662088799574534,
+      "loss": 1.2609,
+      "step": 3302
+    },
+    {
+      "epoch": 0.2899029323269088,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.002661808077513653,
+      "loss": 1.2576,
+      "step": 3303
+    },
+    {
+      "epoch": 0.28999070190981124,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0026615272555913184,
+      "loss": 1.2719,
+      "step": 3304
+    },
+    {
+      "epoch": 0.29007847149271376,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002661246333835246,
+      "loss": 1.263,
+      "step": 3305
+    },
+    {
+      "epoch": 0.2901662410756162,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.002660965312273161,
+      "loss": 1.2305,
+      "step": 3306
+    },
+    {
+      "epoch": 0.29025401065851875,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0026606841909328,
+      "loss": 1.2543,
+      "step": 3307
+    },
+    {
+      "epoch": 0.2903417802414212,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0026604029698419087,
+      "loss": 1.2504,
+      "step": 3308
+    },
+    {
+      "epoch": 0.2904295498243237,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0026601216490282427,
+      "loss": 1.2578,
+      "step": 3309
+    },
+    {
+      "epoch": 0.2905173194072262,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0026598402285195662,
+      "loss": 1.2319,
+      "step": 3310
+    },
+    {
+      "epoch": 0.29060508899012866,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0026595587083436552,
+      "loss": 1.2379,
+      "step": 3311
+    },
+    {
+      "epoch": 0.2906928585730312,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.0026592770885282943,
+      "loss": 1.229,
+      "step": 3312
+    },
+    {
+      "epoch": 0.29078062815593364,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.002658995369101278,
+      "loss": 1.2887,
+      "step": 3313
+    },
+    {
+      "epoch": 0.2908683977388361,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.0026587135500904116,
+      "loss": 1.2987,
+      "step": 3314
+    },
+    {
+      "epoch": 0.2909561673217386,
+      "grad_norm": 0.06201171875,
+      "learning_rate": 0.002658431631523509,
+      "loss": 1.2273,
+      "step": 3315
+    },
+    {
+      "epoch": 0.2910439369046411,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0026581496134283943,
+      "loss": 1.2565,
+      "step": 3316
+    },
+    {
+      "epoch": 0.29113170648754355,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.002657867495832902,
+      "loss": 1.2901,
+      "step": 3317
+    },
+    {
+      "epoch": 0.29121947607044607,
+      "grad_norm": 0.06103515625,
+      "learning_rate": 0.002657585278764876,
+      "loss": 1.2591,
+      "step": 3318
+    },
+    {
+      "epoch": 0.29130724565334853,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.002657302962252169,
+      "loss": 1.2941,
+      "step": 3319
+    },
+    {
+      "epoch": 0.29139501523625105,
+      "grad_norm": 0.0595703125,
+      "learning_rate": 0.002657020546322646,
+      "loss": 1.3409,
+      "step": 3320
+    },
+    {
+      "epoch": 0.2914827848191535,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0026567380310041793,
+      "loss": 1.2749,
+      "step": 3321
+    },
+    {
+      "epoch": 0.291570554402056,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.0026564554163246525,
+      "loss": 1.2671,
+      "step": 3322
+    },
+    {
+      "epoch": 0.2916583239849585,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.0026561727023119586,
+      "loss": 1.2847,
+      "step": 3323
+    },
+    {
+      "epoch": 0.29174609356786096,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.002655889888994,
+      "loss": 1.2533,
+      "step": 3324
+    },
+    {
+      "epoch": 0.2918338631507635,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.00265560697639869,
+      "loss": 1.3036,
+      "step": 3325
+    },
+    {
+      "epoch": 0.29192163273366595,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.0026553239645539495,
+      "loss": 1.2192,
+      "step": 3326
+    },
+    {
+      "epoch": 0.2920094023165684,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0026550408534877124,
+      "loss": 1.2501,
+      "step": 3327
+    },
+    {
+      "epoch": 0.29209717189947093,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0026547576432279195,
+      "loss": 1.2671,
+      "step": 3328
+    },
+    {
+      "epoch": 0.2921849414823734,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002654474333802523,
+      "loss": 1.2793,
+      "step": 3329
+    },
+    {
+      "epoch": 0.29227271106527586,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.002654190925239484,
+      "loss": 1.2288,
+      "step": 3330
+    },
+    {
+      "epoch": 0.2923604806481784,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.002653907417566774,
+      "loss": 1.2461,
+      "step": 3331
+    },
+    {
+      "epoch": 0.29244825023108084,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.002653623810812374,
+      "loss": 1.2731,
+      "step": 3332
+    },
+    {
+      "epoch": 0.29253601981398336,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0026533401050042754,
+      "loss": 1.2592,
+      "step": 3333
+    },
+    {
+      "epoch": 0.2926237893968858,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0026530563001704787,
+      "loss": 1.2732,
+      "step": 3334
+    },
+    {
+      "epoch": 0.2927115589797883,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0026527723963389936,
+      "loss": 1.2814,
+      "step": 3335
+    },
+    {
+      "epoch": 0.2927993285626908,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.002652488393537841,
+      "loss": 1.2276,
+      "step": 3336
+    },
+    {
+      "epoch": 0.29288709814559327,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0026522042917950504,
+      "loss": 1.2368,
+      "step": 3337
+    },
+    {
+      "epoch": 0.2929748677284958,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.002651920091138662,
+      "loss": 1.2755,
+      "step": 3338
+    },
+    {
+      "epoch": 0.29306263731139826,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0026516357915967246,
+      "loss": 1.2489,
+      "step": 3339
+    },
+    {
+      "epoch": 0.2931504068943007,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.002651351393197298,
+      "loss": 1.2415,
+      "step": 3340
+    },
+    {
+      "epoch": 0.29323817647720324,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002651066895968451,
+      "loss": 1.2498,
+      "step": 3341
+    },
+    {
+      "epoch": 0.2933259460601057,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.002650782299938263,
+      "loss": 1.2397,
+      "step": 3342
+    },
+    {
+      "epoch": 0.29341371564300817,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.002650497605134821,
+      "loss": 1.2603,
+      "step": 3343
+    },
+    {
+      "epoch": 0.2935014852259107,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0026502128115862243,
+      "loss": 1.3261,
+      "step": 3344
+    },
+    {
+      "epoch": 0.29358925480881315,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0026499279193205813,
+      "loss": 1.3283,
+      "step": 3345
+    },
+    {
+      "epoch": 0.29367702439171567,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.002649642928366009,
+      "loss": 1.2498,
+      "step": 3346
+    },
+    {
+      "epoch": 0.29376479397461813,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0026493578387506342,
+      "loss": 1.2134,
+      "step": 3347
+    },
+    {
+      "epoch": 0.2938525635575206,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.002649072650502596,
+      "loss": 1.2576,
+      "step": 3348
+    },
+    {
+      "epoch": 0.2939403331404231,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.00264878736365004,
+      "loss": 1.2196,
+      "step": 3349
+    },
+    {
+      "epoch": 0.2940281027233256,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.002648501978221123,
+      "loss": 1.2563,
+      "step": 3350
+    },
+    {
+      "epoch": 0.2941158723062281,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.002648216494244012,
+      "loss": 1.3172,
+      "step": 3351
+    },
+    {
+      "epoch": 0.29420364188913056,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0026479309117468826,
+      "loss": 1.2335,
+      "step": 3352
+    },
+    {
+      "epoch": 0.294291411472033,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0026476452307579207,
+      "loss": 1.3113,
+      "step": 3353
+    },
+    {
+      "epoch": 0.29437918105493555,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0026473594513053223,
+      "loss": 1.2783,
+      "step": 3354
+    },
+    {
+      "epoch": 0.294466950637838,
+      "grad_norm": 0.06787109375,
+      "learning_rate": 0.0026470735734172926,
+      "loss": 1.2683,
+      "step": 3355
+    },
+    {
+      "epoch": 0.2945547202207405,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.0026467875971220452,
+      "loss": 1.2716,
+      "step": 3356
+    },
+    {
+      "epoch": 0.294642489803643,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.002646501522447807,
+      "loss": 1.3174,
+      "step": 3357
+    },
+    {
+      "epoch": 0.29473025938654546,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0026462153494228114,
+      "loss": 1.21,
+      "step": 3358
+    },
+    {
+      "epoch": 0.294818028969448,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.002645929078075303,
+      "loss": 1.2611,
+      "step": 3359
+    },
+    {
+      "epoch": 0.29490579855235044,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0026456427084335346,
+      "loss": 1.2069,
+      "step": 3360
+    },
+    {
+      "epoch": 0.2949935681352529,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002645356240525771,
+      "loss": 1.3464,
+      "step": 3361
+    },
+    {
+      "epoch": 0.2950813377181554,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0026450696743802846,
+      "loss": 1.2969,
+      "step": 3362
+    },
+    {
+      "epoch": 0.2951691073010579,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.0026447830100253585,
+      "loss": 1.2954,
+      "step": 3363
+    },
+    {
+      "epoch": 0.2952568768839604,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0026444962474892857,
+      "loss": 1.301,
+      "step": 3364
+    },
+    {
+      "epoch": 0.29534464646686287,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0026442093868003684,
+      "loss": 1.2894,
+      "step": 3365
+    },
+    {
+      "epoch": 0.29543241604976533,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0026439224279869184,
+      "loss": 1.313,
+      "step": 3366
+    },
+    {
+      "epoch": 0.29552018563266785,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.002643635371077257,
+      "loss": 1.2582,
+      "step": 3367
+    },
+    {
+      "epoch": 0.2956079552155703,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002643348216099717,
+      "loss": 1.2957,
+      "step": 3368
+    },
+    {
+      "epoch": 0.29569572479847284,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.002643060963082638,
+      "loss": 1.2871,
+      "step": 3369
+    },
+    {
+      "epoch": 0.2957834943813753,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0026427736120543718,
+      "loss": 1.307,
+      "step": 3370
+    },
+    {
+      "epoch": 0.29587126396427776,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0026424861630432775,
+      "loss": 1.2383,
+      "step": 3371
+    },
+    {
+      "epoch": 0.2959590335471803,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0026421986160777263,
+      "loss": 1.244,
+      "step": 3372
+    },
+    {
+      "epoch": 0.29604680313008275,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.0026419109711860976,
+      "loss": 1.2831,
+      "step": 3373
+    },
+    {
+      "epoch": 0.2961345727129852,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.002641623228396781,
+      "loss": 1.2764,
+      "step": 3374
+    },
+    {
+      "epoch": 0.29622234229588773,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.002641335387738175,
+      "loss": 1.2788,
+      "step": 3375
+    },
+    {
+      "epoch": 0.2963101118787902,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.0026410474492386893,
+      "loss": 1.2458,
+      "step": 3376
+    },
+    {
+      "epoch": 0.2963978814616927,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.002640759412926742,
+      "loss": 1.2588,
+      "step": 3377
+    },
+    {
+      "epoch": 0.2964856510445952,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0026404712788307603,
+      "loss": 1.2749,
+      "step": 3378
+    },
+    {
+      "epoch": 0.29657342062749764,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0026401830469791826,
+      "loss": 1.3258,
+      "step": 3379
+    },
+    {
+      "epoch": 0.29666119021040016,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.002639894717400456,
+      "loss": 1.2941,
+      "step": 3380
+    },
+    {
+      "epoch": 0.2967489597933026,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002639606290123038,
+      "loss": 1.2627,
+      "step": 3381
+    },
+    {
+      "epoch": 0.29683672937620514,
+      "grad_norm": 0.060546875,
+      "learning_rate": 0.0026393177651753944,
+      "loss": 1.2329,
+      "step": 3382
+    },
+    {
+      "epoch": 0.2969244989591076,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.002639029142586002,
+      "loss": 1.2544,
+      "step": 3383
+    },
+    {
+      "epoch": 0.29701226854201007,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.002638740422383347,
+      "loss": 1.2586,
+      "step": 3384
+    },
+    {
+      "epoch": 0.2971000381249126,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.002638451604595924,
+      "loss": 1.2558,
+      "step": 3385
+    },
+    {
+      "epoch": 0.29718780770781505,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002638162689252239,
+      "loss": 1.2762,
+      "step": 3386
+    },
+    {
+      "epoch": 0.2972755772907175,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.0026378736763808066,
+      "loss": 1.2893,
+      "step": 3387
+    },
+    {
+      "epoch": 0.29736334687362004,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.002637584566010151,
+      "loss": 1.2352,
+      "step": 3388
+    },
+    {
+      "epoch": 0.2974511164565225,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.0026372953581688062,
+      "loss": 1.25,
+      "step": 3389
+    },
+    {
+      "epoch": 0.297538886039425,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.0026370060528853167,
+      "loss": 1.2761,
+      "step": 3390
+    },
+    {
+      "epoch": 0.2976266556223275,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0026367166501882347,
+      "loss": 1.3038,
+      "step": 3391
+    },
+    {
+      "epoch": 0.29771442520522995,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.002636427150106124,
+      "loss": 1.2012,
+      "step": 3392
+    },
+    {
+      "epoch": 0.29780219478813247,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.002636137552667556,
+      "loss": 1.2477,
+      "step": 3393
+    },
+    {
+      "epoch": 0.29788996437103493,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 0.002635847857901114,
+      "loss": 1.2499,
+      "step": 3394
+    },
+    {
+      "epoch": 0.29797773395393745,
+      "grad_norm": 0.068359375,
+      "learning_rate": 0.0026355580658353893,
+      "loss": 1.2579,
+      "step": 3395
+    },
+    {
+      "epoch": 0.2980655035368399,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002635268176498983,
+      "loss": 1.2835,
+      "step": 3396
+    },
+    {
+      "epoch": 0.2981532731197424,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.002634978189920506,
+      "loss": 1.3178,
+      "step": 3397
+    },
+    {
+      "epoch": 0.2982410427026449,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.002634688106128579,
+      "loss": 1.2855,
+      "step": 3398
+    },
+    {
+      "epoch": 0.29832881228554736,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.0026343979251518324,
+      "loss": 1.2734,
+      "step": 3399
+    },
+    {
+      "epoch": 0.2984165818684498,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.0026341076470189056,
+      "loss": 1.2831,
+      "step": 3400
+    },
+    {
+      "epoch": 0.29850435145135235,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.002633817271758448,
+      "loss": 1.271,
+      "step": 3401
+    },
+    {
+      "epoch": 0.2985921210342548,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0026335267993991183,
+      "loss": 1.209,
+      "step": 3402
+    },
+    {
+      "epoch": 0.29867989061715733,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.002633236229969585,
+      "loss": 1.2606,
+      "step": 3403
+    },
+    {
+      "epoch": 0.2987676602000598,
+      "grad_norm": 0.06591796875,
+      "learning_rate": 0.0026329455634985267,
+      "loss": 1.3079,
+      "step": 3404
+    },
+    {
+      "epoch": 0.29885542978296226,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0026326548000146304,
+      "loss": 1.2775,
+      "step": 3405
+    },
+    {
+      "epoch": 0.2989431993658648,
+      "grad_norm": 0.062255859375,
+      "learning_rate": 0.0026323639395465933,
+      "loss": 1.2821,
+      "step": 3406
+    },
+    {
+      "epoch": 0.29903096894876724,
+      "grad_norm": 0.06591796875,
+      "learning_rate": 0.0026320729821231225,
+      "loss": 1.2218,
+      "step": 3407
+    },
+    {
+      "epoch": 0.29911873853166976,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0026317819277729346,
+      "loss": 1.2085,
+      "step": 3408
+    },
+    {
+      "epoch": 0.2992065081145722,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.002631490776524755,
+      "loss": 1.1999,
+      "step": 3409
+    },
+    {
+      "epoch": 0.2992942776974747,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0026311995284073186,
+      "loss": 1.2382,
+      "step": 3410
+    },
+    {
+      "epoch": 0.2993820472803772,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.002630908183449372,
+      "loss": 1.2776,
+      "step": 3411
+    },
+    {
+      "epoch": 0.29946981686327967,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.0026306167416796683,
+      "loss": 1.2725,
+      "step": 3412
+    },
+    {
+      "epoch": 0.29955758644618213,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.002630325203126973,
+      "loss": 1.2999,
+      "step": 3413
+    },
+    {
+      "epoch": 0.29964535602908465,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.0026300335678200587,
+      "loss": 1.2479,
+      "step": 3414
+    },
+    {
+      "epoch": 0.2997331256119871,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.002629741835787709,
+      "loss": 1.2683,
+      "step": 3415
+    },
+    {
+      "epoch": 0.29982089519488964,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0026294500070587166,
+      "loss": 1.2631,
+      "step": 3416
+    },
+    {
+      "epoch": 0.2999086647777921,
+      "grad_norm": 0.0615234375,
+      "learning_rate": 0.002629158081661885,
+      "loss": 1.2528,
+      "step": 3417
+    },
+    {
+      "epoch": 0.29999643436069456,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.002628866059626024,
+      "loss": 1.2879,
+      "step": 3418
+    },
+    {
+      "epoch": 0.3000842039435971,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.0026285739409799566,
+      "loss": 1.2246,
+      "step": 3419
+    },
+    {
+      "epoch": 0.30017197352649955,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.0026282817257525133,
+      "loss": 1.3056,
+      "step": 3420
+    },
+    {
+      "epoch": 0.30025974310940207,
+      "grad_norm": 0.06201171875,
+      "learning_rate": 0.0026279894139725336,
+      "loss": 1.2958,
+      "step": 3421
+    },
+    {
+      "epoch": 0.30034751269230453,
+      "grad_norm": 0.05810546875,
+      "learning_rate": 0.00262769700566887,
+      "loss": 1.2394,
+      "step": 3422
+    },
+    {
+      "epoch": 0.300435282275207,
+      "grad_norm": 0.057373046875,
+      "learning_rate": 0.0026274045008703793,
+      "loss": 1.2265,
+      "step": 3423
+    },
+    {
+      "epoch": 0.3005230518581095,
+      "grad_norm": 0.06787109375,
+      "learning_rate": 0.002627111899605932,
+      "loss": 1.2423,
+      "step": 3424
+    },
+    {
+      "epoch": 0.300610821441012,
+      "grad_norm": 0.058837890625,
+      "learning_rate": 0.0026268192019044062,
+      "loss": 1.2272,
+      "step": 3425
+    },
+    {
+      "epoch": 0.30069859102391444,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.0026265264077946914,
+      "loss": 1.2883,
+      "step": 3426
+    },
+    {
+      "epoch": 0.30078636060681696,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0026262335173056826,
+      "loss": 1.2927,
+      "step": 3427
+    },
+    {
+      "epoch": 0.3008741301897194,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0026259405304662896,
+      "loss": 1.2348,
+      "step": 3428
+    },
+    {
+      "epoch": 0.30096189977262194,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.002625647447305427,
+      "loss": 1.3007,
+      "step": 3429
+    },
+    {
+      "epoch": 0.3010496693555244,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.002625354267852022,
+      "loss": 1.2587,
+      "step": 3430
+    },
+    {
+      "epoch": 0.30113743893842687,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0026250609921350107,
+      "loss": 1.2667,
+      "step": 3431
+    },
+    {
+      "epoch": 0.3012252085213294,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002624767620183337,
+      "loss": 1.2568,
+      "step": 3432
+    },
+    {
+      "epoch": 0.30131297810423185,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.0026244741520259564,
+      "loss": 1.2506,
+      "step": 3433
+    },
+    {
+      "epoch": 0.3014007476871344,
+      "grad_norm": 0.0634765625,
+      "learning_rate": 0.0026241805876918325,
+      "loss": 1.2402,
+      "step": 3434
+    },
+    {
+      "epoch": 0.30148851727003684,
+      "grad_norm": 0.060302734375,
+      "learning_rate": 0.0026238869272099392,
+      "loss": 1.2475,
+      "step": 3435
+    },
+    {
+      "epoch": 0.3015762868529393,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.00262359317060926,
+      "loss": 1.2976,
+      "step": 3436
+    },
+    {
+      "epoch": 0.3016640564358418,
+      "grad_norm": 0.059326171875,
+      "learning_rate": 0.002623299317918787,
+      "loss": 1.2157,
+      "step": 3437
+    },
+    {
+      "epoch": 0.3017518260187443,
+      "grad_norm": 0.060791015625,
+      "learning_rate": 0.0026230053691675224,
+      "loss": 1.2716,
+      "step": 3438
+    },
+    {
+      "epoch": 0.3018395956016468,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.002622711324384478,
+      "loss": 1.2368,
+      "step": 3439
+    },
+    {
+      "epoch": 0.30192736518454927,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.002622417183598674,
+      "loss": 1.2473,
+      "step": 3440
+    },
+    {
+      "epoch": 0.30201513476745173,
+      "grad_norm": 0.05859375,
+      "learning_rate": 0.0026221229468391424,
+      "loss": 1.2557,
+      "step": 3441
+    },
+    {
+      "epoch": 0.30210290435035425,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.0026218286141349227,
+      "loss": 1.2355,
+      "step": 3442
+    },
+    {
+      "epoch": 0.3021906739332567,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.0026215341855150634,
+      "loss": 1.308,
+      "step": 3443
+    },
+    {
+      "epoch": 0.3022784435161592,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.0026212396610086246,
+      "loss": 1.2519,
+      "step": 3444
+    },
+    {
+      "epoch": 0.3023662130990617,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.0026209450406446742,
+      "loss": 1.223,
+      "step": 3445
+    },
+    {
+      "epoch": 0.30245398268196416,
+      "grad_norm": 0.07470703125,
+      "learning_rate": 0.00262065032445229,
+      "loss": 1.2368,
+      "step": 3446
+    },
+    {
+      "epoch": 0.3025417522648667,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.0026203555124605597,
+      "loss": 1.2802,
+      "step": 3447
+    },
+    {
+      "epoch": 0.30262952184776915,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0026200606046985794,
+      "loss": 1.2761,
+      "step": 3448
+    },
+    {
+      "epoch": 0.3027172914306716,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0026197656011954563,
+      "loss": 1.2377,
+      "step": 3449
+    },
+    {
+      "epoch": 0.30280506101357413,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0026194705019803046,
+      "loss": 1.2664,
+      "step": 3450
+    },
+    {
+      "epoch": 0.3028928305964766,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.002619175307082251,
+      "loss": 1.28,
+      "step": 3451
+    },
+    {
+      "epoch": 0.3029806001793791,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0026188800165304294,
+      "loss": 1.2825,
+      "step": 3452
+    },
+    {
+      "epoch": 0.3030683697622816,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.002618584630353984,
+      "loss": 1.2874,
+      "step": 3453
+    },
+    {
+      "epoch": 0.30315613934518404,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.002618289148582068,
+      "loss": 1.2591,
+      "step": 3454
+    },
+    {
+      "epoch": 0.30324390892808656,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.002617993571243844,
+      "loss": 1.2472,
+      "step": 3455
+    },
+    {
+      "epoch": 0.303331678510989,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0026176978983684855,
+      "loss": 1.2675,
+      "step": 3456
+    },
+    {
+      "epoch": 0.3034194480938915,
+      "grad_norm": 0.06787109375,
+      "learning_rate": 0.002617402129985173,
+      "loss": 1.2374,
+      "step": 3457
+    },
+    {
+      "epoch": 0.303507217676794,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0026171062661230976,
+      "loss": 1.2856,
+      "step": 3458
+    },
+    {
+      "epoch": 0.30359498725969647,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.002616810306811461,
+      "loss": 1.3043,
+      "step": 3459
+    },
+    {
+      "epoch": 0.303682756842599,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.002616514252079472,
+      "loss": 1.2512,
+      "step": 3460
+    },
+    {
+      "epoch": 0.30377052642550145,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002616218101956351,
+      "loss": 1.3342,
+      "step": 3461
+    },
+    {
+      "epoch": 0.3038582960084039,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.0026159218564713264,
+      "loss": 1.239,
+      "step": 3462
+    },
+    {
+      "epoch": 0.30394606559130644,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002615625515653637,
+      "loss": 1.2244,
+      "step": 3463
+    },
+    {
+      "epoch": 0.3040338351742089,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.002615329079532529,
+      "loss": 1.2765,
+      "step": 3464
+    },
+    {
+      "epoch": 0.3041216047571114,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0026150325481372615,
+      "loss": 1.2157,
+      "step": 3465
+    },
+    {
+      "epoch": 0.3042093743400139,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.002614735921497099,
+      "loss": 1.3023,
+      "step": 3466
+    },
+    {
+      "epoch": 0.30429714392291635,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.0026144391996413188,
+      "loss": 1.2464,
+      "step": 3467
+    },
+    {
+      "epoch": 0.30438491350581887,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0026141423825992056,
+      "loss": 1.2427,
+      "step": 3468
+    },
+    {
+      "epoch": 0.30447268308872133,
+      "grad_norm": 0.056396484375,
+      "learning_rate": 0.002613845470400054,
+      "loss": 1.2725,
+      "step": 3469
+    },
+    {
+      "epoch": 0.3045604526716238,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002613548463073169,
+      "loss": 1.2515,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3046482222545263,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0026132513606478623,
+      "loss": 1.2545,
+      "step": 3471
+    },
+    {
+      "epoch": 0.3047359918374288,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.002612954163153458,
+      "loss": 1.2926,
+      "step": 3472
+    },
+    {
+      "epoch": 0.3048237614203313,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.002612656870619288,
+      "loss": 1.1954,
+      "step": 3473
+    },
+    {
+      "epoch": 0.30491153100323376,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.0026123594830746943,
+      "loss": 1.2229,
+      "step": 3474
+    },
+    {
+      "epoch": 0.3049993005861362,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0026120620005490275,
+      "loss": 1.2445,
+      "step": 3475
+    },
+    {
+      "epoch": 0.30508707016903874,
+      "grad_norm": 0.06103515625,
+      "learning_rate": 0.0026117644230716475,
+      "loss": 1.2419,
+      "step": 3476
+    },
+    {
+      "epoch": 0.3051748397519412,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0026114667506719246,
+      "loss": 1.3121,
+      "step": 3477
+    },
+    {
+      "epoch": 0.3052626093348437,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.0026111689833792375,
+      "loss": 1.2266,
+      "step": 3478
+    },
+    {
+      "epoch": 0.3053503789177462,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.002610871121222975,
+      "loss": 1.2657,
+      "step": 3479
+    },
+    {
+      "epoch": 0.30543814850064865,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.0026105731642325353,
+      "loss": 1.2014,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3055259180835512,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002610275112437325,
+      "loss": 1.2845,
+      "step": 3481
+    },
+    {
+      "epoch": 0.30561368766645364,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0026099769658667602,
+      "loss": 1.2589,
+      "step": 3482
+    },
+    {
+      "epoch": 0.3057014572493561,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.0026096787245502673,
+      "loss": 1.3053,
+      "step": 3483
+    },
+    {
+      "epoch": 0.3057892268322586,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.002609380388517282,
+      "loss": 1.2686,
+      "step": 3484
+    },
+    {
+      "epoch": 0.3058769964151611,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0026090819577972483,
+      "loss": 1.297,
+      "step": 3485
+    },
+    {
+      "epoch": 0.3059647659980636,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0026087834324196204,
+      "loss": 1.2643,
+      "step": 3486
+    },
+    {
+      "epoch": 0.30605253558096607,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.002608484812413861,
+      "loss": 1.2466,
+      "step": 3487
+    },
+    {
+      "epoch": 0.30614030516386853,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0026081860978094435,
+      "loss": 1.2822,
+      "step": 3488
+    },
+    {
+      "epoch": 0.30622807474677105,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002607887288635849,
+      "loss": 1.2487,
+      "step": 3489
+    },
+    {
+      "epoch": 0.3063158443296735,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0026075883849225696,
+      "loss": 1.3226,
+      "step": 3490
+    },
+    {
+      "epoch": 0.30640361391257603,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.002607289386699106,
+      "loss": 1.2883,
+      "step": 3491
+    },
+    {
+      "epoch": 0.3064913834954785,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.002606990293994967,
+      "loss": 1.307,
+      "step": 3492
+    },
+    {
+      "epoch": 0.30657915307838096,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0026066911068396733,
+      "loss": 1.2099,
+      "step": 3493
+    },
+    {
+      "epoch": 0.3066669226612835,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.0026063918252627522,
+      "loss": 1.2483,
+      "step": 3494
+    },
+    {
+      "epoch": 0.30675469224418594,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0026060924492937424,
+      "loss": 1.2298,
+      "step": 3495
+    },
+    {
+      "epoch": 0.3068424618270884,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.0026057929789621913,
+      "loss": 1.3231,
+      "step": 3496
+    },
+    {
+      "epoch": 0.30693023140999093,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0026054934142976547,
+      "loss": 1.2646,
+      "step": 3497
+    },
+    {
+      "epoch": 0.3070180009928934,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.002605193755329699,
+      "loss": 1.2266,
+      "step": 3498
+    },
+    {
+      "epoch": 0.3071057705757959,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.002604894002087899,
+      "loss": 1.2036,
+      "step": 3499
+    },
+    {
+      "epoch": 0.3071935401586984,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.002604594154601839,
+      "loss": 1.254,
+      "step": 3500
+    },
+    {
+      "epoch": 0.3071935401586984,
+      "eval_loss": 1.246621012687683,
+      "eval_runtime": 437.6804,
+      "eval_samples_per_second": 33.678,
+      "eval_steps_per_second": 8.419,
+      "step": 3500
+    },
+    {
+      "epoch": 0.30728130974160084,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0026042942129011135,
+      "loss": 1.1991,
+      "step": 3501
+    },
+    {
+      "epoch": 0.30736907932450336,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.0026039941770153252,
+      "loss": 1.2096,
+      "step": 3502
+    },
+    {
+      "epoch": 0.3074568489074058,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.0026036940469740862,
+      "loss": 1.2317,
+      "step": 3503
+    },
+    {
+      "epoch": 0.30754461849030834,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.002603393822807018,
+      "loss": 1.2436,
+      "step": 3504
+    },
+    {
+      "epoch": 0.3076323880732108,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.0026030935045437526,
+      "loss": 1.2749,
+      "step": 3505
+    },
+    {
+      "epoch": 0.30772015765611327,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002602793092213929,
+      "loss": 1.2819,
+      "step": 3506
+    },
+    {
+      "epoch": 0.3078079272390158,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0026024925858471977,
+      "loss": 1.2134,
+      "step": 3507
+    },
+    {
+      "epoch": 0.30789569682191825,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0026021919854732163,
+      "loss": 1.2651,
+      "step": 3508
+    },
+    {
+      "epoch": 0.30798346640482077,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.002601891291121654,
+      "loss": 1.2901,
+      "step": 3509
+    },
+    {
+      "epoch": 0.30807123598772324,
+      "grad_norm": 0.05908203125,
+      "learning_rate": 0.002601590502822188,
+      "loss": 1.2187,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3081590055706257,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.002601289620604504,
+      "loss": 1.2439,
+      "step": 3511
+    },
+    {
+      "epoch": 0.3082467751535282,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.002600988644498299,
+      "loss": 1.2828,
+      "step": 3512
+    },
+    {
+      "epoch": 0.3083345447364307,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0026006875745332777,
+      "loss": 1.2665,
+      "step": 3513
+    },
+    {
+      "epoch": 0.30842231431933315,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002600386410739154,
+      "loss": 1.2554,
+      "step": 3514
+    },
+    {
+      "epoch": 0.30851008390223567,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0026000851531456527,
+      "loss": 1.2307,
+      "step": 3515
+    },
+    {
+      "epoch": 0.30859785348513813,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.002599783801782505,
+      "loss": 1.2609,
+      "step": 3516
+    },
+    {
+      "epoch": 0.30868562306804065,
+      "grad_norm": 0.056884765625,
+      "learning_rate": 0.002599482356679455,
+      "loss": 1.2158,
+      "step": 3517
+    },
+    {
+      "epoch": 0.3087733926509431,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002599180817866253,
+      "loss": 1.2437,
+      "step": 3518
+    },
+    {
+      "epoch": 0.3088611622338456,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.00259887918537266,
+      "loss": 1.2729,
+      "step": 3519
+    },
+    {
+      "epoch": 0.3089489318167481,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002598577459228447,
+      "loss": 1.2513,
+      "step": 3520
+    },
+    {
+      "epoch": 0.30903670139965056,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002598275639463391,
+      "loss": 1.2668,
+      "step": 3521
+    },
+    {
+      "epoch": 0.3091244709825531,
+      "grad_norm": 0.060302734375,
+      "learning_rate": 0.002597973726107281,
+      "loss": 1.2166,
+      "step": 3522
+    },
+    {
+      "epoch": 0.30921224056545554,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002597671719189916,
+      "loss": 1.3082,
+      "step": 3523
+    },
+    {
+      "epoch": 0.309300010148358,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0025973696187411026,
+      "loss": 1.2779,
+      "step": 3524
+    },
+    {
+      "epoch": 0.3093877797312605,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0025970674247906556,
+      "loss": 1.312,
+      "step": 3525
+    },
+    {
+      "epoch": 0.309475549314163,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002596765137368401,
+      "loss": 1.2172,
+      "step": 3526
+    },
+    {
+      "epoch": 0.30956331889706545,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.002596462756504174,
+      "loss": 1.2589,
+      "step": 3527
+    },
+    {
+      "epoch": 0.309651088479968,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.002596160282227817,
+      "loss": 1.3159,
+      "step": 3528
+    },
+    {
+      "epoch": 0.30973885806287044,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.002595857714569185,
+      "loss": 1.27,
+      "step": 3529
+    },
+    {
+      "epoch": 0.30982662764577296,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0025955550535581387,
+      "loss": 1.3142,
+      "step": 3530
+    },
+    {
+      "epoch": 0.3099143972286754,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.0025952522992245502,
+      "loss": 1.2416,
+      "step": 3531
+    },
+    {
+      "epoch": 0.3100021668115779,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0025949494515982996,
+      "loss": 1.2216,
+      "step": 3532
+    },
+    {
+      "epoch": 0.3100899363944804,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.002594646510709277,
+      "loss": 1.3537,
+      "step": 3533
+    },
+    {
+      "epoch": 0.31017770597738287,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0025943434765873822,
+      "loss": 1.2775,
+      "step": 3534
+    },
+    {
+      "epoch": 0.3102654755602854,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0025940403492625223,
+      "loss": 1.2768,
+      "step": 3535
+    },
+    {
+      "epoch": 0.31035324514318785,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.0025937371287646153,
+      "loss": 1.2531,
+      "step": 3536
+    },
+    {
+      "epoch": 0.3104410147260903,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.002593433815123588,
+      "loss": 1.258,
+      "step": 3537
+    },
+    {
+      "epoch": 0.31052878430899283,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.002593130408369376,
+      "loss": 1.2718,
+      "step": 3538
+    },
+    {
+      "epoch": 0.3106165538918953,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.002592826908531925,
+      "loss": 1.243,
+      "step": 3539
+    },
+    {
+      "epoch": 0.31070432347479776,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.0025925233156411883,
+      "loss": 1.2287,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3107920930577003,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0025922196297271296,
+      "loss": 1.2506,
+      "step": 3541
+    },
+    {
+      "epoch": 0.31087986264060274,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.0025919158508197217,
+      "loss": 1.3043,
+      "step": 3542
+    },
+    {
+      "epoch": 0.31096763222350526,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0025916119789489464,
+      "loss": 1.2521,
+      "step": 3543
+    },
+    {
+      "epoch": 0.3110554018064077,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.0025913080141447943,
+      "loss": 1.275,
+      "step": 3544
+    },
+    {
+      "epoch": 0.3111431713893102,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0025910039564372663,
+      "loss": 1.2509,
+      "step": 3545
+    },
+    {
+      "epoch": 0.3112309409722127,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.002590699805856371,
+      "loss": 1.2654,
+      "step": 3546
+    },
+    {
+      "epoch": 0.3113187105551152,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.002590395562432127,
+      "loss": 1.2131,
+      "step": 3547
+    },
+    {
+      "epoch": 0.3114064801380177,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0025900912261945614,
+      "loss": 1.2613,
+      "step": 3548
+    },
+    {
+      "epoch": 0.31149424972092016,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.002589786797173712,
+      "loss": 1.3005,
+      "step": 3549
+    },
+    {
+      "epoch": 0.3115820193038226,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.0025894822753996243,
+      "loss": 1.244,
+      "step": 3550
+    },
+    {
+      "epoch": 0.31166978888672514,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0025891776609023535,
+      "loss": 1.2597,
+      "step": 3551
+    },
+    {
+      "epoch": 0.3117575584696276,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0025888729537119634,
+      "loss": 1.2941,
+      "step": 3552
+    },
+    {
+      "epoch": 0.31184532805253007,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0025885681538585278,
+      "loss": 1.2666,
+      "step": 3553
+    },
+    {
+      "epoch": 0.3119330976354326,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.002588263261372129,
+      "loss": 1.222,
+      "step": 3554
+    },
+    {
+      "epoch": 0.31202086721833505,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002587958276282859,
+      "loss": 1.282,
+      "step": 3555
+    },
+    {
+      "epoch": 0.31210863680123757,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0025876531986208187,
+      "loss": 1.2138,
+      "step": 3556
+    },
+    {
+      "epoch": 0.31219640638414003,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0025873480284161173,
+      "loss": 1.2079,
+      "step": 3557
+    },
+    {
+      "epoch": 0.3122841759670425,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.0025870427656988744,
+      "loss": 1.2885,
+      "step": 3558
+    },
+    {
+      "epoch": 0.312371945549945,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.002586737410499219,
+      "loss": 1.2305,
+      "step": 3559
+    },
+    {
+      "epoch": 0.3124597151328475,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0025864319628472864,
+      "loss": 1.2768,
+      "step": 3560
+    },
+    {
+      "epoch": 0.31254748471575,
+      "grad_norm": 0.06591796875,
+      "learning_rate": 0.0025861264227732255,
+      "loss": 1.1599,
+      "step": 3561
+    },
+    {
+      "epoch": 0.31263525429865247,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0025858207903071903,
+      "loss": 1.3309,
+      "step": 3562
+    },
+    {
+      "epoch": 0.31272302388155493,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.002585515065479346,
+      "loss": 1.227,
+      "step": 3563
+    },
+    {
+      "epoch": 0.31281079346445745,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002585209248319866,
+      "loss": 1.3205,
+      "step": 3564
+    },
+    {
+      "epoch": 0.3128985630473599,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.0025849033388589343,
+      "loss": 1.2751,
+      "step": 3565
+    },
+    {
+      "epoch": 0.3129863326302624,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002584597337126742,
+      "loss": 1.2963,
+      "step": 3566
+    },
+    {
+      "epoch": 0.3130741022131649,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.00258429124315349,
+      "loss": 1.3122,
+      "step": 3567
+    },
+    {
+      "epoch": 0.31316187179606736,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0025839850569693897,
+      "loss": 1.2886,
+      "step": 3568
+    },
+    {
+      "epoch": 0.3132496413789699,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.0025836787786046598,
+      "loss": 1.2859,
+      "step": 3569
+    },
+    {
+      "epoch": 0.31333741096187234,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.0025833724080895288,
+      "loss": 1.2981,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3134251805447748,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.002583065945454234,
+      "loss": 1.224,
+      "step": 3571
+    },
+    {
+      "epoch": 0.3135129501276773,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.002582759390729023,
+      "loss": 1.2456,
+      "step": 3572
+    },
+    {
+      "epoch": 0.3136007197105798,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0025824527439441505,
+      "loss": 1.2585,
+      "step": 3573
+    },
+    {
+      "epoch": 0.3136884892934823,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.0025821460051298813,
+      "loss": 1.2229,
+      "step": 3574
+    },
+    {
+      "epoch": 0.3137762588763848,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0025818391743164903,
+      "loss": 1.2405,
+      "step": 3575
+    },
+    {
+      "epoch": 0.31386402845928724,
+      "grad_norm": 0.0634765625,
+      "learning_rate": 0.0025815322515342597,
+      "loss": 1.1832,
+      "step": 3576
+    },
+    {
+      "epoch": 0.31395179804218976,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.002581225236813482,
+      "loss": 1.2483,
+      "step": 3577
+    },
+    {
+      "epoch": 0.3140395676250922,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0025809181301844576,
+      "loss": 1.2632,
+      "step": 3578
+    },
+    {
+      "epoch": 0.31412733720799474,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0025806109316774974,
+      "loss": 1.2861,
+      "step": 3579
+    },
+    {
+      "epoch": 0.3142151067908972,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0025803036413229205,
+      "loss": 1.2116,
+      "step": 3580
+    },
+    {
+      "epoch": 0.31430287637379967,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.0025799962591510554,
+      "loss": 1.2423,
+      "step": 3581
+    },
+    {
+      "epoch": 0.3143906459567022,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0025796887851922395,
+      "loss": 1.2125,
+      "step": 3582
+    },
+    {
+      "epoch": 0.31447841553960465,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.0025793812194768186,
+      "loss": 1.2657,
+      "step": 3583
+    },
+    {
+      "epoch": 0.3145661851225071,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0025790735620351488,
+      "loss": 1.2553,
+      "step": 3584
+    },
+    {
+      "epoch": 0.31465395470540963,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0025787658128975945,
+      "loss": 1.2947,
+      "step": 3585
+    },
+    {
+      "epoch": 0.3147417242883121,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.00257845797209453,
+      "loss": 1.1999,
+      "step": 3586
+    },
+    {
+      "epoch": 0.3148294938712146,
+      "grad_norm": 0.054931640625,
+      "learning_rate": 0.0025781500396563368,
+      "loss": 1.2308,
+      "step": 3587
+    },
+    {
+      "epoch": 0.3149172634541171,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002577842015613407,
+      "loss": 1.2637,
+      "step": 3588
+    },
+    {
+      "epoch": 0.31500503303701954,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0025775338999961418,
+      "loss": 1.2466,
+      "step": 3589
+    },
+    {
+      "epoch": 0.31509280261992206,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.002577225692834951,
+      "loss": 1.2522,
+      "step": 3590
+    },
+    {
+      "epoch": 0.3151805722028245,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0025769173941602532,
+      "loss": 1.3115,
+      "step": 3591
+    },
+    {
+      "epoch": 0.31526834178572705,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.002576609004002476,
+      "loss": 1.2314,
+      "step": 3592
+    },
+    {
+      "epoch": 0.3153561113686295,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.002576300522392057,
+      "loss": 1.2635,
+      "step": 3593
+    },
+    {
+      "epoch": 0.315443880951532,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0025759919493594406,
+      "loss": 1.2546,
+      "step": 3594
+    },
+    {
+      "epoch": 0.3155316505344345,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0025756832849350834,
+      "loss": 1.2556,
+      "step": 3595
+    },
+    {
+      "epoch": 0.31561942011733696,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0025753745291494487,
+      "loss": 1.2822,
+      "step": 3596
+    },
+    {
+      "epoch": 0.3157071897002394,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0025750656820330096,
+      "loss": 1.243,
+      "step": 3597
+    },
+    {
+      "epoch": 0.31579495928314194,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.002574756743616248,
+      "loss": 1.2971,
+      "step": 3598
+    },
+    {
+      "epoch": 0.3158827288660444,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0025744477139296542,
+      "loss": 1.2585,
+      "step": 3599
+    },
+    {
+      "epoch": 0.3159704984489469,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0025741385930037296,
+      "loss": 1.2598,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3160582680318494,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0025738293808689823,
+      "loss": 1.2685,
+      "step": 3601
+    },
+    {
+      "epoch": 0.31614603761475185,
+      "grad_norm": 0.06787109375,
+      "learning_rate": 0.002573520077555931,
+      "loss": 1.2809,
+      "step": 3602
+    },
+    {
+      "epoch": 0.31623380719765437,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0025732106830951016,
+      "loss": 1.2518,
+      "step": 3603
+    },
+    {
+      "epoch": 0.31632157678055683,
+      "grad_norm": 0.068359375,
+      "learning_rate": 0.0025729011975170308,
+      "loss": 1.2493,
+      "step": 3604
+    },
+    {
+      "epoch": 0.31640934636345935,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0025725916208522636,
+      "loss": 1.2884,
+      "step": 3605
+    },
+    {
+      "epoch": 0.3164971159463618,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.002572281953131354,
+      "loss": 1.21,
+      "step": 3606
+    },
+    {
+      "epoch": 0.3165848855292643,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.002571972194384865,
+      "loss": 1.3251,
+      "step": 3607
+    },
+    {
+      "epoch": 0.3166726551121668,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.0025716623446433683,
+      "loss": 1.288,
+      "step": 3608
+    },
+    {
+      "epoch": 0.31676042469506926,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.002571352403937445,
+      "loss": 1.2037,
+      "step": 3609
+    },
+    {
+      "epoch": 0.31684819427797173,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0025710423722976852,
+      "loss": 1.2591,
+      "step": 3610
+    },
+    {
+      "epoch": 0.31693596386087425,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0025707322497546875,
+      "loss": 1.2781,
+      "step": 3611
+    },
+    {
+      "epoch": 0.3170237334437767,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.00257042203633906,
+      "loss": 1.353,
+      "step": 3612
+    },
+    {
+      "epoch": 0.31711150302667923,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.0025701117320814193,
+      "loss": 1.2505,
+      "step": 3613
+    },
+    {
+      "epoch": 0.3171992726095817,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0025698013370123915,
+      "loss": 1.2587,
+      "step": 3614
+    },
+    {
+      "epoch": 0.31728704219248416,
+      "grad_norm": 0.059326171875,
+      "learning_rate": 0.002569490851162611,
+      "loss": 1.2979,
+      "step": 3615
+    },
+    {
+      "epoch": 0.3173748117753867,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.002569180274562722,
+      "loss": 1.3164,
+      "step": 3616
+    },
+    {
+      "epoch": 0.31746258135828914,
+      "grad_norm": 0.0556640625,
+      "learning_rate": 0.0025688696072433768,
+      "loss": 1.1939,
+      "step": 3617
+    },
+    {
+      "epoch": 0.31755035094119166,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.002568558849235237,
+      "loss": 1.1871,
+      "step": 3618
+    },
+    {
+      "epoch": 0.3176381205240941,
+      "grad_norm": 0.058837890625,
+      "learning_rate": 0.002568248000568974,
+      "loss": 1.2734,
+      "step": 3619
+    },
+    {
+      "epoch": 0.3177258901069966,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.002567937061275266,
+      "loss": 1.2382,
+      "step": 3620
+    },
+    {
+      "epoch": 0.3178136596898991,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0025676260313848022,
+      "loss": 1.2273,
+      "step": 3621
+    },
+    {
+      "epoch": 0.3179014292728016,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.0025673149109282805,
+      "loss": 1.2013,
+      "step": 3622
+    },
+    {
+      "epoch": 0.31798919885570404,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.002567003699936406,
+      "loss": 1.2256,
+      "step": 3623
+    },
+    {
+      "epoch": 0.31807696843860656,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.002566692398439895,
+      "loss": 1.2752,
+      "step": 3624
+    },
+    {
+      "epoch": 0.318164738021509,
+      "grad_norm": 0.05712890625,
+      "learning_rate": 0.0025663810064694716,
+      "loss": 1.2639,
+      "step": 3625
+    },
+    {
+      "epoch": 0.31825250760441154,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0025660695240558684,
+      "loss": 1.192,
+      "step": 3626
+    },
+    {
+      "epoch": 0.318340277187314,
+      "grad_norm": 0.056396484375,
+      "learning_rate": 0.0025657579512298285,
+      "loss": 1.2568,
+      "step": 3627
+    },
+    {
+      "epoch": 0.31842804677021647,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0025654462880221017,
+      "loss": 1.2127,
+      "step": 3628
+    },
+    {
+      "epoch": 0.318515816353119,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.0025651345344634485,
+      "loss": 1.2482,
+      "step": 3629
+    },
+    {
+      "epoch": 0.31860358593602145,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.002564822690584638,
+      "loss": 1.2515,
+      "step": 3630
+    },
+    {
+      "epoch": 0.31869135551892397,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0025645107564164476,
+      "loss": 1.2398,
+      "step": 3631
+    },
+    {
+      "epoch": 0.31877912510182643,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.0025641987319896644,
+      "loss": 1.2943,
+      "step": 3632
+    },
+    {
+      "epoch": 0.3188668946847289,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.0025638866173350837,
+      "loss": 1.2504,
+      "step": 3633
+    },
+    {
+      "epoch": 0.3189546642676314,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.002563574412483509,
+      "loss": 1.2551,
+      "step": 3634
+    },
+    {
+      "epoch": 0.3190424338505339,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.0025632621174657556,
+      "loss": 1.21,
+      "step": 3635
+    },
+    {
+      "epoch": 0.31913020343343634,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0025629497323126447,
+      "loss": 1.2433,
+      "step": 3636
+    },
+    {
+      "epoch": 0.31921797301633886,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0025626372570550075,
+      "loss": 1.2733,
+      "step": 3637
+    },
+    {
+      "epoch": 0.3193057425992413,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.002562324691723684,
+      "loss": 1.2792,
+      "step": 3638
+    },
+    {
+      "epoch": 0.31939351218214385,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0025620120363495226,
+      "loss": 1.2842,
+      "step": 3639
+    },
+    {
+      "epoch": 0.3194812817650463,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.002561699290963383,
+      "loss": 1.23,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3195690513479488,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0025613864555961304,
+      "loss": 1.221,
+      "step": 3641
+    },
+    {
+      "epoch": 0.3196568209308513,
+      "grad_norm": 0.06884765625,
+      "learning_rate": 0.002561073530278641,
+      "loss": 1.292,
+      "step": 3642
+    },
+    {
+      "epoch": 0.31974459051375376,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0025607605150417987,
+      "loss": 1.2336,
+      "step": 3643
+    },
+    {
+      "epoch": 0.3198323600966563,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.002560447409916497,
+      "loss": 1.2731,
+      "step": 3644
+    },
+    {
+      "epoch": 0.31992012967955874,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0025601342149336394,
+      "loss": 1.2563,
+      "step": 3645
+    },
+    {
+      "epoch": 0.3200078992624612,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0025598209301241352,
+      "loss": 1.2311,
+      "step": 3646
+    },
+    {
+      "epoch": 0.3200956688453637,
+      "grad_norm": 0.0751953125,
+      "learning_rate": 0.002559507555518906,
+      "loss": 1.2805,
+      "step": 3647
+    },
+    {
+      "epoch": 0.3201834384282662,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.002559194091148879,
+      "loss": 1.2659,
+      "step": 3648
+    },
+    {
+      "epoch": 0.3202712080111687,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.0025588805370449926,
+      "loss": 1.3416,
+      "step": 3649
+    },
+    {
+      "epoch": 0.32035897759407117,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.002558566893238194,
+      "loss": 1.2211,
+      "step": 3650
+    },
+    {
+      "epoch": 0.32044674717697363,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.002558253159759438,
+      "loss": 1.2193,
+      "step": 3651
+    },
+    {
+      "epoch": 0.32053451675987615,
+      "grad_norm": 0.056884765625,
+      "learning_rate": 0.0025579393366396887,
+      "loss": 1.264,
+      "step": 3652
+    },
+    {
+      "epoch": 0.3206222863427786,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.0025576254239099193,
+      "loss": 1.3171,
+      "step": 3653
+    },
+    {
+      "epoch": 0.3207100559256811,
+      "grad_norm": 0.060791015625,
+      "learning_rate": 0.002557311421601112,
+      "loss": 1.3002,
+      "step": 3654
+    },
+    {
+      "epoch": 0.3207978255085836,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.0025569973297442576,
+      "loss": 1.2464,
+      "step": 3655
+    },
+    {
+      "epoch": 0.32088559509148606,
+      "grad_norm": 0.068359375,
+      "learning_rate": 0.0025566831483703556,
+      "loss": 1.1842,
+      "step": 3656
+    },
+    {
+      "epoch": 0.3209733646743886,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.002556368877510414,
+      "loss": 1.277,
+      "step": 3657
+    },
+    {
+      "epoch": 0.32106113425729105,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.002556054517195451,
+      "loss": 1.3148,
+      "step": 3658
+    },
+    {
+      "epoch": 0.3211489038401935,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.002555740067456492,
+      "loss": 1.2964,
+      "step": 3659
+    },
+    {
+      "epoch": 0.32123667342309603,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.002555425528324572,
+      "loss": 1.269,
+      "step": 3660
+    },
+    {
+      "epoch": 0.3213244430059985,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.0025551108998307353,
+      "loss": 1.2718,
+      "step": 3661
+    },
+    {
+      "epoch": 0.321412212588901,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0025547961820060337,
+      "loss": 1.2604,
+      "step": 3662
+    },
+    {
+      "epoch": 0.3214999821718035,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.00255448137488153,
+      "loss": 1.2569,
+      "step": 3663
+    },
+    {
+      "epoch": 0.32158775175470594,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002554166478488292,
+      "loss": 1.2961,
+      "step": 3664
+    },
+    {
+      "epoch": 0.32167552133760846,
+      "grad_norm": 0.060791015625,
+      "learning_rate": 0.002553851492857401,
+      "loss": 1.2614,
+      "step": 3665
+    },
+    {
+      "epoch": 0.3217632909205109,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.002553536418019944,
+      "loss": 1.219,
+      "step": 3666
+    },
+    {
+      "epoch": 0.3218510605034134,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0025532212540070173,
+      "loss": 1.2551,
+      "step": 3667
+    },
+    {
+      "epoch": 0.3219388300863159,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.002552906000849727,
+      "loss": 1.2483,
+      "step": 3668
+    },
+    {
+      "epoch": 0.32202659966921837,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0025525906585791873,
+      "loss": 1.2635,
+      "step": 3669
+    },
+    {
+      "epoch": 0.3221143692521209,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0025522752272265207,
+      "loss": 1.2612,
+      "step": 3670
+    },
+    {
+      "epoch": 0.32220213883502336,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0025519597068228593,
+      "loss": 1.2294,
+      "step": 3671
+    },
+    {
+      "epoch": 0.3222899084179258,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0025516440973993434,
+      "loss": 1.2335,
+      "step": 3672
+    },
+    {
+      "epoch": 0.32237767800082834,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.002551328398987123,
+      "loss": 1.2709,
+      "step": 3673
+    },
+    {
+      "epoch": 0.3224654475837308,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0025510126116173562,
+      "loss": 1.1987,
+      "step": 3674
+    },
+    {
+      "epoch": 0.3225532171666333,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0025506967353212096,
+      "loss": 1.2795,
+      "step": 3675
+    },
+    {
+      "epoch": 0.3226409867495358,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.0025503807701298593,
+      "loss": 1.2423,
+      "step": 3676
+    },
+    {
+      "epoch": 0.32272875633243825,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0025500647160744895,
+      "loss": 1.316,
+      "step": 3677
+    },
+    {
+      "epoch": 0.32281652591534077,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0025497485731862933,
+      "loss": 1.3196,
+      "step": 3678
+    },
+    {
+      "epoch": 0.32290429549824323,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.002549432341496473,
+      "loss": 1.2743,
+      "step": 3679
+    },
+    {
+      "epoch": 0.3229920650811457,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0025491160210362397,
+      "loss": 1.2443,
+      "step": 3680
+    },
+    {
+      "epoch": 0.3230798346640482,
+      "grad_norm": 0.0634765625,
+      "learning_rate": 0.002548799611836813,
+      "loss": 1.2591,
+      "step": 3681
+    },
+    {
+      "epoch": 0.3231676042469507,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.002548483113929421,
+      "loss": 1.2808,
+      "step": 3682
+    },
+    {
+      "epoch": 0.3232553738298532,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0025481665273453004,
+      "loss": 1.2841,
+      "step": 3683
+    },
+    {
+      "epoch": 0.32334314341275566,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.002547849852115697,
+      "loss": 1.2251,
+      "step": 3684
+    },
+    {
+      "epoch": 0.3234309129956581,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.002547533088271866,
+      "loss": 1.2241,
+      "step": 3685
+    },
+    {
+      "epoch": 0.32351868257856065,
+      "grad_norm": 0.061767578125,
+      "learning_rate": 0.0025472162358450715,
+      "loss": 1.2915,
+      "step": 3686
+    },
+    {
+      "epoch": 0.3236064521614631,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.002546899294866584,
+      "loss": 1.2427,
+      "step": 3687
+    },
+    {
+      "epoch": 0.32369422174436563,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.002546582265367685,
+      "loss": 1.2149,
+      "step": 3688
+    },
+    {
+      "epoch": 0.3237819913272681,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002546265147379664,
+      "loss": 1.2542,
+      "step": 3689
+    },
+    {
+      "epoch": 0.32386976091017056,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0025459479409338194,
+      "loss": 1.2619,
+      "step": 3690
+    },
+    {
+      "epoch": 0.3239575304930731,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0025456306460614575,
+      "loss": 1.3024,
+      "step": 3691
+    },
+    {
+      "epoch": 0.32404530007597554,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002545313262793895,
+      "loss": 1.2757,
+      "step": 3692
+    },
+    {
+      "epoch": 0.324133069658878,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.0025449957911624563,
+      "loss": 1.2264,
+      "step": 3693
+    },
+    {
+      "epoch": 0.3242208392417805,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0025446782311984747,
+      "loss": 1.3051,
+      "step": 3694
+    },
+    {
+      "epoch": 0.324308608824683,
+      "grad_norm": 0.06982421875,
+      "learning_rate": 0.0025443605829332905,
+      "loss": 1.2478,
+      "step": 3695
+    },
+    {
+      "epoch": 0.3243963784075855,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.002544042846398257,
+      "loss": 1.2214,
+      "step": 3696
+    },
+    {
+      "epoch": 0.32448414799048797,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0025437250216247312,
+      "loss": 1.2114,
+      "step": 3697
+    },
+    {
+      "epoch": 0.32457191757339043,
+      "grad_norm": 0.068359375,
+      "learning_rate": 0.002543407108644082,
+      "loss": 1.2714,
+      "step": 3698
+    },
+    {
+      "epoch": 0.32465968715629295,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.002543089107487687,
+      "loss": 1.2723,
+      "step": 3699
+    },
+    {
+      "epoch": 0.3247474567391954,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00254277101818693,
+      "loss": 1.2854,
+      "step": 3700
+    },
+    {
+      "epoch": 0.32483522632209794,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0025424528407732064,
+      "loss": 1.2899,
+      "step": 3701
+    },
+    {
+      "epoch": 0.3249229959050004,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0025421345752779188,
+      "loss": 1.2391,
+      "step": 3702
+    },
+    {
+      "epoch": 0.32501076548790286,
+      "grad_norm": 0.060791015625,
+      "learning_rate": 0.0025418162217324784,
+      "loss": 1.251,
+      "step": 3703
+    },
+    {
+      "epoch": 0.3250985350708054,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.002541497780168306,
+      "loss": 1.2149,
+      "step": 3704
+    },
+    {
+      "epoch": 0.32518630465370785,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.0025411792506168296,
+      "loss": 1.2652,
+      "step": 3705
+    },
+    {
+      "epoch": 0.3252740742366103,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0025408606331094874,
+      "loss": 1.258,
+      "step": 3706
+    },
+    {
+      "epoch": 0.32536184381951283,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.002540541927677726,
+      "loss": 1.2684,
+      "step": 3707
+    },
+    {
+      "epoch": 0.3254496134024153,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0025402231343530004,
+      "loss": 1.2957,
+      "step": 3708
+    },
+    {
+      "epoch": 0.3255373829853178,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.0025399042531667735,
+      "loss": 1.293,
+      "step": 3709
+    },
+    {
+      "epoch": 0.3256251525682203,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0025395852841505182,
+      "loss": 1.2639,
+      "step": 3710
+    },
+    {
+      "epoch": 0.32571292215112274,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.0025392662273357146,
+      "loss": 1.2412,
+      "step": 3711
+    },
+    {
+      "epoch": 0.32580069173402526,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0025389470827538534,
+      "loss": 1.2482,
+      "step": 3712
+    },
+    {
+      "epoch": 0.3258884613169277,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0025386278504364325,
+      "loss": 1.225,
+      "step": 3713
+    },
+    {
+      "epoch": 0.32597623089983024,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0025383085304149593,
+      "loss": 1.2702,
+      "step": 3714
+    },
+    {
+      "epoch": 0.3260640004827327,
+      "grad_norm": 0.06591796875,
+      "learning_rate": 0.002537989122720949,
+      "loss": 1.2097,
+      "step": 3715
+    },
+    {
+      "epoch": 0.32615177006563517,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.002537669627385925,
+      "loss": 1.2717,
+      "step": 3716
+    },
+    {
+      "epoch": 0.3262395396485377,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.0025373500444414217,
+      "loss": 1.2309,
+      "step": 3717
+    },
+    {
+      "epoch": 0.32632730923144015,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.00253703037391898,
+      "loss": 1.2578,
+      "step": 3718
+    },
+    {
+      "epoch": 0.3264150788143427,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.00253671061585015,
+      "loss": 1.3245,
+      "step": 3719
+    },
+    {
+      "epoch": 0.32650284839724514,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002536390770266491,
+      "loss": 1.2825,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3265906179801476,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.00253607083719957,
+      "loss": 1.3416,
+      "step": 3721
+    },
+    {
+      "epoch": 0.3266783875630501,
+      "grad_norm": 0.06103515625,
+      "learning_rate": 0.002535750816680964,
+      "loss": 1.1654,
+      "step": 3722
+    },
+    {
+      "epoch": 0.3267661571459526,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0025354307087422567,
+      "loss": 1.2169,
+      "step": 3723
+    },
+    {
+      "epoch": 0.32685392672885505,
+      "grad_norm": 0.0625,
+      "learning_rate": 0.002535110513415042,
+      "loss": 1.2615,
+      "step": 3724
+    },
+    {
+      "epoch": 0.32694169631175757,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0025347902307309217,
+      "loss": 1.2649,
+      "step": 3725
+    },
+    {
+      "epoch": 0.32702946589466003,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.002534469860721507,
+      "loss": 1.2465,
+      "step": 3726
+    },
+    {
+      "epoch": 0.32711723547756255,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.002534149403418416,
+      "loss": 1.214,
+      "step": 3727
+    },
+    {
+      "epoch": 0.327205005060465,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0025338288588532777,
+      "loss": 1.2594,
+      "step": 3728
+    },
+    {
+      "epoch": 0.3272927746433675,
+      "grad_norm": 0.0634765625,
+      "learning_rate": 0.0025335082270577284,
+      "loss": 1.2803,
+      "step": 3729
+    },
+    {
+      "epoch": 0.32738054422627,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0025331875080634126,
+      "loss": 1.2449,
+      "step": 3730
+    },
+    {
+      "epoch": 0.32746831380917246,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.0025328667019019844,
+      "loss": 1.2621,
+      "step": 3731
+    },
+    {
+      "epoch": 0.327556083392075,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002532545808605106,
+      "loss": 1.203,
+      "step": 3732
+    },
+    {
+      "epoch": 0.32764385297497745,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002532224828204448,
+      "loss": 1.2359,
+      "step": 3733
+    },
+    {
+      "epoch": 0.3277316225578799,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0025319037607316915,
+      "loss": 1.1992,
+      "step": 3734
+    },
+    {
+      "epoch": 0.32781939214078243,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0025315826062185217,
+      "loss": 1.3304,
+      "step": 3735
+    },
+    {
+      "epoch": 0.3279071617236849,
+      "grad_norm": 0.06103515625,
+      "learning_rate": 0.0025312613646966383,
+      "loss": 1.2537,
+      "step": 3736
+    },
+    {
+      "epoch": 0.32799493130658736,
+      "grad_norm": 0.125,
+      "learning_rate": 0.002530940036197744,
+      "loss": 1.2027,
+      "step": 3737
+    },
+    {
+      "epoch": 0.3280827008894899,
+      "grad_norm": 0.0634765625,
+      "learning_rate": 0.002530618620753555,
+      "loss": 1.2073,
+      "step": 3738
+    },
+    {
+      "epoch": 0.32817047047239234,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002530297118395792,
+      "loss": 1.2433,
+      "step": 3739
+    },
+    {
+      "epoch": 0.32825824005529486,
+      "grad_norm": 0.05859375,
+      "learning_rate": 0.0025299755291561866,
+      "loss": 1.2212,
+      "step": 3740
+    },
+    {
+      "epoch": 0.3283460096381973,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.002529653853066478,
+      "loss": 1.222,
+      "step": 3741
+    },
+    {
+      "epoch": 0.3284337792210998,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.002529332090158415,
+      "loss": 1.2265,
+      "step": 3742
+    },
+    {
+      "epoch": 0.3285215488040023,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0025290102404637543,
+      "loss": 1.1789,
+      "step": 3743
+    },
+    {
+      "epoch": 0.32860931838690477,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0025286883040142607,
+      "loss": 1.2608,
+      "step": 3744
+    },
+    {
+      "epoch": 0.3286970879698073,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.002528366280841708,
+      "loss": 1.287,
+      "step": 3745
+    },
+    {
+      "epoch": 0.32878485755270975,
+      "grad_norm": 0.060791015625,
+      "learning_rate": 0.0025280441709778795,
+      "loss": 1.2626,
+      "step": 3746
+    },
+    {
+      "epoch": 0.3288726271356122,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002527721974454565,
+      "loss": 1.2261,
+      "step": 3747
+    },
+    {
+      "epoch": 0.32896039671851474,
+      "grad_norm": 0.056396484375,
+      "learning_rate": 0.0025273996913035652,
+      "loss": 1.2195,
+      "step": 3748
+    },
+    {
+      "epoch": 0.3290481663014172,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0025270773215566875,
+      "loss": 1.2956,
+      "step": 3749
+    },
+    {
+      "epoch": 0.32913593588431966,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.002526754865245748,
+      "loss": 1.2899,
+      "step": 3750
+    },
+    {
+      "epoch": 0.3292237054672222,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.002526432322402573,
+      "loss": 1.2792,
+      "step": 3751
+    },
+    {
+      "epoch": 0.32931147505012465,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.002526109693058995,
+      "loss": 1.2438,
+      "step": 3752
+    },
+    {
+      "epoch": 0.32939924463302717,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.0025257869772468576,
+      "loss": 1.3401,
+      "step": 3753
+    },
+    {
+      "epoch": 0.32948701421592963,
+      "grad_norm": 0.060302734375,
+      "learning_rate": 0.0025254641749980107,
+      "loss": 1.3386,
+      "step": 3754
+    },
+    {
+      "epoch": 0.3295747837988321,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.0025251412863443136,
+      "loss": 1.2555,
+      "step": 3755
+    },
+    {
+      "epoch": 0.3296625533817346,
+      "grad_norm": 0.055419921875,
+      "learning_rate": 0.002524818311317634,
+      "loss": 1.2637,
+      "step": 3756
+    },
+    {
+      "epoch": 0.3297503229646371,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.002524495249949849,
+      "loss": 1.2575,
+      "step": 3757
+    },
+    {
+      "epoch": 0.3298380925475396,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.002524172102272843,
+      "loss": 1.2357,
+      "step": 3758
+    },
+    {
+      "epoch": 0.32992586213044206,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.0025238488683185095,
+      "loss": 1.238,
+      "step": 3759
+    },
+    {
+      "epoch": 0.3300136317133445,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.00252352554811875,
+      "loss": 1.2975,
+      "step": 3760
+    },
+    {
+      "epoch": 0.33010140129624704,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.002523202141705475,
+      "loss": 1.2368,
+      "step": 3761
+    },
+    {
+      "epoch": 0.3301891708791495,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0025228786491106043,
+      "loss": 1.2287,
+      "step": 3762
+    },
+    {
+      "epoch": 0.33027694046205197,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0025225550703660646,
+      "loss": 1.2193,
+      "step": 3763
+    },
+    {
+      "epoch": 0.3303647100449545,
+      "grad_norm": 0.0615234375,
+      "learning_rate": 0.002522231405503792,
+      "loss": 1.2226,
+      "step": 3764
+    },
+    {
+      "epoch": 0.33045247962785695,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.002521907654555731,
+      "loss": 1.3005,
+      "step": 3765
+    },
+    {
+      "epoch": 0.3305402492107595,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0025215838175538345,
+      "loss": 1.2229,
+      "step": 3766
+    },
+    {
+      "epoch": 0.33062801879366194,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002521259894530064,
+      "loss": 1.262,
+      "step": 3767
+    },
+    {
+      "epoch": 0.3307157883765644,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.0025209358855163887,
+      "loss": 1.2642,
+      "step": 3768
+    },
+    {
+      "epoch": 0.3308035579594669,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.002520611790544788,
+      "loss": 1.2599,
+      "step": 3769
+    },
+    {
+      "epoch": 0.3308913275423694,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.0025202876096472484,
+      "loss": 1.2076,
+      "step": 3770
+    },
+    {
+      "epoch": 0.3309790971252719,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002519963342855765,
+      "loss": 1.2606,
+      "step": 3771
+    },
+    {
+      "epoch": 0.33106686670817437,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0025196389902023427,
+      "loss": 1.2664,
+      "step": 3772
+    },
+    {
+      "epoch": 0.33115463629107683,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0025193145517189925,
+      "loss": 1.2137,
+      "step": 3773
+    },
+    {
+      "epoch": 0.33124240587397935,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0025189900274377357,
+      "loss": 1.2974,
+      "step": 3774
+    },
+    {
+      "epoch": 0.3313301754568818,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0025186654173906015,
+      "loss": 1.3374,
+      "step": 3775
+    },
+    {
+      "epoch": 0.3314179450397843,
+      "grad_norm": 0.125,
+      "learning_rate": 0.002518340721609628,
+      "loss": 1.2484,
+      "step": 3776
+    },
+    {
+      "epoch": 0.3315057146226868,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.002518015940126861,
+      "loss": 1.3006,
+      "step": 3777
+    },
+    {
+      "epoch": 0.33159348420558926,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0025176910729743556,
+      "loss": 1.2739,
+      "step": 3778
+    },
+    {
+      "epoch": 0.3316812537884918,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.0025173661201841743,
+      "loss": 1.2801,
+      "step": 3779
+    },
+    {
+      "epoch": 0.33176902337139424,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0025170410817883892,
+      "loss": 1.2251,
+      "step": 3780
+    },
+    {
+      "epoch": 0.3318567929542967,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.0025167159578190794,
+      "loss": 1.2788,
+      "step": 3781
+    },
+    {
+      "epoch": 0.33194456253719923,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.0025163907483083344,
+      "loss": 1.2707,
+      "step": 3782
+    },
+    {
+      "epoch": 0.3320323321201017,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0025160654532882507,
+      "loss": 1.2502,
+      "step": 3783
+    },
+    {
+      "epoch": 0.3321201017030042,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0025157400727909337,
+      "loss": 1.2705,
+      "step": 3784
+    },
+    {
+      "epoch": 0.3322078712859067,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.002515414606848497,
+      "loss": 1.2243,
+      "step": 3785
+    },
+    {
+      "epoch": 0.33229564086880914,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0025150890554930627,
+      "loss": 1.2388,
+      "step": 3786
+    },
+    {
+      "epoch": 0.33238341045171166,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.0025147634187567616,
+      "loss": 1.1998,
+      "step": 3787
+    },
+    {
+      "epoch": 0.3324711800346141,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0025144376966717333,
+      "loss": 1.2723,
+      "step": 3788
+    },
+    {
+      "epoch": 0.33255894961751664,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0025141118892701244,
+      "loss": 1.2278,
+      "step": 3789
+    },
+    {
+      "epoch": 0.3326467192004191,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002513785996584091,
+      "loss": 1.2397,
+      "step": 3790
+    },
+    {
+      "epoch": 0.33273448878332157,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.0025134600186457975,
+      "loss": 1.199,
+      "step": 3791
+    },
+    {
+      "epoch": 0.3328222583662241,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0025131339554874173,
+      "loss": 1.2355,
+      "step": 3792
+    },
+    {
+      "epoch": 0.33291002794912655,
+      "grad_norm": 0.060546875,
+      "learning_rate": 0.00251280780714113,
+      "loss": 1.2621,
+      "step": 3793
+    },
+    {
+      "epoch": 0.332997797532029,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.002512481573639127,
+      "loss": 1.3712,
+      "step": 3794
+    },
+    {
+      "epoch": 0.33308556711493154,
+      "grad_norm": 0.0625,
+      "learning_rate": 0.0025121552550136043,
+      "loss": 1.2313,
+      "step": 3795
+    },
+    {
+      "epoch": 0.333173336697834,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0025118288512967707,
+      "loss": 1.2302,
+      "step": 3796
+    },
+    {
+      "epoch": 0.3332611062807365,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.0025115023625208385,
+      "loss": 1.2087,
+      "step": 3797
+    },
+    {
+      "epoch": 0.333348875863639,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0025111757887180323,
+      "loss": 1.3146,
+      "step": 3798
+    },
+    {
+      "epoch": 0.33343664544654145,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.0025108491299205833,
+      "loss": 1.2671,
+      "step": 3799
+    },
+    {
+      "epoch": 0.33352441502944397,
+      "grad_norm": 0.078125,
+      "learning_rate": 0.0025105223861607313,
+      "loss": 1.2379,
+      "step": 3800
+    },
+    {
+      "epoch": 0.33361218461234643,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0025101955574707248,
+      "loss": 1.2303,
+      "step": 3801
+    },
+    {
+      "epoch": 0.33369995419524895,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.0025098686438828205,
+      "loss": 1.3177,
+      "step": 3802
+    },
+    {
+      "epoch": 0.3337877237781514,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.002509541645429283,
+      "loss": 1.3069,
+      "step": 3803
+    },
+    {
+      "epoch": 0.3338754933610539,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.0025092145621423857,
+      "loss": 1.2034,
+      "step": 3804
+    },
+    {
+      "epoch": 0.3339632629439564,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002508887394054412,
+      "loss": 1.2386,
+      "step": 3805
+    },
+    {
+      "epoch": 0.33405103252685886,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.0025085601411976503,
+      "loss": 1.2443,
+      "step": 3806
+    },
+    {
+      "epoch": 0.3341388021097613,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.0025082328036043996,
+      "loss": 1.2725,
+      "step": 3807
+    },
+    {
+      "epoch": 0.33422657169266384,
+      "grad_norm": 0.059814453125,
+      "learning_rate": 0.0025079053813069677,
+      "loss": 1.2294,
+      "step": 3808
+    },
+    {
+      "epoch": 0.3343143412755663,
+      "grad_norm": 0.0634765625,
+      "learning_rate": 0.0025075778743376687,
+      "loss": 1.1634,
+      "step": 3809
+    },
+    {
+      "epoch": 0.3344021108584688,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.0025072502827288273,
+      "loss": 1.2462,
+      "step": 3810
+    },
+    {
+      "epoch": 0.3344898804413713,
+      "grad_norm": 0.062255859375,
+      "learning_rate": 0.0025069226065127744,
+      "loss": 1.2371,
+      "step": 3811
+    },
+    {
+      "epoch": 0.33457765002427375,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0025065948457218506,
+      "loss": 1.3102,
+      "step": 3812
+    },
+    {
+      "epoch": 0.3346654196071763,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.002506267000388406,
+      "loss": 1.216,
+      "step": 3813
+    },
+    {
+      "epoch": 0.33475318919007874,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0025059390705447957,
+      "loss": 1.2178,
+      "step": 3814
+    },
+    {
+      "epoch": 0.33484095877298126,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.0025056110562233862,
+      "loss": 1.261,
+      "step": 3815
+    },
+    {
+      "epoch": 0.3349287283558837,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.0025052829574565513,
+      "loss": 1.2298,
+      "step": 3816
+    },
+    {
+      "epoch": 0.3350164979387862,
+      "grad_norm": 0.0546875,
+      "learning_rate": 0.002504954774276672,
+      "loss": 1.262,
+      "step": 3817
+    },
+    {
+      "epoch": 0.3351042675216887,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0025046265067161398,
+      "loss": 1.232,
+      "step": 3818
+    },
+    {
+      "epoch": 0.33519203710459117,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.002504298154807353,
+      "loss": 1.2552,
+      "step": 3819
+    },
+    {
+      "epoch": 0.33527980668749363,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.002503969718582718,
+      "loss": 1.1982,
+      "step": 3820
+    },
+    {
+      "epoch": 0.33536757627039615,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0025036411980746513,
+      "loss": 1.2486,
+      "step": 3821
+    },
+    {
+      "epoch": 0.3354553458532986,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0025033125933155756,
+      "loss": 1.2,
+      "step": 3822
+    },
+    {
+      "epoch": 0.33554311543620113,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.002502983904337924,
+      "loss": 1.2906,
+      "step": 3823
+    },
+    {
+      "epoch": 0.3356308850191036,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0025026551311741357,
+      "loss": 1.2372,
+      "step": 3824
+    },
+    {
+      "epoch": 0.33571865460200606,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.00250232627385666,
+      "loss": 1.2027,
+      "step": 3825
+    },
+    {
+      "epoch": 0.3358064241849086,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.0025019973324179536,
+      "loss": 1.2718,
+      "step": 3826
+    },
+    {
+      "epoch": 0.33589419376781104,
+      "grad_norm": 0.05517578125,
+      "learning_rate": 0.0025016683068904816,
+      "loss": 1.1979,
+      "step": 3827
+    },
+    {
+      "epoch": 0.33598196335071356,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.002501339197306717,
+      "loss": 1.2185,
+      "step": 3828
+    },
+    {
+      "epoch": 0.336069732933616,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0025010100036991427,
+      "loss": 1.2562,
+      "step": 3829
+    },
+    {
+      "epoch": 0.3361575025165185,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0025006807261002484,
+      "loss": 1.1998,
+      "step": 3830
+    },
+    {
+      "epoch": 0.336245272099421,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.002500351364542532,
+      "loss": 1.2845,
+      "step": 3831
+    },
+    {
+      "epoch": 0.3363330416823235,
+      "grad_norm": 0.0712890625,
+      "learning_rate": 0.002500021919058501,
+      "loss": 1.3241,
+      "step": 3832
+    },
+    {
+      "epoch": 0.33642081126522594,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.00249969238968067,
+      "loss": 1.2449,
+      "step": 3833
+    },
+    {
+      "epoch": 0.33650858084812846,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002499362776441562,
+      "loss": 1.3042,
+      "step": 3834
+    },
+    {
+      "epoch": 0.3365963504310309,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0024990330793737093,
+      "loss": 1.2679,
+      "step": 3835
+    },
+    {
+      "epoch": 0.33668412001393344,
+      "grad_norm": 0.05908203125,
+      "learning_rate": 0.00249870329850965,
+      "loss": 1.246,
+      "step": 3836
+    },
+    {
+      "epoch": 0.3367718895968359,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0024983734338819338,
+      "loss": 1.2822,
+      "step": 3837
+    },
+    {
+      "epoch": 0.33685965917973837,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0024980434855231166,
+      "loss": 1.178,
+      "step": 3838
+    },
+    {
+      "epoch": 0.3369474287626409,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.002497713453465763,
+      "loss": 1.3513,
+      "step": 3839
+    },
+    {
+      "epoch": 0.33703519834554335,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0024973833377424457,
+      "loss": 1.198,
+      "step": 3840
+    },
+    {
+      "epoch": 0.33712296792844587,
+      "grad_norm": 0.072265625,
+      "learning_rate": 0.002497053138385746,
+      "loss": 1.2593,
+      "step": 3841
+    },
+    {
+      "epoch": 0.33721073751134834,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.002496722855428253,
+      "loss": 1.2104,
+      "step": 3842
+    },
+    {
+      "epoch": 0.3372985070942508,
+      "grad_norm": 0.060546875,
+      "learning_rate": 0.0024963924889025648,
+      "loss": 1.2327,
+      "step": 3843
+    },
+    {
+      "epoch": 0.3373862766771533,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0024960620388412864,
+      "loss": 1.2432,
+      "step": 3844
+    },
+    {
+      "epoch": 0.3374740462600558,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.0024957315052770326,
+      "loss": 1.2453,
+      "step": 3845
+    },
+    {
+      "epoch": 0.33756181584295825,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0024954008882424254,
+      "loss": 1.2313,
+      "step": 3846
+    },
+    {
+      "epoch": 0.33764958542586077,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.002495070187770096,
+      "loss": 1.2881,
+      "step": 3847
+    },
+    {
+      "epoch": 0.33773735500876323,
+      "grad_norm": 0.076171875,
+      "learning_rate": 0.0024947394038926833,
+      "loss": 1.2224,
+      "step": 3848
+    },
+    {
+      "epoch": 0.33782512459166575,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0024944085366428335,
+      "loss": 1.2479,
+      "step": 3849
+    },
+    {
+      "epoch": 0.3379128941745682,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.0024940775860532024,
+      "loss": 1.2176,
+      "step": 3850
+    },
+    {
+      "epoch": 0.3380006637574707,
+      "grad_norm": 0.060302734375,
+      "learning_rate": 0.0024937465521564527,
+      "loss": 1.2383,
+      "step": 3851
+    },
+    {
+      "epoch": 0.3380884333403732,
+      "grad_norm": 0.061767578125,
+      "learning_rate": 0.0024934154349852576,
+      "loss": 1.2914,
+      "step": 3852
+    },
+    {
+      "epoch": 0.33817620292327566,
+      "grad_norm": 0.0625,
+      "learning_rate": 0.002493084234572296,
+      "loss": 1.2776,
+      "step": 3853
+    },
+    {
+      "epoch": 0.3382639725061782,
+      "grad_norm": 0.05712890625,
+      "learning_rate": 0.002492752950950256,
+      "loss": 1.2459,
+      "step": 3854
+    },
+    {
+      "epoch": 0.33835174208908064,
+      "grad_norm": 0.058349609375,
+      "learning_rate": 0.0024924215841518353,
+      "loss": 1.2197,
+      "step": 3855
+    },
+    {
+      "epoch": 0.3384395116719831,
+      "grad_norm": 0.068359375,
+      "learning_rate": 0.002492090134209737,
+      "loss": 1.276,
+      "step": 3856
+    },
+    {
+      "epoch": 0.3385272812548856,
+      "grad_norm": 0.06787109375,
+      "learning_rate": 0.002491758601156674,
+      "loss": 1.2357,
+      "step": 3857
+    },
+    {
+      "epoch": 0.3386150508377881,
+      "grad_norm": 0.0615234375,
+      "learning_rate": 0.0024914269850253683,
+      "loss": 1.21,
+      "step": 3858
+    },
+    {
+      "epoch": 0.3387028204206906,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.0024910952858485484,
+      "loss": 1.2655,
+      "step": 3859
+    },
+    {
+      "epoch": 0.3387905900035931,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.002490763503658952,
+      "loss": 1.2379,
+      "step": 3860
+    },
+    {
+      "epoch": 0.33887835958649554,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0024904316384893243,
+      "loss": 1.2728,
+      "step": 3861
+    },
+    {
+      "epoch": 0.33896612916939806,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0024900996903724193,
+      "loss": 1.2496,
+      "step": 3862
+    },
+    {
+      "epoch": 0.3390538987523005,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0024897676593409993,
+      "loss": 1.2273,
+      "step": 3863
+    },
+    {
+      "epoch": 0.339141668335203,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.0024894355454278335,
+      "loss": 1.2697,
+      "step": 3864
+    },
+    {
+      "epoch": 0.3392294379181055,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.0024891033486657012,
+      "loss": 1.282,
+      "step": 3865
+    },
+    {
+      "epoch": 0.33931720750100797,
+      "grad_norm": 0.06689453125,
+      "learning_rate": 0.0024887710690873885,
+      "loss": 1.2563,
+      "step": 3866
+    },
+    {
+      "epoch": 0.3394049770839105,
+      "grad_norm": 0.0634765625,
+      "learning_rate": 0.0024884387067256905,
+      "loss": 1.1868,
+      "step": 3867
+    },
+    {
+      "epoch": 0.33949274666681295,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.0024881062616134092,
+      "loss": 1.1902,
+      "step": 3868
+    },
+    {
+      "epoch": 0.3395805162497154,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0024877737337833565,
+      "loss": 1.2901,
+      "step": 3869
+    },
+    {
+      "epoch": 0.33966828583261793,
+      "grad_norm": 0.059326171875,
+      "learning_rate": 0.0024874411232683508,
+      "loss": 1.2237,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3397560554155204,
+      "grad_norm": 0.0625,
+      "learning_rate": 0.00248710843010122,
+      "loss": 1.2257,
+      "step": 3871
+    },
+    {
+      "epoch": 0.3398438249984229,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.002486775654314799,
+      "loss": 1.2259,
+      "step": 3872
+    },
+    {
+      "epoch": 0.3399315945813254,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0024864427959419323,
+      "loss": 1.2533,
+      "step": 3873
+    },
+    {
+      "epoch": 0.34001936416422784,
+      "grad_norm": 0.060302734375,
+      "learning_rate": 0.002486109855015472,
+      "loss": 1.2246,
+      "step": 3874
+    },
+    {
+      "epoch": 0.34010713374713036,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.0024857768315682763,
+      "loss": 1.265,
+      "step": 3875
+    },
+    {
+      "epoch": 0.3401949033300328,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.0024854437256332145,
+      "loss": 1.2477,
+      "step": 3876
+    },
+    {
+      "epoch": 0.3402826729129353,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0024851105372431623,
+      "loss": 1.2539,
+      "step": 3877
+    },
+    {
+      "epoch": 0.3403704424958378,
+      "grad_norm": 0.06103515625,
+      "learning_rate": 0.0024847772664310054,
+      "loss": 1.2641,
+      "step": 3878
+    },
+    {
+      "epoch": 0.3404582120787403,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.0024844439132296345,
+      "loss": 1.2266,
+      "step": 3879
+    },
+    {
+      "epoch": 0.3405459816616428,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.002484110477671952,
+      "loss": 1.2298,
+      "step": 3880
+    },
+    {
+      "epoch": 0.34063375124454526,
+      "grad_norm": 0.0673828125,
+      "learning_rate": 0.0024837769597908644,
+      "loss": 1.223,
+      "step": 3881
+    },
+    {
+      "epoch": 0.3407215208274477,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.0024834433596192906,
+      "loss": 1.2716,
+      "step": 3882
+    },
+    {
+      "epoch": 0.34080929041035024,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.002483109677190156,
+      "loss": 1.2197,
+      "step": 3883
+    },
+    {
+      "epoch": 0.3408970599932527,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.0024827759125363915,
+      "loss": 1.2683,
+      "step": 3884
+    },
+    {
+      "epoch": 0.3409848295761552,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0024824420656909397,
+      "loss": 1.2819,
+      "step": 3885
+    },
+    {
+      "epoch": 0.3410725991590577,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.00248210813668675,
+      "loss": 1.2713,
+      "step": 3886
+    },
+    {
+      "epoch": 0.34116036874196015,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0024817741255567798,
+      "loss": 1.2244,
+      "step": 3887
+    },
+    {
+      "epoch": 0.34124813832486267,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0024814400323339945,
+      "loss": 1.2013,
+      "step": 3888
+    },
+    {
+      "epoch": 0.34133590790776513,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0024811058570513678,
+      "loss": 1.2884,
+      "step": 3889
+    },
+    {
+      "epoch": 0.3414236774906676,
+      "grad_norm": 0.07373046875,
+      "learning_rate": 0.0024807715997418817,
+      "loss": 1.2264,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3415114470735701,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0024804372604385256,
+      "loss": 1.1952,
+      "step": 3891
+    },
+    {
+      "epoch": 0.3415992166564726,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.002480102839174298,
+      "loss": 1.2594,
+      "step": 3892
+    },
+    {
+      "epoch": 0.3416869862393751,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.002479768335982205,
+      "loss": 1.2732,
+      "step": 3893
+    },
+    {
+      "epoch": 0.34177475582227757,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0024794337508952597,
+      "loss": 1.1971,
+      "step": 3894
+    },
+    {
+      "epoch": 0.34186252540518003,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.002479099083946486,
+      "loss": 1.3004,
+      "step": 3895
+    },
+    {
+      "epoch": 0.34195029498808255,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.002478764335168912,
+      "loss": 1.2729,
+      "step": 3896
+    },
+    {
+      "epoch": 0.342038064570985,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0024784295045955787,
+      "loss": 1.2337,
+      "step": 3897
+    },
+    {
+      "epoch": 0.34212583415388753,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.0024780945922595306,
+      "loss": 1.2409,
+      "step": 3898
+    },
+    {
+      "epoch": 0.34221360373679,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.002477759598193823,
+      "loss": 1.2426,
+      "step": 3899
+    },
+    {
+      "epoch": 0.34230137331969246,
+      "grad_norm": 0.058349609375,
+      "learning_rate": 0.0024774245224315184,
+      "loss": 1.2438,
+      "step": 3900
+    },
+    {
+      "epoch": 0.342389142902595,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.002477089365005687,
+      "loss": 1.2612,
+      "step": 3901
+    },
+    {
+      "epoch": 0.34247691248549744,
+      "grad_norm": 0.059326171875,
+      "learning_rate": 0.0024767541259494087,
+      "loss": 1.2024,
+      "step": 3902
+    },
+    {
+      "epoch": 0.3425646820683999,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.002476418805295768,
+      "loss": 1.201,
+      "step": 3903
+    },
+    {
+      "epoch": 0.3426524516513024,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.0024760834030778628,
+      "loss": 1.2228,
+      "step": 3904
+    },
+    {
+      "epoch": 0.3427402212342049,
+      "grad_norm": 0.06591796875,
+      "learning_rate": 0.002475747919328794,
+      "loss": 1.2355,
+      "step": 3905
+    },
+    {
+      "epoch": 0.3428279908171074,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0024754123540816726,
+      "loss": 1.2358,
+      "step": 3906
+    },
+    {
+      "epoch": 0.3429157604000099,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.002475076707369618,
+      "loss": 1.2443,
+      "step": 3907
+    },
+    {
+      "epoch": 0.34300352998291234,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0024747409792257565,
+      "loss": 1.2954,
+      "step": 3908
+    },
+    {
+      "epoch": 0.34309129956581486,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.0024744051696832243,
+      "loss": 1.2233,
+      "step": 3909
+    },
+    {
+      "epoch": 0.3431790691487173,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0024740692787751633,
+      "loss": 1.2086,
+      "step": 3910
+    },
+    {
+      "epoch": 0.34326683873161984,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.002473733306534726,
+      "loss": 1.2927,
+      "step": 3911
+    },
+    {
+      "epoch": 0.3433546083145223,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0024733972529950698,
+      "loss": 1.2207,
+      "step": 3912
+    },
+    {
+      "epoch": 0.34344237789742477,
+      "grad_norm": 0.0615234375,
+      "learning_rate": 0.0024730611181893636,
+      "loss": 1.2145,
+      "step": 3913
+    },
+    {
+      "epoch": 0.3435301474803273,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.002472724902150781,
+      "loss": 1.2483,
+      "step": 3914
+    },
+    {
+      "epoch": 0.34361791706322975,
+      "grad_norm": 0.057861328125,
+      "learning_rate": 0.0024723886049125058,
+      "loss": 1.2293,
+      "step": 3915
+    },
+    {
+      "epoch": 0.34370568664613227,
+      "grad_norm": 0.064453125,
+      "learning_rate": 0.00247205222650773,
+      "loss": 1.2995,
+      "step": 3916
+    },
+    {
+      "epoch": 0.34379345622903473,
+      "grad_norm": 0.060302734375,
+      "learning_rate": 0.002471715766969652,
+      "loss": 1.2245,
+      "step": 3917
+    },
+    {
+      "epoch": 0.3438812258119372,
+      "grad_norm": 0.061767578125,
+      "learning_rate": 0.002471379226331479,
+      "loss": 1.2291,
+      "step": 3918
+    },
+    {
+      "epoch": 0.3439689953948397,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.0024710426046264263,
+      "loss": 1.2931,
+      "step": 3919
+    },
+    {
+      "epoch": 0.3440567649777422,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.0024707059018877174,
+      "loss": 1.2876,
+      "step": 3920
+    },
+    {
+      "epoch": 0.34414453456064464,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.0024703691181485834,
+      "loss": 1.1831,
+      "step": 3921
+    },
+    {
+      "epoch": 0.34423230414354716,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0024700322534422633,
+      "loss": 1.2332,
+      "step": 3922
+    },
+    {
+      "epoch": 0.3443200737264496,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.002469695307802005,
+      "loss": 1.2615,
+      "step": 3923
+    },
+    {
+      "epoch": 0.34440784330935215,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0024693582812610622,
+      "loss": 1.2372,
+      "step": 3924
+    },
+    {
+      "epoch": 0.3444956128922546,
+      "grad_norm": 0.0732421875,
+      "learning_rate": 0.0024690211738526992,
+      "loss": 1.2727,
+      "step": 3925
+    },
+    {
+      "epoch": 0.3445833824751571,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.002468683985610188,
+      "loss": 1.2459,
+      "step": 3926
+    },
+    {
+      "epoch": 0.3446711520580596,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.002468346716566806,
+      "loss": 1.2815,
+      "step": 3927
+    },
+    {
+      "epoch": 0.34475892164096206,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002468009366755841,
+      "loss": 1.199,
+      "step": 3928
+    },
+    {
+      "epoch": 0.3448466912238646,
+      "grad_norm": 0.06787109375,
+      "learning_rate": 0.002467671936210589,
+      "loss": 1.1999,
+      "step": 3929
+    },
+    {
+      "epoch": 0.34493446080676704,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.0024673344249643517,
+      "loss": 1.2399,
+      "step": 3930
+    },
+    {
+      "epoch": 0.3450222303896695,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002466996833050441,
+      "loss": 1.2403,
+      "step": 3931
+    },
+    {
+      "epoch": 0.345109999972572,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.0024666591605021753,
+      "loss": 1.2722,
+      "step": 3932
+    },
+    {
+      "epoch": 0.3451977695554745,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0024663214073528824,
+      "loss": 1.2229,
+      "step": 3933
+    },
+    {
+      "epoch": 0.34528553913837695,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.002465983573635896,
+      "loss": 1.2623,
+      "step": 3934
+    },
+    {
+      "epoch": 0.34537330872127947,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0024656456593845594,
+      "loss": 1.2396,
+      "step": 3935
+    },
+    {
+      "epoch": 0.34546107830418193,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0024653076646322237,
+      "loss": 1.2818,
+      "step": 3936
+    },
+    {
+      "epoch": 0.34554884788708445,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.0024649695894122478,
+      "loss": 1.1482,
+      "step": 3937
+    },
+    {
+      "epoch": 0.3456366174699869,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.002464631433757998,
+      "loss": 1.2432,
+      "step": 3938
+    },
+    {
+      "epoch": 0.3457243870528894,
+      "grad_norm": 0.080078125,
+      "learning_rate": 0.002464293197702849,
+      "loss": 1.2769,
+      "step": 3939
+    },
+    {
+      "epoch": 0.3458121566357919,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0024639548812801833,
+      "loss": 1.2549,
+      "step": 3940
+    },
+    {
+      "epoch": 0.34589992621869436,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002463616484523392,
+      "loss": 1.2159,
+      "step": 3941
+    },
+    {
+      "epoch": 0.3459876958015969,
+      "grad_norm": 0.07666015625,
+      "learning_rate": 0.002463278007465872,
+      "loss": 1.2639,
+      "step": 3942
+    },
+    {
+      "epoch": 0.34607546538449935,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.0024629394501410316,
+      "loss": 1.2568,
+      "step": 3943
+    },
+    {
+      "epoch": 0.3461632349674018,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.0024626008125822833,
+      "loss": 1.2463,
+      "step": 3944
+    },
+    {
+      "epoch": 0.34625100455030433,
+      "grad_norm": 0.05908203125,
+      "learning_rate": 0.0024622620948230507,
+      "loss": 1.2087,
+      "step": 3945
+    },
+    {
+      "epoch": 0.3463387741332068,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0024619232968967632,
+      "loss": 1.268,
+      "step": 3946
+    },
+    {
+      "epoch": 0.34642654371610926,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0024615844188368586,
+      "loss": 1.2481,
+      "step": 3947
+    },
+    {
+      "epoch": 0.3465143132990118,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.002461245460676783,
+      "loss": 1.3106,
+      "step": 3948
+    },
+    {
+      "epoch": 0.34660208288191424,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0024609064224499913,
+      "loss": 1.3123,
+      "step": 3949
+    },
+    {
+      "epoch": 0.34668985246481676,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0024605673041899435,
+      "loss": 1.1776,
+      "step": 3950
+    },
+    {
+      "epoch": 0.3467776220477192,
+      "grad_norm": 0.068359375,
+      "learning_rate": 0.0024602281059301107,
+      "loss": 1.2278,
+      "step": 3951
+    },
+    {
+      "epoch": 0.3468653916306217,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.002459888827703969,
+      "loss": 1.236,
+      "step": 3952
+    },
+    {
+      "epoch": 0.3469531612135242,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.002459549469545005,
+      "loss": 1.2221,
+      "step": 3953
+    },
+    {
+      "epoch": 0.34704093079642667,
+      "grad_norm": 0.062255859375,
+      "learning_rate": 0.002459210031486711,
+      "loss": 1.2717,
+      "step": 3954
+    },
+    {
+      "epoch": 0.3471287003793292,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0024588705135625897,
+      "loss": 1.258,
+      "step": 3955
+    },
+    {
+      "epoch": 0.34721646996223166,
+      "grad_norm": 0.0595703125,
+      "learning_rate": 0.0024585309158061484,
+      "loss": 1.31,
+      "step": 3956
+    },
+    {
+      "epoch": 0.3473042395451341,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0024581912382509053,
+      "loss": 1.2284,
+      "step": 3957
+    },
+    {
+      "epoch": 0.34739200912803664,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.002457851480930385,
+      "loss": 1.234,
+      "step": 3958
+    },
+    {
+      "epoch": 0.3474797787109391,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.00245751164387812,
+      "loss": 1.2052,
+      "step": 3959
+    },
+    {
+      "epoch": 0.34756754829384157,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.0024571717271276504,
+      "loss": 1.3198,
+      "step": 3960
+    },
+    {
+      "epoch": 0.3476553178767441,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0024568317307125257,
+      "loss": 1.2388,
+      "step": 3961
+    },
+    {
+      "epoch": 0.34774308745964655,
+      "grad_norm": 0.058837890625,
+      "learning_rate": 0.0024564916546663015,
+      "loss": 1.2584,
+      "step": 3962
+    },
+    {
+      "epoch": 0.34783085704254907,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.002456151499022542,
+      "loss": 1.2585,
+      "step": 3963
+    },
+    {
+      "epoch": 0.34791862662545153,
+      "grad_norm": 0.0595703125,
+      "learning_rate": 0.0024558112638148194,
+      "loss": 1.2513,
+      "step": 3964
+    },
+    {
+      "epoch": 0.348006396208354,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002455470949076714,
+      "loss": 1.2528,
+      "step": 3965
+    },
+    {
+      "epoch": 0.3480941657912565,
+      "grad_norm": 0.059814453125,
+      "learning_rate": 0.002455130554841812,
+      "loss": 1.2199,
+      "step": 3966
+    },
+    {
+      "epoch": 0.348181935374159,
+      "grad_norm": 0.0625,
+      "learning_rate": 0.0024547900811437108,
+      "loss": 1.1906,
+      "step": 3967
+    },
+    {
+      "epoch": 0.3482697049570615,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0024544495280160125,
+      "loss": 1.1924,
+      "step": 3968
+    },
+    {
+      "epoch": 0.34835747453996396,
+      "grad_norm": 0.060546875,
+      "learning_rate": 0.002454108895492329,
+      "loss": 1.2129,
+      "step": 3969
+    },
+    {
+      "epoch": 0.3484452441228664,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.0024537681836062793,
+      "loss": 1.2634,
+      "step": 3970
+    },
+    {
+      "epoch": 0.34853301370576895,
+      "grad_norm": 0.059814453125,
+      "learning_rate": 0.00245342739239149,
+      "loss": 1.2644,
+      "step": 3971
+    },
+    {
+      "epoch": 0.3486207832886714,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.0024530865218815963,
+      "loss": 1.2404,
+      "step": 3972
+    },
+    {
+      "epoch": 0.3487085528715739,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.0024527455721102404,
+      "loss": 1.2037,
+      "step": 3973
+    },
+    {
+      "epoch": 0.3487963224544764,
+      "grad_norm": 0.059326171875,
+      "learning_rate": 0.0024524045431110735,
+      "loss": 1.2171,
+      "step": 3974
+    },
+    {
+      "epoch": 0.34888409203737886,
+      "grad_norm": 0.060546875,
+      "learning_rate": 0.002452063434917752,
+      "loss": 1.2733,
+      "step": 3975
+    },
+    {
+      "epoch": 0.3489718616202814,
+      "grad_norm": 0.06298828125,
+      "learning_rate": 0.0024517222475639437,
+      "loss": 1.2401,
+      "step": 3976
+    },
+    {
+      "epoch": 0.34905963120318384,
+      "grad_norm": 0.07080078125,
+      "learning_rate": 0.0024513809810833218,
+      "loss": 1.3198,
+      "step": 3977
+    },
+    {
+      "epoch": 0.3491474007860863,
+      "grad_norm": 0.055419921875,
+      "learning_rate": 0.0024510396355095676,
+      "loss": 1.2371,
+      "step": 3978
+    },
+    {
+      "epoch": 0.3492351703689888,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.0024506982108763718,
+      "loss": 1.2926,
+      "step": 3979
+    },
+    {
+      "epoch": 0.3493229399518913,
+      "grad_norm": 0.054443359375,
+      "learning_rate": 0.00245035670721743,
+      "loss": 1.2289,
+      "step": 3980
+    },
+    {
+      "epoch": 0.3494107095347938,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0024500151245664486,
+      "loss": 1.2357,
+      "step": 3981
+    },
+    {
+      "epoch": 0.34949847911769627,
+      "grad_norm": 0.07177734375,
+      "learning_rate": 0.002449673462957139,
+      "loss": 1.1955,
+      "step": 3982
+    },
+    {
+      "epoch": 0.34958624870059873,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.002449331722423224,
+      "loss": 1.3234,
+      "step": 3983
+    },
+    {
+      "epoch": 0.34967401828350125,
+      "grad_norm": 0.06494140625,
+      "learning_rate": 0.00244898990299843,
+      "loss": 1.264,
+      "step": 3984
+    },
+    {
+      "epoch": 0.3497617878664037,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.002448648004716494,
+      "loss": 1.2428,
+      "step": 3985
+    },
+    {
+      "epoch": 0.34984955744930624,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0024483060276111602,
+      "loss": 1.2085,
+      "step": 3986
+    },
+    {
+      "epoch": 0.3499373270322087,
+      "grad_norm": 0.07763671875,
+      "learning_rate": 0.0024479639717161802,
+      "loss": 1.2003,
+      "step": 3987
+    },
+    {
+      "epoch": 0.35002509661511116,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0024476218370653134,
+      "loss": 1.2714,
+      "step": 3988
+    },
+    {
+      "epoch": 0.3501128661980137,
+      "grad_norm": 0.06396484375,
+      "learning_rate": 0.002447279623692328,
+      "loss": 1.2583,
+      "step": 3989
+    },
+    {
+      "epoch": 0.35020063578091615,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.002446937331630998,
+      "loss": 1.2962,
+      "step": 3990
+    },
+    {
+      "epoch": 0.3502884053638186,
+      "grad_norm": 0.0654296875,
+      "learning_rate": 0.0024465949609151065,
+      "loss": 1.2294,
+      "step": 3991
+    },
+    {
+      "epoch": 0.35037617494672113,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.002446252511578444,
+      "loss": 1.2503,
+      "step": 3992
+    },
+    {
+      "epoch": 0.3504639445296236,
+      "grad_norm": 0.0771484375,
+      "learning_rate": 0.00244590998365481,
+      "loss": 1.222,
+      "step": 3993
+    },
+    {
+      "epoch": 0.3505517141125261,
+      "grad_norm": 0.06640625,
+      "learning_rate": 0.0024455673771780094,
+      "loss": 1.2856,
+      "step": 3994
+    },
+    {
+      "epoch": 0.3506394836954286,
+      "grad_norm": 0.07421875,
+      "learning_rate": 0.002445224692181857,
+      "loss": 1.2407,
+      "step": 3995
+    },
+    {
+      "epoch": 0.35072725327833104,
+      "grad_norm": 0.07958984375,
+      "learning_rate": 0.0024448819287001735,
+      "loss": 1.2514,
+      "step": 3996
+    },
+    {
+      "epoch": 0.35081502286123356,
+      "grad_norm": 0.0703125,
+      "learning_rate": 0.002444539086766789,
+      "loss": 1.2559,
+      "step": 3997
+    },
+    {
+      "epoch": 0.350902792444136,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.0024441961664155402,
+      "loss": 1.235,
+      "step": 3998
+    },
+    {
+      "epoch": 0.35099056202703854,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0024438531676802726,
+      "loss": 1.2386,
+      "step": 3999
+    },
+    {
+      "epoch": 0.351078331609941,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.002443510090594839,
+      "loss": 1.2591,
+      "step": 4000
+    },
+    {
+      "epoch": 0.351078331609941,
+      "eval_loss": 1.231809139251709,
+      "eval_runtime": 437.7571,
+      "eval_samples_per_second": 33.672,
+      "eval_steps_per_second": 8.418,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 11000,
+  "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": 1.9795135168512e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}