diff --git "a/checkpoint-3108/trainer_state.json" "b/checkpoint-3108/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-3108/trainer_state.json"
@@ -0,0 +1,21817 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 622,
+  "global_step": 3108,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2006484270095825,
+      "learning_rate": 2.5e-06,
+      "loss": 5.0998,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 5.15006685256958,
+      "eval_runtime": 28.119,
+      "eval_samples_per_second": 3.023,
+      "eval_steps_per_second": 1.529,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6314904689788818,
+      "learning_rate": 5e-06,
+      "loss": 5.6374,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3538857698440552,
+      "learning_rate": 7.5e-06,
+      "loss": 5.0656,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9769891500473022,
+      "learning_rate": 1e-05,
+      "loss": 5.1906,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4604706764221191,
+      "learning_rate": 1.25e-05,
+      "loss": 5.1107,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2502201795578003,
+      "learning_rate": 1.5e-05,
+      "loss": 4.8447,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1797655820846558,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 5.0042,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3177151679992676,
+      "learning_rate": 2e-05,
+      "loss": 4.878,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1382358074188232,
+      "learning_rate": 2.2499999999999998e-05,
+      "loss": 5.106,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1523857116699219,
+      "learning_rate": 2.5e-05,
+      "loss": 5.241,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.994879961013794,
+      "learning_rate": 2.75e-05,
+      "loss": 4.4475,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0009862184524536,
+      "learning_rate": 3e-05,
+      "loss": 4.615,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1249079704284668,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 4.3469,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1036263704299927,
+      "learning_rate": 3.5000000000000004e-05,
+      "loss": 4.7748,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1240339279174805,
+      "learning_rate": 3.75e-05,
+      "loss": 4.8696,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9816920757293701,
+      "learning_rate": 4e-05,
+      "loss": 4.5595,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1084593534469604,
+      "learning_rate": 4.25e-05,
+      "loss": 4.2922,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.997968316078186,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 4.3057,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.0401185750961304,
+      "learning_rate": 4.75e-05,
+      "loss": 4.364,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.0804249048233032,
+      "learning_rate": 5e-05,
+      "loss": 4.8663,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2384392023086548,
+      "learning_rate": 5.25e-05,
+      "loss": 4.2992,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.231845736503601,
+      "learning_rate": 5.5e-05,
+      "loss": 4.8637,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.242187261581421,
+      "learning_rate": 5.75e-05,
+      "loss": 3.865,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.2873363494873047,
+      "learning_rate": 6e-05,
+      "loss": 3.8585,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.1170119047164917,
+      "learning_rate": 6.25e-05,
+      "loss": 3.8948,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1938223838806152,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 3.8009,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3068759441375732,
+      "learning_rate": 6.75e-05,
+      "loss": 3.6811,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.280858039855957,
+      "learning_rate": 7.000000000000001e-05,
+      "loss": 3.9975,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3869880437850952,
+      "learning_rate": 7.25e-05,
+      "loss": 3.5562,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3908005952835083,
+      "learning_rate": 7.5e-05,
+      "loss": 3.296,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.622360348701477,
+      "learning_rate": 7.75e-05,
+      "loss": 3.8233,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5797035694122314,
+      "learning_rate": 8e-05,
+      "loss": 3.6274,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.816627025604248,
+      "learning_rate": 8.25e-05,
+      "loss": 3.3872,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1191844940185547,
+      "learning_rate": 8.5e-05,
+      "loss": 3.5982,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.9559025764465332,
+      "learning_rate": 8.75e-05,
+      "loss": 3.5972,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6593563556671143,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 3.3982,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0360360145568848,
+      "learning_rate": 9.25e-05,
+      "loss": 3.5185,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9218908548355103,
+      "learning_rate": 9.5e-05,
+      "loss": 3.5998,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6969695091247559,
+      "learning_rate": 9.750000000000001e-05,
+      "loss": 3.2894,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6107378005981445,
+      "learning_rate": 0.0001,
+      "loss": 3.3074,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6574798822402954,
+      "learning_rate": 0.0001025,
+      "loss": 3.4148,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8693658113479614,
+      "learning_rate": 0.000105,
+      "loss": 3.4726,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4956166744232178,
+      "learning_rate": 0.0001075,
+      "loss": 2.9462,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6204662322998047,
+      "learning_rate": 0.00011,
+      "loss": 3.1303,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8228565454483032,
+      "learning_rate": 0.00011250000000000001,
+      "loss": 3.9787,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6097328662872314,
+      "learning_rate": 0.000115,
+      "loss": 2.9556,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.843204140663147,
+      "learning_rate": 0.0001175,
+      "loss": 3.0469,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8468220233917236,
+      "learning_rate": 0.00012,
+      "loss": 3.1549,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8635084629058838,
+      "learning_rate": 0.0001225,
+      "loss": 3.2309,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9674320220947266,
+      "learning_rate": 0.000125,
+      "loss": 3.264,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.514456868171692,
+      "learning_rate": 0.0001275,
+      "loss": 2.895,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8026971817016602,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 2.9692,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.872318983078003,
+      "learning_rate": 0.00013250000000000002,
+      "loss": 3.7424,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7249802350997925,
+      "learning_rate": 0.000135,
+      "loss": 2.941,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.677011489868164,
+      "learning_rate": 0.0001375,
+      "loss": 3.3312,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.9033194780349731,
+      "learning_rate": 0.00014000000000000001,
+      "loss": 2.5066,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9987455606460571,
+      "learning_rate": 0.0001425,
+      "loss": 3.4693,
+      "step": 57
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9788786172866821,
+      "learning_rate": 0.000145,
+      "loss": 2.8805,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7737230062484741,
+      "learning_rate": 0.0001475,
+      "loss": 2.8758,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7360577583312988,
+      "learning_rate": 0.00015,
+      "loss": 3.4,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.982911229133606,
+      "learning_rate": 0.0001525,
+      "loss": 2.779,
+      "step": 61
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.9741348028182983,
+      "learning_rate": 0.000155,
+      "loss": 3.3381,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.060384750366211,
+      "learning_rate": 0.0001575,
+      "loss": 3.6285,
+      "step": 63
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.6591861248016357,
+      "learning_rate": 0.00016,
+      "loss": 3.1691,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.8308738470077515,
+      "learning_rate": 0.00016250000000000002,
+      "loss": 3.3354,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.8891240358352661,
+      "learning_rate": 0.000165,
+      "loss": 3.2649,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.697945237159729,
+      "learning_rate": 0.0001675,
+      "loss": 2.8464,
+      "step": 67
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.6797360181808472,
+      "learning_rate": 0.00017,
+      "loss": 3.3645,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.964936375617981,
+      "learning_rate": 0.0001725,
+      "loss": 2.6052,
+      "step": 69
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.081937313079834,
+      "learning_rate": 0.000175,
+      "loss": 3.3901,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8318378925323486,
+      "learning_rate": 0.0001775,
+      "loss": 3.2655,
+      "step": 71
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.757249116897583,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 3.254,
+      "step": 72
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.8252121210098267,
+      "learning_rate": 0.0001825,
+      "loss": 2.5852,
+      "step": 73
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0867161750793457,
+      "learning_rate": 0.000185,
+      "loss": 2.8474,
+      "step": 74
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1990573406219482,
+      "learning_rate": 0.0001875,
+      "loss": 2.5813,
+      "step": 75
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1143388748168945,
+      "learning_rate": 0.00019,
+      "loss": 3.1759,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.1437313556671143,
+      "learning_rate": 0.00019250000000000002,
+      "loss": 2.7499,
+      "step": 77
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0984110832214355,
+      "learning_rate": 0.00019500000000000002,
+      "loss": 2.6248,
+      "step": 78
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.28216290473938,
+      "learning_rate": 0.0001975,
+      "loss": 2.5725,
+      "step": 79
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.966034173965454,
+      "learning_rate": 0.0002,
+      "loss": 3.0547,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.117560625076294,
+      "learning_rate": 0.00020250000000000002,
+      "loss": 3.1316,
+      "step": 81
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.8826152086257935,
+      "learning_rate": 0.000205,
+      "loss": 2.8674,
+      "step": 82
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.997240424156189,
+      "learning_rate": 0.0002075,
+      "loss": 2.4458,
+      "step": 83
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.6871142387390137,
+      "learning_rate": 0.00021,
+      "loss": 2.814,
+      "step": 84
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.066634178161621,
+      "learning_rate": 0.0002125,
+      "loss": 3.2066,
+      "step": 85
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9600077867507935,
+      "learning_rate": 0.000215,
+      "loss": 3.2272,
+      "step": 86
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1469485759735107,
+      "learning_rate": 0.0002175,
+      "loss": 3.4447,
+      "step": 87
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.944505214691162,
+      "learning_rate": 0.00022,
+      "loss": 2.5898,
+      "step": 88
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3109183311462402,
+      "learning_rate": 0.00022250000000000001,
+      "loss": 2.771,
+      "step": 89
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.174560070037842,
+      "learning_rate": 0.00022500000000000002,
+      "loss": 3.0631,
+      "step": 90
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8415343761444092,
+      "learning_rate": 0.0002275,
+      "loss": 3.123,
+      "step": 91
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.104952096939087,
+      "learning_rate": 0.00023,
+      "loss": 3.1166,
+      "step": 92
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.3860461711883545,
+      "learning_rate": 0.0002325,
+      "loss": 3.048,
+      "step": 93
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.132197856903076,
+      "learning_rate": 0.000235,
+      "loss": 2.7315,
+      "step": 94
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7788770198822021,
+      "learning_rate": 0.0002375,
+      "loss": 3.127,
+      "step": 95
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8100048303604126,
+      "learning_rate": 0.00024,
+      "loss": 2.3229,
+      "step": 96
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7176936864852905,
+      "learning_rate": 0.00024249999999999999,
+      "loss": 2.8595,
+      "step": 97
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.920607089996338,
+      "learning_rate": 0.000245,
+      "loss": 2.6285,
+      "step": 98
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8976895809173584,
+      "learning_rate": 0.0002475,
+      "loss": 2.3511,
+      "step": 99
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0189640522003174,
+      "learning_rate": 0.00025,
+      "loss": 2.7418,
+      "step": 100
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9701827764511108,
+      "learning_rate": 0.00024999993182517067,
+      "loss": 2.0604,
+      "step": 101
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.252211570739746,
+      "learning_rate": 0.00024999972730075704,
+      "loss": 2.6496,
+      "step": 102
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.1773853302001953,
+      "learning_rate": 0.0002499993864269822,
+      "loss": 2.8563,
+      "step": 103
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.173621654510498,
+      "learning_rate": 0.00024999890920421796,
+      "loss": 2.5704,
+      "step": 104
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8827345371246338,
+      "learning_rate": 0.0002499982956329849,
+      "loss": 2.7194,
+      "step": 105
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.4898743629455566,
+      "learning_rate": 0.00024999754571395234,
+      "loss": 3.4996,
+      "step": 106
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.006110191345215,
+      "learning_rate": 0.00024999665944793815,
+      "loss": 2.7179,
+      "step": 107
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0414044857025146,
+      "learning_rate": 0.00024999563683590924,
+      "loss": 2.9323,
+      "step": 108
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1107375621795654,
+      "learning_rate": 0.0002499944778789809,
+      "loss": 2.4901,
+      "step": 109
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8949179649353027,
+      "learning_rate": 0.0002499931825784175,
+      "loss": 2.6771,
+      "step": 110
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.852339267730713,
+      "learning_rate": 0.00024999175093563185,
+      "loss": 2.5123,
+      "step": 111
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7306336164474487,
+      "learning_rate": 0.0002499901829521856,
+      "loss": 2.7333,
+      "step": 112
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1602630615234375,
+      "learning_rate": 0.000249988478629789,
+      "loss": 3.2803,
+      "step": 113
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8918505907058716,
+      "learning_rate": 0.0002499866379703013,
+      "loss": 2.7467,
+      "step": 114
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.9145008325576782,
+      "learning_rate": 0.0002499846609757302,
+      "loss": 2.7413,
+      "step": 115
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8250600099563599,
+      "learning_rate": 0.00024998254764823215,
+      "loss": 2.5852,
+      "step": 116
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1568727493286133,
+      "learning_rate": 0.0002499802979901124,
+      "loss": 2.7673,
+      "step": 117
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7749338150024414,
+      "learning_rate": 0.0002499779120038249,
+      "loss": 2.4479,
+      "step": 118
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7486765384674072,
+      "learning_rate": 0.0002499753896919723,
+      "loss": 1.8766,
+      "step": 119
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.076768159866333,
+      "learning_rate": 0.0002499727310573059,
+      "loss": 1.8791,
+      "step": 120
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.604339599609375,
+      "learning_rate": 0.00024996993610272567,
+      "loss": 2.0573,
+      "step": 121
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.116293430328369,
+      "learning_rate": 0.00024996700483128037,
+      "loss": 2.7077,
+      "step": 122
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.233560800552368,
+      "learning_rate": 0.0002499639372461675,
+      "loss": 3.1111,
+      "step": 123
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.750420093536377,
+      "learning_rate": 0.0002499607333507331,
+      "loss": 2.4861,
+      "step": 124
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8790860176086426,
+      "learning_rate": 0.000249957393148472,
+      "loss": 2.5195,
+      "step": 125
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.7548282146453857,
+      "learning_rate": 0.0002499539166430276,
+      "loss": 2.5072,
+      "step": 126
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8878798484802246,
+      "learning_rate": 0.0002499503038381922,
+      "loss": 2.3164,
+      "step": 127
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.040093421936035,
+      "learning_rate": 0.00024994655473790657,
+      "loss": 2.0806,
+      "step": 128
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9570492506027222,
+      "learning_rate": 0.0002499426693462602,
+      "loss": 2.7483,
+      "step": 129
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9189451932907104,
+      "learning_rate": 0.0002499386476674913,
+      "loss": 2.4365,
+      "step": 130
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.840690016746521,
+      "learning_rate": 0.0002499344897059867,
+      "loss": 2.2675,
+      "step": 131
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0497002601623535,
+      "learning_rate": 0.0002499301954662818,
+      "loss": 2.3665,
+      "step": 132
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.057999849319458,
+      "learning_rate": 0.0002499257649530609,
+      "loss": 2.2258,
+      "step": 133
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1040830612182617,
+      "learning_rate": 0.00024992119817115674,
+      "loss": 2.9013,
+      "step": 134
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1239287853240967,
+      "learning_rate": 0.0002499164951255507,
+      "loss": 2.4111,
+      "step": 135
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.246864080429077,
+      "learning_rate": 0.00024991165582137286,
+      "loss": 2.1288,
+      "step": 136
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.854050874710083,
+      "learning_rate": 0.000249906680263902,
+      "loss": 2.308,
+      "step": 137
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7639249563217163,
+      "learning_rate": 0.0002499015684585654,
+      "loss": 2.1624,
+      "step": 138
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9622340202331543,
+      "learning_rate": 0.0002498963204109389,
+      "loss": 1.7293,
+      "step": 139
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9182345867156982,
+      "learning_rate": 0.0002498909361267472,
+      "loss": 2.5676,
+      "step": 140
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.546036720275879,
+      "learning_rate": 0.0002498854156118634,
+      "loss": 2.0416,
+      "step": 141
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7901947498321533,
+      "learning_rate": 0.0002498797588723093,
+      "loss": 2.3867,
+      "step": 142
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.682002305984497,
+      "learning_rate": 0.00024987396591425517,
+      "loss": 2.2,
+      "step": 143
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.1231274604797363,
+      "learning_rate": 0.00024986803674402003,
+      "loss": 2.7393,
+      "step": 144
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7583919763565063,
+      "learning_rate": 0.0002498619713680714,
+      "loss": 2.2841,
+      "step": 145
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6599968671798706,
+      "learning_rate": 0.00024985576979302533,
+      "loss": 1.9753,
+      "step": 146
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5316509008407593,
+      "learning_rate": 0.00024984943202564655,
+      "loss": 1.9288,
+      "step": 147
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8190314769744873,
+      "learning_rate": 0.0002498429580728482,
+      "loss": 2.2112,
+      "step": 148
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7357500791549683,
+      "learning_rate": 0.00024983634794169214,
+      "loss": 2.2501,
+      "step": 149
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8484951257705688,
+      "learning_rate": 0.0002498296016393886,
+      "loss": 1.7479,
+      "step": 150
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.765267252922058,
+      "learning_rate": 0.00024982271917329646,
+      "loss": 2.5703,
+      "step": 151
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6502829790115356,
+      "learning_rate": 0.0002498157005509231,
+      "loss": 2.7001,
+      "step": 152
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.010906934738159,
+      "learning_rate": 0.0002498085457799244,
+      "loss": 2.8633,
+      "step": 153
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.850074052810669,
+      "learning_rate": 0.00024980125486810477,
+      "loss": 2.2063,
+      "step": 154
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9126758575439453,
+      "learning_rate": 0.00024979382782341713,
+      "loss": 2.1976,
+      "step": 155
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0436511039733887,
+      "learning_rate": 0.00024978626465396286,
+      "loss": 2.3515,
+      "step": 156
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3156819343566895,
+      "learning_rate": 0.0002497785653679919,
+      "loss": 2.7545,
+      "step": 157
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9115028381347656,
+      "learning_rate": 0.00024977072997390247,
+      "loss": 2.206,
+      "step": 158
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8286609649658203,
+      "learning_rate": 0.00024976275848024156,
+      "loss": 2.371,
+      "step": 159
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6878124475479126,
+      "learning_rate": 0.0002497546508957044,
+      "loss": 2.1747,
+      "step": 160
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.893005132675171,
+      "learning_rate": 0.00024974640722913465,
+      "loss": 1.7254,
+      "step": 161
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.11098313331604,
+      "learning_rate": 0.0002497380274895246,
+      "loss": 2.5951,
+      "step": 162
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7469336986541748,
+      "learning_rate": 0.00024972951168601476,
+      "loss": 2.2754,
+      "step": 163
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.835694670677185,
+      "learning_rate": 0.00024972085982789415,
+      "loss": 2.1267,
+      "step": 164
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6754062175750732,
+      "learning_rate": 0.0002497120719246002,
+      "loss": 1.9974,
+      "step": 165
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8209201097488403,
+      "learning_rate": 0.0002497031479857188,
+      "loss": 2.5032,
+      "step": 166
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.76840341091156,
+      "learning_rate": 0.000249694088020984,
+      "loss": 1.7389,
+      "step": 167
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.552156925201416,
+      "learning_rate": 0.00024968489204027863,
+      "loss": 1.8085,
+      "step": 168
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.761948585510254,
+      "learning_rate": 0.0002496755600536334,
+      "loss": 2.0805,
+      "step": 169
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.200204372406006,
+      "learning_rate": 0.0002496660920712277,
+      "loss": 2.0036,
+      "step": 170
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8660489320755005,
+      "learning_rate": 0.0002496564881033892,
+      "loss": 2.1293,
+      "step": 171
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.33247709274292,
+      "learning_rate": 0.00024964674816059393,
+      "loss": 2.3811,
+      "step": 172
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8773770332336426,
+      "learning_rate": 0.00024963687225346604,
+      "loss": 1.9161,
+      "step": 173
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1089746952056885,
+      "learning_rate": 0.0002496268603927783,
+      "loss": 2.5818,
+      "step": 174
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.0784616470336914,
+      "learning_rate": 0.00024961671258945156,
+      "loss": 2.0916,
+      "step": 175
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8207252025604248,
+      "learning_rate": 0.000249606428854555,
+      "loss": 2.3204,
+      "step": 176
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9220213890075684,
+      "learning_rate": 0.00024959600919930607,
+      "loss": 2.5432,
+      "step": 177
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8817466497421265,
+      "learning_rate": 0.0002495854536350706,
+      "loss": 2.4326,
+      "step": 178
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8520444631576538,
+      "learning_rate": 0.0002495747621733625,
+      "loss": 1.9673,
+      "step": 179
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7097537517547607,
+      "learning_rate": 0.00024956393482584397,
+      "loss": 2.0189,
+      "step": 180
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8573436737060547,
+      "learning_rate": 0.0002495529716043254,
+      "loss": 2.0942,
+      "step": 181
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9761507511138916,
+      "learning_rate": 0.00024954187252076564,
+      "loss": 2.189,
+      "step": 182
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8549177646636963,
+      "learning_rate": 0.00024953063758727137,
+      "loss": 2.0821,
+      "step": 183
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7199348211288452,
+      "learning_rate": 0.00024951926681609767,
+      "loss": 1.9875,
+      "step": 184
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.6276706457138062,
+      "learning_rate": 0.00024950776021964775,
+      "loss": 2.0707,
+      "step": 185
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.5845235586166382,
+      "learning_rate": 0.000249496117810473,
+      "loss": 1.5413,
+      "step": 186
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7384014129638672,
+      "learning_rate": 0.00024948433960127284,
+      "loss": 2.1632,
+      "step": 187
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0509490966796875,
+      "learning_rate": 0.000249472425604895,
+      "loss": 1.758,
+      "step": 188
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9983670711517334,
+      "learning_rate": 0.0002494603758343352,
+      "loss": 2.5845,
+      "step": 189
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0809295177459717,
+      "learning_rate": 0.0002494481903027373,
+      "loss": 2.6146,
+      "step": 190
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7669973373413086,
+      "learning_rate": 0.0002494358690233933,
+      "loss": 2.0003,
+      "step": 191
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7728503942489624,
+      "learning_rate": 0.0002494234120097431,
+      "loss": 2.0072,
+      "step": 192
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9957607984542847,
+      "learning_rate": 0.0002494108192753748,
+      "loss": 2.2449,
+      "step": 193
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0449464321136475,
+      "learning_rate": 0.0002493980908340246,
+      "loss": 2.0996,
+      "step": 194
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8009361028671265,
+      "learning_rate": 0.0002493852266995766,
+      "loss": 2.0876,
+      "step": 195
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9847387075424194,
+      "learning_rate": 0.00024937222688606303,
+      "loss": 1.8041,
+      "step": 196
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.0888640880584717,
+      "learning_rate": 0.00024935909140766393,
+      "loss": 2.4709,
+      "step": 197
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6796460151672363,
+      "learning_rate": 0.0002493458202787075,
+      "loss": 1.9681,
+      "step": 198
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6660256385803223,
+      "learning_rate": 0.0002493324135136699,
+      "loss": 2.5484,
+      "step": 199
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8919657468795776,
+      "learning_rate": 0.0002493188711271751,
+      "loss": 2.5249,
+      "step": 200
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8183636665344238,
+      "learning_rate": 0.0002493051931339952,
+      "loss": 1.8468,
+      "step": 201
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6800963878631592,
+      "learning_rate": 0.0002492913795490501,
+      "loss": 1.9648,
+      "step": 202
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8262184858322144,
+      "learning_rate": 0.00024927743038740747,
+      "loss": 2.0492,
+      "step": 203
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1556880474090576,
+      "learning_rate": 0.0002492633456642832,
+      "loss": 2.4633,
+      "step": 204
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7076188325881958,
+      "learning_rate": 0.0002492491253950408,
+      "loss": 2.1612,
+      "step": 205
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8892126083374023,
+      "learning_rate": 0.0002492347695951917,
+      "loss": 1.9447,
+      "step": 206
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.873687744140625,
+      "learning_rate": 0.0002492202782803952,
+      "loss": 2.7744,
+      "step": 207
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7962186336517334,
+      "learning_rate": 0.0002492056514664583,
+      "loss": 2.0622,
+      "step": 208
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9481604099273682,
+      "learning_rate": 0.00024919088916933597,
+      "loss": 1.9443,
+      "step": 209
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7974705696105957,
+      "learning_rate": 0.0002491759914051308,
+      "loss": 2.3544,
+      "step": 210
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.5557405948638916,
+      "learning_rate": 0.00024916095819009336,
+      "loss": 2.2321,
+      "step": 211
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.729710340499878,
+      "learning_rate": 0.0002491457895406218,
+      "loss": 1.9179,
+      "step": 212
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6658893823623657,
+      "learning_rate": 0.00024913048547326193,
+      "loss": 2.088,
+      "step": 213
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6439865827560425,
+      "learning_rate": 0.0002491150460047075,
+      "loss": 1.8413,
+      "step": 214
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.595862865447998,
+      "learning_rate": 0.00024909947115179983,
+      "loss": 2.0446,
+      "step": 215
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.21343994140625,
+      "learning_rate": 0.00024908376093152784,
+      "loss": 2.841,
+      "step": 216
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7837032079696655,
+      "learning_rate": 0.0002490679153610283,
+      "loss": 1.8496,
+      "step": 217
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6079756021499634,
+      "learning_rate": 0.00024905193445758545,
+      "loss": 1.9118,
+      "step": 218
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9322482347488403,
+      "learning_rate": 0.00024903581823863125,
+      "loss": 2.2768,
+      "step": 219
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9454126358032227,
+      "learning_rate": 0.0002490195667217452,
+      "loss": 2.0224,
+      "step": 220
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8890690803527832,
+      "learning_rate": 0.00024900317992465447,
+      "loss": 1.9491,
+      "step": 221
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.844327449798584,
+      "learning_rate": 0.0002489866578652337,
+      "loss": 2.3316,
+      "step": 222
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9561625719070435,
+      "learning_rate": 0.00024897000056150505,
+      "loss": 1.8929,
+      "step": 223
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.812215805053711,
+      "learning_rate": 0.0002489532080316383,
+      "loss": 1.5681,
+      "step": 224
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7232577800750732,
+      "learning_rate": 0.0002489362802939507,
+      "loss": 2.2427,
+      "step": 225
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.5524508953094482,
+      "learning_rate": 0.00024891921736690703,
+      "loss": 2.1043,
+      "step": 226
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.5697345733642578,
+      "learning_rate": 0.0002489020192691194,
+      "loss": 1.9664,
+      "step": 227
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.5127205848693848,
+      "learning_rate": 0.0002488846860193475,
+      "loss": 1.5816,
+      "step": 228
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.036536693572998,
+      "learning_rate": 0.0002488672176364984,
+      "loss": 2.7449,
+      "step": 229
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9850804805755615,
+      "learning_rate": 0.0002488496141396265,
+      "loss": 1.8061,
+      "step": 230
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.029428720474243,
+      "learning_rate": 0.0002488318755479337,
+      "loss": 1.7861,
+      "step": 231
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6909549236297607,
+      "learning_rate": 0.00024881400188076923,
+      "loss": 2.1058,
+      "step": 232
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7147257328033447,
+      "learning_rate": 0.0002487959931576296,
+      "loss": 1.6736,
+      "step": 233
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6619194746017456,
+      "learning_rate": 0.00024877784939815863,
+      "loss": 1.7423,
+      "step": 234
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7721213102340698,
+      "learning_rate": 0.0002487595706221476,
+      "loss": 1.4057,
+      "step": 235
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6098127365112305,
+      "learning_rate": 0.00024874115684953485,
+      "loss": 1.6113,
+      "step": 236
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.2047505378723145,
+      "learning_rate": 0.00024872260810040607,
+      "loss": 2.5387,
+      "step": 237
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6279314756393433,
+      "learning_rate": 0.0002487039243949943,
+      "loss": 1.9872,
+      "step": 238
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.795262098312378,
+      "learning_rate": 0.0002486851057536795,
+      "loss": 1.4461,
+      "step": 239
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9134374856948853,
+      "learning_rate": 0.00024866615219698915,
+      "loss": 2.3969,
+      "step": 240
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6883046627044678,
+      "learning_rate": 0.0002486470637455976,
+      "loss": 2.2354,
+      "step": 241
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7430942058563232,
+      "learning_rate": 0.00024862784042032666,
+      "loss": 1.5288,
+      "step": 242
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6902698278427124,
+      "learning_rate": 0.00024860848224214486,
+      "loss": 2.102,
+      "step": 243
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.5325380563735962,
+      "learning_rate": 0.0002485889892321683,
+      "loss": 1.3993,
+      "step": 244
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.5996594429016113,
+      "learning_rate": 0.00024856936141165963,
+      "loss": 1.586,
+      "step": 245
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9675129652023315,
+      "learning_rate": 0.00024854959880202905,
+      "loss": 2.6326,
+      "step": 246
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9281418323516846,
+      "learning_rate": 0.00024852970142483346,
+      "loss": 2.3056,
+      "step": 247
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.0125837326049805,
+      "learning_rate": 0.00024850966930177687,
+      "loss": 2.2092,
+      "step": 248
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.904648780822754,
+      "learning_rate": 0.00024848950245471023,
+      "loss": 2.1627,
+      "step": 249
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7093470096588135,
+      "learning_rate": 0.00024846920090563156,
+      "loss": 1.9851,
+      "step": 250
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7834309339523315,
+      "learning_rate": 0.0002484487646766857,
+      "loss": 2.112,
+      "step": 251
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.5248178243637085,
+      "learning_rate": 0.0002484281937901644,
+      "loss": 1.426,
+      "step": 252
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.4602184295654297,
+      "learning_rate": 0.0002484074882685063,
+      "loss": 1.3696,
+      "step": 253
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.773917555809021,
+      "learning_rate": 0.0002483866481342971,
+      "loss": 2.1261,
+      "step": 254
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.692901611328125,
+      "learning_rate": 0.000248365673410269,
+      "loss": 2.3204,
+      "step": 255
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8946104049682617,
+      "learning_rate": 0.0002483445641193012,
+      "loss": 1.8685,
+      "step": 256
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.919169306755066,
+      "learning_rate": 0.0002483233202844197,
+      "loss": 2.0433,
+      "step": 257
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.84755277633667,
+      "learning_rate": 0.00024830194192879715,
+      "loss": 2.2363,
+      "step": 258
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.112445116043091,
+      "learning_rate": 0.00024828042907575304,
+      "loss": 2.6892,
+      "step": 259
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6115814447402954,
+      "learning_rate": 0.0002482587817487536,
+      "loss": 1.8159,
+      "step": 260
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6731853485107422,
+      "learning_rate": 0.00024823699997141154,
+      "loss": 2.1162,
+      "step": 261
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5666356086730957,
+      "learning_rate": 0.0002482150837674864,
+      "loss": 1.8052,
+      "step": 262
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5038379430770874,
+      "learning_rate": 0.0002481930331608844,
+      "loss": 1.8459,
+      "step": 263
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5254186391830444,
+      "learning_rate": 0.00024817084817565827,
+      "loss": 1.9475,
+      "step": 264
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7081129550933838,
+      "learning_rate": 0.0002481485288360072,
+      "loss": 2.4121,
+      "step": 265
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9189531803131104,
+      "learning_rate": 0.0002481260751662772,
+      "loss": 2.4579,
+      "step": 266
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6706724166870117,
+      "learning_rate": 0.00024810348719096065,
+      "loss": 1.6386,
+      "step": 267
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.539415717124939,
+      "learning_rate": 0.0002480807649346964,
+      "loss": 1.6492,
+      "step": 268
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.747955560684204,
+      "learning_rate": 0.00024805790842226985,
+      "loss": 1.6512,
+      "step": 269
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6738513708114624,
+      "learning_rate": 0.0002480349176786128,
+      "loss": 1.7555,
+      "step": 270
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9724032878875732,
+      "learning_rate": 0.0002480117927288035,
+      "loss": 1.6291,
+      "step": 271
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6531236171722412,
+      "learning_rate": 0.00024798853359806665,
+      "loss": 1.6063,
+      "step": 272
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9296516180038452,
+      "learning_rate": 0.0002479651403117732,
+      "loss": 1.4847,
+      "step": 273
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7940136194229126,
+      "learning_rate": 0.0002479416128954404,
+      "loss": 2.4839,
+      "step": 274
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.010266065597534,
+      "learning_rate": 0.00024791795137473204,
+      "loss": 1.7243,
+      "step": 275
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8504762649536133,
+      "learning_rate": 0.00024789415577545793,
+      "loss": 1.9652,
+      "step": 276
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.4306342601776123,
+      "learning_rate": 0.0002478702261235743,
+      "loss": 1.5664,
+      "step": 277
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6604520082473755,
+      "learning_rate": 0.0002478461624451835,
+      "loss": 2.0659,
+      "step": 278
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0522496700286865,
+      "learning_rate": 0.0002478219647665342,
+      "loss": 1.9088,
+      "step": 279
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6429219245910645,
+      "learning_rate": 0.0002477976331140211,
+      "loss": 1.7491,
+      "step": 280
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.397255301475525,
+      "learning_rate": 0.00024777316751418515,
+      "loss": 1.3155,
+      "step": 281
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.5616679191589355,
+      "learning_rate": 0.00024774856799371326,
+      "loss": 1.8068,
+      "step": 282
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7590296268463135,
+      "learning_rate": 0.00024772383457943864,
+      "loss": 1.7305,
+      "step": 283
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.6376055479049683,
+      "learning_rate": 0.00024769896729834036,
+      "loss": 1.7105,
+      "step": 284
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.385542392730713,
+      "learning_rate": 0.00024767396617754364,
+      "loss": 1.605,
+      "step": 285
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.827626347541809,
+      "learning_rate": 0.0002476488312443195,
+      "loss": 2.032,
+      "step": 286
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.1502676010131836,
+      "learning_rate": 0.00024762356252608527,
+      "loss": 1.9228,
+      "step": 287
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5369927883148193,
+      "learning_rate": 0.00024759816005040384,
+      "loss": 1.3886,
+      "step": 288
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7193357944488525,
+      "learning_rate": 0.0002475726238449842,
+      "loss": 2.0007,
+      "step": 289
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.561558246612549,
+      "learning_rate": 0.00024754695393768114,
+      "loss": 2.6965,
+      "step": 290
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2155981063842773,
+      "learning_rate": 0.0002475211503564954,
+      "loss": 1.678,
+      "step": 291
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.59506356716156,
+      "learning_rate": 0.00024749521312957337,
+      "loss": 1.7036,
+      "step": 292
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5700795650482178,
+      "learning_rate": 0.0002474691422852074,
+      "loss": 1.5835,
+      "step": 293
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8655942678451538,
+      "learning_rate": 0.00024744293785183537,
+      "loss": 1.6957,
+      "step": 294
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.434198260307312,
+      "learning_rate": 0.00024741659985804115,
+      "loss": 1.5576,
+      "step": 295
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.4937015771865845,
+      "learning_rate": 0.0002473901283325541,
+      "loss": 1.7204,
+      "step": 296
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9438872337341309,
+      "learning_rate": 0.00024736352330424923,
+      "loss": 2.2574,
+      "step": 297
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8500808477401733,
+      "learning_rate": 0.0002473367848021473,
+      "loss": 2.2356,
+      "step": 298
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6820452213287354,
+      "learning_rate": 0.00024730991285541455,
+      "loss": 2.7996,
+      "step": 299
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6098867654800415,
+      "learning_rate": 0.0002472829074933628,
+      "loss": 2.0764,
+      "step": 300
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.763376235961914,
+      "learning_rate": 0.00024725576874544956,
+      "loss": 2.1674,
+      "step": 301
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7412651777267456,
+      "learning_rate": 0.0002472284966412776,
+      "loss": 2.2091,
+      "step": 302
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.690738320350647,
+      "learning_rate": 0.00024720109121059524,
+      "loss": 2.1094,
+      "step": 303
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.3498783111572266,
+      "learning_rate": 0.00024717355248329625,
+      "loss": 1.8,
+      "step": 304
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6631073951721191,
+      "learning_rate": 0.00024714588048941987,
+      "loss": 2.251,
+      "step": 305
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7227246761322021,
+      "learning_rate": 0.0002471180752591506,
+      "loss": 1.8822,
+      "step": 306
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7856996059417725,
+      "learning_rate": 0.00024709013682281826,
+      "loss": 1.914,
+      "step": 307
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0128395557403564,
+      "learning_rate": 0.0002470620652108981,
+      "loss": 2.0113,
+      "step": 308
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.4962693452835083,
+      "learning_rate": 0.00024703386045401047,
+      "loss": 1.7985,
+      "step": 309
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5838319063186646,
+      "learning_rate": 0.0002470055225829211,
+      "loss": 1.5443,
+      "step": 310
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6290487051010132,
+      "learning_rate": 0.0002469770516285409,
+      "loss": 1.6863,
+      "step": 311
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.46754789352417,
+      "learning_rate": 0.0002469484476219259,
+      "loss": 1.6632,
+      "step": 312
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8494882583618164,
+      "learning_rate": 0.00024691971059427717,
+      "loss": 1.7439,
+      "step": 313
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.868640661239624,
+      "learning_rate": 0.000246890840576941,
+      "loss": 2.0376,
+      "step": 314
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.553097128868103,
+      "learning_rate": 0.0002468618376014088,
+      "loss": 1.9508,
+      "step": 315
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6028902530670166,
+      "learning_rate": 0.000246832701699317,
+      "loss": 1.5764,
+      "step": 316
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1267387866973877,
+      "learning_rate": 0.0002468034329024468,
+      "loss": 1.8886,
+      "step": 317
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.5905228853225708,
+      "learning_rate": 0.00024677403124272456,
+      "loss": 1.8134,
+      "step": 318
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.4769141674041748,
+      "learning_rate": 0.0002467444967522216,
+      "loss": 1.2115,
+      "step": 319
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.4753649234771729,
+      "learning_rate": 0.000246714829463154,
+      "loss": 1.653,
+      "step": 320
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6038315296173096,
+      "learning_rate": 0.0002466850294078828,
+      "loss": 2.1011,
+      "step": 321
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7277235984802246,
+      "learning_rate": 0.00024665509661891385,
+      "loss": 2.1318,
+      "step": 322
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6639180183410645,
+      "learning_rate": 0.0002466250311288977,
+      "loss": 1.6277,
+      "step": 323
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.660370111465454,
+      "learning_rate": 0.00024659483297062964,
+      "loss": 2.112,
+      "step": 324
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7722567319869995,
+      "learning_rate": 0.0002465645021770499,
+      "loss": 2.0158,
+      "step": 325
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.559173583984375,
+      "learning_rate": 0.00024653403878124305,
+      "loss": 1.8675,
+      "step": 326
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.679623007774353,
+      "learning_rate": 0.0002465034428164386,
+      "loss": 1.844,
+      "step": 327
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.72882878780365,
+      "learning_rate": 0.00024647271431601055,
+      "loss": 2.2323,
+      "step": 328
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.4677948951721191,
+      "learning_rate": 0.00024644185331347735,
+      "loss": 1.8718,
+      "step": 329
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.5333250761032104,
+      "learning_rate": 0.00024641085984250223,
+      "loss": 2.0585,
+      "step": 330
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.307824730873108,
+      "learning_rate": 0.0002463797339368927,
+      "loss": 1.8447,
+      "step": 331
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6973696947097778,
+      "learning_rate": 0.0002463484756306009,
+      "loss": 1.5235,
+      "step": 332
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.3276690244674683,
+      "learning_rate": 0.0002463170849577232,
+      "loss": 1.9472,
+      "step": 333
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.557236671447754,
+      "learning_rate": 0.0002462855619525005,
+      "loss": 1.6783,
+      "step": 334
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6383414268493652,
+      "learning_rate": 0.000246253906649318,
+      "loss": 1.4636,
+      "step": 335
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.4756358861923218,
+      "learning_rate": 0.0002462221190827053,
+      "loss": 1.4215,
+      "step": 336
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6746768951416016,
+      "learning_rate": 0.000246190199287336,
+      "loss": 1.7268,
+      "step": 337
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8532724380493164,
+      "learning_rate": 0.00024615814729802833,
+      "loss": 2.2173,
+      "step": 338
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8261704444885254,
+      "learning_rate": 0.0002461259631497444,
+      "loss": 1.6643,
+      "step": 339
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6665236949920654,
+      "learning_rate": 0.0002460936468775907,
+      "loss": 1.9251,
+      "step": 340
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5993882417678833,
+      "learning_rate": 0.00024606119851681757,
+      "loss": 1.4908,
+      "step": 341
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8725510835647583,
+      "learning_rate": 0.00024602861810281966,
+      "loss": 2.1784,
+      "step": 342
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.803350567817688,
+      "learning_rate": 0.0002459959056711357,
+      "loss": 2.11,
+      "step": 343
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6295034885406494,
+      "learning_rate": 0.00024596306125744815,
+      "loss": 1.777,
+      "step": 344
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.189607858657837,
+      "learning_rate": 0.00024593008489758375,
+      "loss": 2.4662,
+      "step": 345
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.835728645324707,
+      "learning_rate": 0.0002458969766275129,
+      "loss": 1.8764,
+      "step": 346
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6741111278533936,
+      "learning_rate": 0.00024586373648335014,
+      "loss": 1.8256,
+      "step": 347
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5066823959350586,
+      "learning_rate": 0.0002458303645013536,
+      "loss": 1.4491,
+      "step": 348
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.439063549041748,
+      "learning_rate": 0.00024579686071792543,
+      "loss": 1.8896,
+      "step": 349
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.2669867277145386,
+      "learning_rate": 0.0002457632251696115,
+      "loss": 1.7784,
+      "step": 350
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.4590708017349243,
+      "learning_rate": 0.00024572945789310123,
+      "loss": 1.5708,
+      "step": 351
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7421554327011108,
+      "learning_rate": 0.000245695558925228,
+      "loss": 2.0778,
+      "step": 352
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5311520099639893,
+      "learning_rate": 0.00024566152830296875,
+      "loss": 1.4703,
+      "step": 353
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9303866624832153,
+      "learning_rate": 0.0002456273660634438,
+      "loss": 2.0816,
+      "step": 354
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5437086820602417,
+      "learning_rate": 0.0002455930722439174,
+      "loss": 1.4711,
+      "step": 355
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.50086510181427,
+      "learning_rate": 0.0002455586468817971,
+      "loss": 1.8523,
+      "step": 356
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5427656173706055,
+      "learning_rate": 0.00024552409001463393,
+      "loss": 1.6157,
+      "step": 357
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7002308368682861,
+      "learning_rate": 0.00024548940168012253,
+      "loss": 1.724,
+      "step": 358
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5757559537887573,
+      "learning_rate": 0.0002454545819161008,
+      "loss": 1.663,
+      "step": 359
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.4780895709991455,
+      "learning_rate": 0.00024541963076055,
+      "loss": 1.488,
+      "step": 360
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7047303915023804,
+      "learning_rate": 0.00024538454825159486,
+      "loss": 2.0954,
+      "step": 361
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9181485176086426,
+      "learning_rate": 0.00024534933442750317,
+      "loss": 1.5492,
+      "step": 362
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.150815010070801,
+      "learning_rate": 0.0002453139893266861,
+      "loss": 2.1316,
+      "step": 363
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.663496732711792,
+      "learning_rate": 0.00024527851298769803,
+      "loss": 1.9913,
+      "step": 364
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.6102313995361328,
+      "learning_rate": 0.00024524290544923643,
+      "loss": 1.8153,
+      "step": 365
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.003488302230835,
+      "learning_rate": 0.0002452071667501419,
+      "loss": 2.0315,
+      "step": 366
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.4711098670959473,
+      "learning_rate": 0.0002451712969293982,
+      "loss": 1.4064,
+      "step": 367
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.4077496528625488,
+      "learning_rate": 0.0002451352960261319,
+      "loss": 1.8027,
+      "step": 368
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.4813429117202759,
+      "learning_rate": 0.0002450991640796127,
+      "loss": 1.9595,
+      "step": 369
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.755618929862976,
+      "learning_rate": 0.00024506290112925335,
+      "loss": 1.8781,
+      "step": 370
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6948919296264648,
+      "learning_rate": 0.00024502650721460926,
+      "loss": 1.3217,
+      "step": 371
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6204948425292969,
+      "learning_rate": 0.00024498998237537883,
+      "loss": 1.7845,
+      "step": 372
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.4599529504776,
+      "learning_rate": 0.00024495332665140337,
+      "loss": 1.8832,
+      "step": 373
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5336644649505615,
+      "learning_rate": 0.00024491654008266666,
+      "loss": 1.7582,
+      "step": 374
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6207972764968872,
+      "learning_rate": 0.00024487962270929546,
+      "loss": 1.6177,
+      "step": 375
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5989536046981812,
+      "learning_rate": 0.0002448425745715592,
+      "loss": 2.1292,
+      "step": 376
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7636903524398804,
+      "learning_rate": 0.0002448053957098699,
+      "loss": 1.8165,
+      "step": 377
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6132402420043945,
+      "learning_rate": 0.0002447680861647821,
+      "loss": 1.8227,
+      "step": 378
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8587868213653564,
+      "learning_rate": 0.0002447306459769929,
+      "loss": 2.365,
+      "step": 379
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.4449355602264404,
+      "learning_rate": 0.000244693075187342,
+      "loss": 1.8724,
+      "step": 380
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.2807527780532837,
+      "learning_rate": 0.0002446553738368116,
+      "loss": 1.7084,
+      "step": 381
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.2705549001693726,
+      "learning_rate": 0.0002446175419665261,
+      "loss": 1.5953,
+      "step": 382
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.444528341293335,
+      "learning_rate": 0.00024457957961775253,
+      "loss": 1.7351,
+      "step": 383
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6186937093734741,
+      "learning_rate": 0.00024454148683189996,
+      "loss": 2.1234,
+      "step": 384
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.2942836284637451,
+      "learning_rate": 0.00024450326365052,
+      "loss": 1.5367,
+      "step": 385
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.3657950162887573,
+      "learning_rate": 0.0002444649101153064,
+      "loss": 1.417,
+      "step": 386
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.2977917194366455,
+      "learning_rate": 0.0002444264262680951,
+      "loss": 1.6442,
+      "step": 387
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.5234405994415283,
+      "learning_rate": 0.0002443878121508641,
+      "loss": 1.7331,
+      "step": 388
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.580356478691101,
+      "learning_rate": 0.00024434906780573356,
+      "loss": 1.5812,
+      "step": 389
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5635511875152588,
+      "learning_rate": 0.0002443101932749658,
+      "loss": 1.7743,
+      "step": 390
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.4234613180160522,
+      "learning_rate": 0.00024427118860096504,
+      "loss": 1.9202,
+      "step": 391
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8113735914230347,
+      "learning_rate": 0.00024423205382627746,
+      "loss": 2.3395,
+      "step": 392
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5529452562332153,
+      "learning_rate": 0.0002441927889935911,
+      "loss": 1.6039,
+      "step": 393
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.3958035707473755,
+      "learning_rate": 0.000244153394145736,
+      "loss": 1.5429,
+      "step": 394
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.723780632019043,
+      "learning_rate": 0.0002441138693256839,
+      "loss": 1.7384,
+      "step": 395
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.403164029121399,
+      "learning_rate": 0.00024407421457654845,
+      "loss": 0.9538,
+      "step": 396
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.4130743741989136,
+      "learning_rate": 0.00024403442994158487,
+      "loss": 1.9526,
+      "step": 397
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.4831724166870117,
+      "learning_rate": 0.00024399451546419017,
+      "loss": 1.6449,
+      "step": 398
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6771824359893799,
+      "learning_rate": 0.00024395447118790293,
+      "loss": 1.7758,
+      "step": 399
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5150790214538574,
+      "learning_rate": 0.00024391429715640335,
+      "loss": 1.9961,
+      "step": 400
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5657678842544556,
+      "learning_rate": 0.0002438739934135131,
+      "loss": 1.5219,
+      "step": 401
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4939355850219727,
+      "learning_rate": 0.00024383356000319548,
+      "loss": 1.5112,
+      "step": 402
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6075009107589722,
+      "learning_rate": 0.0002437929969695551,
+      "loss": 1.4834,
+      "step": 403
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6066113710403442,
+      "learning_rate": 0.000243752304356838,
+      "loss": 1.655,
+      "step": 404
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.713575005531311,
+      "learning_rate": 0.0002437114822094316,
+      "loss": 1.8589,
+      "step": 405
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6600104570388794,
+      "learning_rate": 0.00024367053057186455,
+      "loss": 1.9319,
+      "step": 406
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8421694040298462,
+      "learning_rate": 0.0002436294494888068,
+      "loss": 2.1056,
+      "step": 407
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7146579027175903,
+      "learning_rate": 0.0002435882390050695,
+      "loss": 2.3835,
+      "step": 408
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.558237075805664,
+      "learning_rate": 0.0002435468991656049,
+      "loss": 1.972,
+      "step": 409
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.5981314182281494,
+      "learning_rate": 0.0002435054300155064,
+      "loss": 1.8716,
+      "step": 410
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.376088261604309,
+      "learning_rate": 0.00024346383160000847,
+      "loss": 1.6165,
+      "step": 411
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.5061323642730713,
+      "learning_rate": 0.0002434221039644865,
+      "loss": 1.6584,
+      "step": 412
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7581522464752197,
+      "learning_rate": 0.00024338024715445688,
+      "loss": 1.9282,
+      "step": 413
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.3032132387161255,
+      "learning_rate": 0.00024333826121557695,
+      "loss": 1.1937,
+      "step": 414
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.632356882095337,
+      "learning_rate": 0.0002432961461936448,
+      "loss": 2.075,
+      "step": 415
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.2259700298309326,
+      "learning_rate": 0.00024325390213459944,
+      "loss": 1.1513,
+      "step": 416
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.721431016921997,
+      "learning_rate": 0.00024321152908452054,
+      "loss": 1.5305,
+      "step": 417
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.5428179502487183,
+      "learning_rate": 0.00024316902708962848,
+      "loss": 1.7885,
+      "step": 418
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6045695543289185,
+      "learning_rate": 0.0002431263961962844,
+      "loss": 1.6599,
+      "step": 419
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6771968603134155,
+      "learning_rate": 0.00024308363645098984,
+      "loss": 1.6666,
+      "step": 420
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5328516960144043,
+      "learning_rate": 0.00024304074790038716,
+      "loss": 1.7956,
+      "step": 421
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6807280778884888,
+      "learning_rate": 0.00024299773059125896,
+      "loss": 1.7099,
+      "step": 422
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.528563380241394,
+      "learning_rate": 0.00024295458457052844,
+      "loss": 1.5751,
+      "step": 423
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.2582366466522217,
+      "learning_rate": 0.00024291130988525917,
+      "loss": 1.5967,
+      "step": 424
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5818768739700317,
+      "learning_rate": 0.00024286790658265507,
+      "loss": 2.0187,
+      "step": 425
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.647545576095581,
+      "learning_rate": 0.00024282437471006033,
+      "loss": 2.2781,
+      "step": 426
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.1883927583694458,
+      "learning_rate": 0.00024278071431495937,
+      "loss": 1.7889,
+      "step": 427
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8100999593734741,
+      "learning_rate": 0.00024273692544497688,
+      "loss": 1.6325,
+      "step": 428
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.4264518022537231,
+      "learning_rate": 0.0002426930081478776,
+      "loss": 1.8201,
+      "step": 429
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.2940385341644287,
+      "learning_rate": 0.00024264896247156643,
+      "loss": 1.508,
+      "step": 430
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.3940926790237427,
+      "learning_rate": 0.00024260478846408823,
+      "loss": 1.317,
+      "step": 431
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.3203697204589844,
+      "learning_rate": 0.00024256048617362792,
+      "loss": 1.5154,
+      "step": 432
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5709110498428345,
+      "learning_rate": 0.00024251605564851032,
+      "loss": 1.4274,
+      "step": 433
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.4669488668441772,
+      "learning_rate": 0.00024247149693720012,
+      "loss": 1.5945,
+      "step": 434
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9155356884002686,
+      "learning_rate": 0.00024242681008830184,
+      "loss": 1.727,
+      "step": 435
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.4303950071334839,
+      "learning_rate": 0.00024238199515055976,
+      "loss": 1.215,
+      "step": 436
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5403741598129272,
+      "learning_rate": 0.0002423370521728579,
+      "loss": 1.2804,
+      "step": 437
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0623433589935303,
+      "learning_rate": 0.00024229198120421993,
+      "loss": 1.5453,
+      "step": 438
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7625672817230225,
+      "learning_rate": 0.00024224678229380913,
+      "loss": 1.923,
+      "step": 439
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7964345216751099,
+      "learning_rate": 0.00024220145549092842,
+      "loss": 1.68,
+      "step": 440
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.8445159196853638,
+      "learning_rate": 0.00024215600084502006,
+      "loss": 1.8517,
+      "step": 441
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7241743803024292,
+      "learning_rate": 0.0002421104184056659,
+      "loss": 1.8859,
+      "step": 442
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.3813362121582031,
+      "learning_rate": 0.00024206470822258713,
+      "loss": 1.3435,
+      "step": 443
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.6187509298324585,
+      "learning_rate": 0.0002420188703456443,
+      "loss": 1.9744,
+      "step": 444
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.5142652988433838,
+      "learning_rate": 0.00024197290482483725,
+      "loss": 1.768,
+      "step": 445
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.3997564315795898,
+      "learning_rate": 0.000241926811710305,
+      "loss": 1.6986,
+      "step": 446
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.248969554901123,
+      "learning_rate": 0.00024188059105232585,
+      "loss": 1.2985,
+      "step": 447
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.3714370727539062,
+      "learning_rate": 0.00024183424290131716,
+      "loss": 1.626,
+      "step": 448
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.3766262531280518,
+      "learning_rate": 0.0002417877673078353,
+      "loss": 1.3913,
+      "step": 449
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.621702790260315,
+      "learning_rate": 0.00024174116432257583,
+      "loss": 1.7584,
+      "step": 450
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6640499830245972,
+      "learning_rate": 0.00024169443399637303,
+      "loss": 1.7288,
+      "step": 451
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7537161111831665,
+      "learning_rate": 0.00024164757638020032,
+      "loss": 1.8502,
+      "step": 452
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.614371657371521,
+      "learning_rate": 0.00024160059152516982,
+      "loss": 1.8407,
+      "step": 453
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.3671376705169678,
+      "learning_rate": 0.0002415534794825325,
+      "loss": 1.7035,
+      "step": 454
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.1287087202072144,
+      "learning_rate": 0.00024150624030367797,
+      "loss": 0.8822,
+      "step": 455
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.104029893875122,
+      "learning_rate": 0.0002414588740401347,
+      "loss": 2.4046,
+      "step": 456
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.5074779987335205,
+      "learning_rate": 0.00024141138074356962,
+      "loss": 1.4848,
+      "step": 457
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.8815460205078125,
+      "learning_rate": 0.00024136376046578835,
+      "loss": 1.759,
+      "step": 458
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.5019757747650146,
+      "learning_rate": 0.00024131601325873487,
+      "loss": 1.4991,
+      "step": 459
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.6193299293518066,
+      "learning_rate": 0.00024126813917449175,
+      "loss": 1.818,
+      "step": 460
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.4965165853500366,
+      "learning_rate": 0.0002412201382652799,
+      "loss": 1.4364,
+      "step": 461
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6111191511154175,
+      "learning_rate": 0.00024117201058345862,
+      "loss": 1.7969,
+      "step": 462
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6606050729751587,
+      "learning_rate": 0.00024112375618152537,
+      "loss": 1.8442,
+      "step": 463
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7113983631134033,
+      "learning_rate": 0.00024107537511211603,
+      "loss": 1.6806,
+      "step": 464
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.5785653591156006,
+      "learning_rate": 0.00024102686742800446,
+      "loss": 1.6221,
+      "step": 465
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.715955138206482,
+      "learning_rate": 0.0002409782331821027,
+      "loss": 1.8423,
+      "step": 466
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7139517068862915,
+      "learning_rate": 0.00024092947242746094,
+      "loss": 1.9045,
+      "step": 467
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.2441296577453613,
+      "learning_rate": 0.00024088058521726718,
+      "loss": 1.809,
+      "step": 468
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.487850546836853,
+      "learning_rate": 0.00024083157160484752,
+      "loss": 1.3382,
+      "step": 469
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.3797458410263062,
+      "learning_rate": 0.00024078243164366586,
+      "loss": 1.3186,
+      "step": 470
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.5861365795135498,
+      "learning_rate": 0.0002407331653873239,
+      "loss": 2.0695,
+      "step": 471
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.4714101552963257,
+      "learning_rate": 0.00024068377288956116,
+      "loss": 1.8704,
+      "step": 472
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.6222436428070068,
+      "learning_rate": 0.00024063425420425485,
+      "loss": 2.0155,
+      "step": 473
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7147741317749023,
+      "learning_rate": 0.00024058460938541982,
+      "loss": 2.2635,
+      "step": 474
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7251660823822021,
+      "learning_rate": 0.0002405348384872085,
+      "loss": 2.3124,
+      "step": 475
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.4105515480041504,
+      "learning_rate": 0.00024048494156391086,
+      "loss": 1.5038,
+      "step": 476
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.4747495651245117,
+      "learning_rate": 0.0002404349186699543,
+      "loss": 1.8607,
+      "step": 477
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.4962126016616821,
+      "learning_rate": 0.00024038476985990364,
+      "loss": 1.9751,
+      "step": 478
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.5216854810714722,
+      "learning_rate": 0.00024033449518846112,
+      "loss": 1.6452,
+      "step": 479
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.7051867246627808,
+      "learning_rate": 0.0002402840947104662,
+      "loss": 1.5337,
+      "step": 480
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.6586822271347046,
+      "learning_rate": 0.00024023356848089557,
+      "loss": 1.5478,
+      "step": 481
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.2107691764831543,
+      "learning_rate": 0.00024018291655486306,
+      "loss": 1.278,
+      "step": 482
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.4238454103469849,
+      "learning_rate": 0.00024013213898761975,
+      "loss": 1.6255,
+      "step": 483
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.3502719402313232,
+      "learning_rate": 0.0002400812358345536,
+      "loss": 1.3436,
+      "step": 484
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2329463958740234,
+      "learning_rate": 0.00024003020715118967,
+      "loss": 2.3051,
+      "step": 485
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6358450651168823,
+      "learning_rate": 0.00023997905299318983,
+      "loss": 2.2955,
+      "step": 486
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5185836553573608,
+      "learning_rate": 0.000239927773416353,
+      "loss": 1.5445,
+      "step": 487
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5232542753219604,
+      "learning_rate": 0.00023987636847661476,
+      "loss": 1.6744,
+      "step": 488
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.4239147901535034,
+      "learning_rate": 0.00023982483823004747,
+      "loss": 1.666,
+      "step": 489
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.3807685375213623,
+      "learning_rate": 0.00023977318273286018,
+      "loss": 1.5856,
+      "step": 490
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.573913812637329,
+      "learning_rate": 0.00023972140204139858,
+      "loss": 1.9832,
+      "step": 491
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.556381344795227,
+      "learning_rate": 0.0002396694962121449,
+      "loss": 1.4315,
+      "step": 492
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7060798406600952,
+      "learning_rate": 0.00023961746530171788,
+      "loss": 1.7671,
+      "step": 493
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.5967423915863037,
+      "learning_rate": 0.0002395653093668727,
+      "loss": 1.9734,
+      "step": 494
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.390377402305603,
+      "learning_rate": 0.00023951302846450087,
+      "loss": 1.4402,
+      "step": 495
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.330527663230896,
+      "learning_rate": 0.00023946062265163033,
+      "loss": 1.591,
+      "step": 496
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.6965402364730835,
+      "learning_rate": 0.00023940809198542512,
+      "loss": 2.3242,
+      "step": 497
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.51449716091156,
+      "learning_rate": 0.0002393554365231856,
+      "loss": 1.619,
+      "step": 498
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.6005566120147705,
+      "learning_rate": 0.00023930265632234816,
+      "loss": 2.0559,
+      "step": 499
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.4369415044784546,
+      "learning_rate": 0.00023924975144048533,
+      "loss": 1.4451,
+      "step": 500
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.4336689710617065,
+      "learning_rate": 0.00023919672193530565,
+      "loss": 1.7534,
+      "step": 501
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.449487566947937,
+      "learning_rate": 0.00023914356786465348,
+      "loss": 1.8109,
+      "step": 502
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.439719319343567,
+      "learning_rate": 0.00023909028928650918,
+      "loss": 1.6258,
+      "step": 503
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.702451229095459,
+      "learning_rate": 0.00023903688625898887,
+      "loss": 1.9753,
+      "step": 504
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.1217689514160156,
+      "learning_rate": 0.00023898335884034444,
+      "loss": 1.0806,
+      "step": 505
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.43269944190979,
+      "learning_rate": 0.00023892970708896342,
+      "loss": 2.2551,
+      "step": 506
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.3758916854858398,
+      "learning_rate": 0.00023887593106336903,
+      "loss": 1.6455,
+      "step": 507
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.348081350326538,
+      "learning_rate": 0.00023882203082221996,
+      "loss": 1.6481,
+      "step": 508
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.44756019115448,
+      "learning_rate": 0.00023876800642431048,
+      "loss": 1.6608,
+      "step": 509
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5149917602539062,
+      "learning_rate": 0.00023871385792857027,
+      "loss": 1.6434,
+      "step": 510
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.436800241470337,
+      "learning_rate": 0.00023865958539406434,
+      "loss": 1.9541,
+      "step": 511
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.563488483428955,
+      "learning_rate": 0.00023860518887999305,
+      "loss": 1.921,
+      "step": 512
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.2544896602630615,
+      "learning_rate": 0.00023855066844569194,
+      "loss": 1.285,
+      "step": 513
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.1815474033355713,
+      "learning_rate": 0.00023849602415063176,
+      "loss": 1.5958,
+      "step": 514
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.3179055452346802,
+      "learning_rate": 0.00023844125605441835,
+      "loss": 1.5887,
+      "step": 515
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.3224471807479858,
+      "learning_rate": 0.0002383863642167926,
+      "loss": 1.3453,
+      "step": 516
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.3743826150894165,
+      "learning_rate": 0.00023833134869763041,
+      "loss": 1.6145,
+      "step": 517
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7813752889633179,
+      "learning_rate": 0.00023827620955694248,
+      "loss": 1.5723,
+      "step": 518
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.457780361175537,
+      "learning_rate": 0.00023822094685487451,
+      "loss": 1.5773,
+      "step": 519
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.4680794477462769,
+      "learning_rate": 0.00023816556065170693,
+      "loss": 1.7291,
+      "step": 520
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5272331237792969,
+      "learning_rate": 0.00023811005100785471,
+      "loss": 1.6012,
+      "step": 521
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.6784194707870483,
+      "learning_rate": 0.00023805441798386778,
+      "loss": 1.6109,
+      "step": 522
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5598907470703125,
+      "learning_rate": 0.00023799866164043044,
+      "loss": 2.0627,
+      "step": 523
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.470335841178894,
+      "learning_rate": 0.0002379427820383615,
+      "loss": 1.8267,
+      "step": 524
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.677306890487671,
+      "learning_rate": 0.00023788677923861434,
+      "loss": 2.2845,
+      "step": 525
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.2650090456008911,
+      "learning_rate": 0.00023783065330227666,
+      "loss": 1.6407,
+      "step": 526
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.6766515970230103,
+      "learning_rate": 0.00023777440429057043,
+      "loss": 1.7039,
+      "step": 527
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.3705694675445557,
+      "learning_rate": 0.00023771803226485197,
+      "loss": 1.8251,
+      "step": 528
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.3645520210266113,
+      "learning_rate": 0.00023766153728661173,
+      "loss": 1.0754,
+      "step": 529
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.5679587125778198,
+      "learning_rate": 0.00023760491941747419,
+      "loss": 1.7131,
+      "step": 530
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.4037436246871948,
+      "learning_rate": 0.0002375481787191981,
+      "loss": 1.7182,
+      "step": 531
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.4753473997116089,
+      "learning_rate": 0.000237491315253676,
+      "loss": 1.9983,
+      "step": 532
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.307569980621338,
+      "learning_rate": 0.00023743432908293437,
+      "loss": 1.4774,
+      "step": 533
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.5248976945877075,
+      "learning_rate": 0.00023737722026913357,
+      "loss": 1.678,
+      "step": 534
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.6276731491088867,
+      "learning_rate": 0.00023731998887456778,
+      "loss": 1.9258,
+      "step": 535
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.5609790086746216,
+      "learning_rate": 0.0002372626349616649,
+      "loss": 1.7404,
+      "step": 536
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.3915233612060547,
+      "learning_rate": 0.0002372051585929863,
+      "loss": 1.5489,
+      "step": 537
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.6098966598510742,
+      "learning_rate": 0.0002371475598312271,
+      "loss": 1.7612,
+      "step": 538
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.3052818775177002,
+      "learning_rate": 0.00023708983873921586,
+      "loss": 1.2804,
+      "step": 539
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.3293620347976685,
+      "learning_rate": 0.00023703199537991466,
+      "loss": 1.4353,
+      "step": 540
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.4364746809005737,
+      "learning_rate": 0.00023697402981641877,
+      "loss": 1.4233,
+      "step": 541
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.7351473569869995,
+      "learning_rate": 0.00023691594211195695,
+      "loss": 1.9575,
+      "step": 542
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.3488695621490479,
+      "learning_rate": 0.00023685773232989105,
+      "loss": 2.0196,
+      "step": 543
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.455739140510559,
+      "learning_rate": 0.0002367994005337162,
+      "loss": 2.1026,
+      "step": 544
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.2512130737304688,
+      "learning_rate": 0.00023674094678706052,
+      "loss": 1.3881,
+      "step": 545
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6788513660430908,
+      "learning_rate": 0.0002366823711536852,
+      "loss": 1.5795,
+      "step": 546
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.0756875276565552,
+      "learning_rate": 0.00023662367369748442,
+      "loss": 1.3518,
+      "step": 547
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.5534878969192505,
+      "learning_rate": 0.00023656485448248516,
+      "loss": 1.4272,
+      "step": 548
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.276198387145996,
+      "learning_rate": 0.00023650591357284728,
+      "loss": 1.3384,
+      "step": 549
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.4616369009017944,
+      "learning_rate": 0.00023644685103286337,
+      "loss": 1.6083,
+      "step": 550
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.487240195274353,
+      "learning_rate": 0.00023638766692695869,
+      "loss": 1.2336,
+      "step": 551
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.3095462322235107,
+      "learning_rate": 0.00023632836131969105,
+      "loss": 1.3381,
+      "step": 552
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.327471971511841,
+      "learning_rate": 0.0002362689342757509,
+      "loss": 2.192,
+      "step": 553
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9238693714141846,
+      "learning_rate": 0.00023620938585996106,
+      "loss": 1.6526,
+      "step": 554
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6897307634353638,
+      "learning_rate": 0.00023614971613727684,
+      "loss": 1.6713,
+      "step": 555
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.526841163635254,
+      "learning_rate": 0.00023608992517278578,
+      "loss": 2.0541,
+      "step": 556
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5254669189453125,
+      "learning_rate": 0.00023603001303170763,
+      "loss": 1.3508,
+      "step": 557
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.57151198387146,
+      "learning_rate": 0.0002359699797793945,
+      "loss": 1.3197,
+      "step": 558
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5244841575622559,
+      "learning_rate": 0.00023590982548133045,
+      "loss": 1.4338,
+      "step": 559
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.4043896198272705,
+      "learning_rate": 0.0002358495502031316,
+      "loss": 1.9264,
+      "step": 560
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.4733903408050537,
+      "learning_rate": 0.00023578915401054607,
+      "loss": 1.59,
+      "step": 561
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7210224866867065,
+      "learning_rate": 0.0002357286369694539,
+      "loss": 1.7421,
+      "step": 562
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.291479468345642,
+      "learning_rate": 0.00023566799914586688,
+      "loss": 0.9787,
+      "step": 563
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6497067213058472,
+      "learning_rate": 0.00023560724060592858,
+      "loss": 1.5709,
+      "step": 564
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6274982690811157,
+      "learning_rate": 0.00023554636141591426,
+      "loss": 1.5853,
+      "step": 565
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.4311622381210327,
+      "learning_rate": 0.00023548536164223077,
+      "loss": 1.7003,
+      "step": 566
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6563549041748047,
+      "learning_rate": 0.00023542424135141656,
+      "loss": 1.4655,
+      "step": 567
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6957557201385498,
+      "learning_rate": 0.00023536300061014132,
+      "loss": 1.8792,
+      "step": 568
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.2087243795394897,
+      "learning_rate": 0.00023530163948520645,
+      "loss": 1.262,
+      "step": 569
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.3456618785858154,
+      "learning_rate": 0.00023524015804354437,
+      "loss": 1.6861,
+      "step": 570
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.5029468536376953,
+      "learning_rate": 0.00023517855635221896,
+      "loss": 1.7885,
+      "step": 571
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.4606847763061523,
+      "learning_rate": 0.00023511683447842514,
+      "loss": 1.4353,
+      "step": 572
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.4862562417984009,
+      "learning_rate": 0.00023505499248948898,
+      "loss": 1.7917,
+      "step": 573
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.3696837425231934,
+      "learning_rate": 0.00023499303045286751,
+      "loss": 1.2735,
+      "step": 574
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.4719946384429932,
+      "learning_rate": 0.00023493094843614877,
+      "loss": 1.867,
+      "step": 575
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.2641010284423828,
+      "learning_rate": 0.0002348687465070517,
+      "loss": 1.3491,
+      "step": 576
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.2275723218917847,
+      "learning_rate": 0.00023480642473342592,
+      "loss": 1.5747,
+      "step": 577
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.214805245399475,
+      "learning_rate": 0.00023474398318325192,
+      "loss": 1.5349,
+      "step": 578
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.4233876466751099,
+      "learning_rate": 0.0002346814219246407,
+      "loss": 2.0182,
+      "step": 579
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.363861322402954,
+      "learning_rate": 0.00023461874102583395,
+      "loss": 1.8423,
+      "step": 580
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.2210170030593872,
+      "learning_rate": 0.00023455594055520385,
+      "loss": 1.1782,
+      "step": 581
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.3371609449386597,
+      "learning_rate": 0.0002344930205812529,
+      "loss": 1.7994,
+      "step": 582
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.424874186515808,
+      "learning_rate": 0.00023442998117261416,
+      "loss": 1.637,
+      "step": 583
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.1814637184143066,
+      "learning_rate": 0.00023436682239805074,
+      "loss": 1.334,
+      "step": 584
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.3004404306411743,
+      "learning_rate": 0.00023430354432645606,
+      "loss": 1.1703,
+      "step": 585
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.332945704460144,
+      "learning_rate": 0.00023424014702685375,
+      "loss": 1.3147,
+      "step": 586
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.5192474126815796,
+      "learning_rate": 0.00023417663056839733,
+      "loss": 1.2995,
+      "step": 587
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.5328348875045776,
+      "learning_rate": 0.00023411299502037043,
+      "loss": 1.4266,
+      "step": 588
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7186061143875122,
+      "learning_rate": 0.00023404924045218652,
+      "loss": 1.2097,
+      "step": 589
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.892555832862854,
+      "learning_rate": 0.0002339853669333889,
+      "loss": 1.9445,
+      "step": 590
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.4742870330810547,
+      "learning_rate": 0.0002339213745336506,
+      "loss": 1.4,
+      "step": 591
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.3957617282867432,
+      "learning_rate": 0.00023385726332277441,
+      "loss": 1.0397,
+      "step": 592
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.274652123451233,
+      "learning_rate": 0.00023379303337069262,
+      "loss": 1.5659,
+      "step": 593
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0992603302001953,
+      "learning_rate": 0.00023372868474746714,
+      "loss": 1.8078,
+      "step": 594
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.4869073629379272,
+      "learning_rate": 0.00023366421752328922,
+      "loss": 1.3452,
+      "step": 595
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.4044032096862793,
+      "learning_rate": 0.00023359963176847957,
+      "loss": 1.3954,
+      "step": 596
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.3737423419952393,
+      "learning_rate": 0.0002335349275534881,
+      "loss": 1.2106,
+      "step": 597
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.022343873977661,
+      "learning_rate": 0.00023347010494889406,
+      "loss": 1.8948,
+      "step": 598
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.3906071186065674,
+      "learning_rate": 0.00023340516402540573,
+      "loss": 1.5172,
+      "step": 599
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.825240135192871,
+      "learning_rate": 0.0002333401048538605,
+      "loss": 1.0034,
+      "step": 600
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.4067100286483765,
+      "learning_rate": 0.00023327492750522475,
+      "loss": 1.7208,
+      "step": 601
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.222434401512146,
+      "learning_rate": 0.0002332096320505937,
+      "loss": 1.1459,
+      "step": 602
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7354257106781006,
+      "learning_rate": 0.00023314421856119154,
+      "loss": 1.5793,
+      "step": 603
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.384118914604187,
+      "learning_rate": 0.00023307868710837105,
+      "loss": 1.0095,
+      "step": 604
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.459686040878296,
+      "learning_rate": 0.00023301303776361378,
+      "loss": 1.528,
+      "step": 605
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.4759612083435059,
+      "learning_rate": 0.0002329472705985299,
+      "loss": 1.378,
+      "step": 606
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.5465447902679443,
+      "learning_rate": 0.00023288138568485802,
+      "loss": 1.5053,
+      "step": 607
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.2388064861297607,
+      "learning_rate": 0.0002328153830944652,
+      "loss": 1.2588,
+      "step": 608
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.4053964614868164,
+      "learning_rate": 0.00023274926289934688,
+      "loss": 1.014,
+      "step": 609
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.5368363857269287,
+      "learning_rate": 0.00023268302517162688,
+      "loss": 1.6706,
+      "step": 610
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.3434786796569824,
+      "learning_rate": 0.00023261666998355702,
+      "loss": 1.1686,
+      "step": 611
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.3576295375823975,
+      "learning_rate": 0.00023255019740751743,
+      "loss": 1.5393,
+      "step": 612
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.285962700843811,
+      "learning_rate": 0.0002324836075160162,
+      "loss": 1.7462,
+      "step": 613
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.13918399810791,
+      "learning_rate": 0.0002324169003816894,
+      "loss": 1.4822,
+      "step": 614
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.7046306133270264,
+      "learning_rate": 0.0002323500760773009,
+      "loss": 1.1588,
+      "step": 615
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6164623498916626,
+      "learning_rate": 0.00023228313467574263,
+      "loss": 1.9175,
+      "step": 616
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.4854605197906494,
+      "learning_rate": 0.00023221607625003402,
+      "loss": 1.7528,
+      "step": 617
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.6400645971298218,
+      "learning_rate": 0.00023214890087332218,
+      "loss": 1.9289,
+      "step": 618
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.5701171159744263,
+      "learning_rate": 0.00023208160861888186,
+      "loss": 1.3207,
+      "step": 619
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.5051968097686768,
+      "learning_rate": 0.00023201419956011526,
+      "loss": 1.4028,
+      "step": 620
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.1874357461929321,
+      "learning_rate": 0.000231946673770552,
+      "loss": 1.3708,
+      "step": 621
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.3110204935073853,
+      "learning_rate": 0.00023187903132384903,
+      "loss": 1.3477,
+      "step": 622
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 1.6304495334625244,
+      "eval_runtime": 28.33,
+      "eval_samples_per_second": 3.0,
+      "eval_steps_per_second": 1.518,
+      "step": 622
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.3583319187164307,
+      "learning_rate": 0.00023181127229379058,
+      "loss": 1.6669,
+      "step": 623
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.400608777999878,
+      "learning_rate": 0.00023174339675428792,
+      "loss": 1.1634,
+      "step": 624
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.403623342514038,
+      "learning_rate": 0.0002316754047793796,
+      "loss": 1.4847,
+      "step": 625
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.511208176612854,
+      "learning_rate": 0.00023160729644323099,
+      "loss": 1.3538,
+      "step": 626
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.1673988103866577,
+      "learning_rate": 0.00023153907182013456,
+      "loss": 1.1773,
+      "step": 627
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.5797700881958008,
+      "learning_rate": 0.00023147073098450953,
+      "loss": 1.6715,
+      "step": 628
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7258118391036987,
+      "learning_rate": 0.00023140227401090185,
+      "loss": 1.9853,
+      "step": 629
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.3571829795837402,
+      "learning_rate": 0.00023133370097398425,
+      "loss": 1.624,
+      "step": 630
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.462410807609558,
+      "learning_rate": 0.00023126501194855597,
+      "loss": 1.0793,
+      "step": 631
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.2532066106796265,
+      "learning_rate": 0.00023119620700954286,
+      "loss": 1.3601,
+      "step": 632
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.077949285507202,
+      "learning_rate": 0.0002311272862319971,
+      "loss": 1.9824,
+      "step": 633
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.5285255908966064,
+      "learning_rate": 0.00023105824969109736,
+      "loss": 1.3991,
+      "step": 634
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.544295310974121,
+      "learning_rate": 0.0002309890974621484,
+      "loss": 1.8974,
+      "step": 635
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.4555768966674805,
+      "learning_rate": 0.00023091982962058138,
+      "loss": 1.4607,
+      "step": 636
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.2902973890304565,
+      "learning_rate": 0.00023085044624195344,
+      "loss": 1.7837,
+      "step": 637
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.6322664022445679,
+      "learning_rate": 0.0002307809474019478,
+      "loss": 1.3603,
+      "step": 638
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.295804738998413,
+      "learning_rate": 0.00023071133317637357,
+      "loss": 1.3386,
+      "step": 639
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.3429510593414307,
+      "learning_rate": 0.0002306416036411658,
+      "loss": 1.2507,
+      "step": 640
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.6754704713821411,
+      "learning_rate": 0.0002305717588723852,
+      "loss": 1.4455,
+      "step": 641
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.1602115631103516,
+      "learning_rate": 0.00023050179894621835,
+      "loss": 1.6829,
+      "step": 642
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.412318468093872,
+      "learning_rate": 0.00023043172393897728,
+      "loss": 1.5495,
+      "step": 643
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.4912530183792114,
+      "learning_rate": 0.00023036153392709963,
+      "loss": 1.6909,
+      "step": 644
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.4303503036499023,
+      "learning_rate": 0.0002302912289871485,
+      "loss": 1.5794,
+      "step": 645
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.3841469287872314,
+      "learning_rate": 0.00023022080919581227,
+      "loss": 1.9355,
+      "step": 646
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.5403145551681519,
+      "learning_rate": 0.00023015027462990474,
+      "loss": 1.6106,
+      "step": 647
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.370869517326355,
+      "learning_rate": 0.00023007962536636474,
+      "loss": 1.4042,
+      "step": 648
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.6991840600967407,
+      "learning_rate": 0.00023000886148225635,
+      "loss": 1.5327,
+      "step": 649
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7101843357086182,
+      "learning_rate": 0.0002299379830547686,
+      "loss": 1.4779,
+      "step": 650
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.3180490732192993,
+      "learning_rate": 0.00022986699016121545,
+      "loss": 1.5081,
+      "step": 651
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.180501937866211,
+      "learning_rate": 0.0002297958828790358,
+      "loss": 1.0716,
+      "step": 652
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.322432279586792,
+      "learning_rate": 0.00022972466128579327,
+      "loss": 1.277,
+      "step": 653
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.6292730569839478,
+      "learning_rate": 0.00022965332545917623,
+      "loss": 1.631,
+      "step": 654
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7111241817474365,
+      "learning_rate": 0.00022958187547699748,
+      "loss": 1.3783,
+      "step": 655
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.3672375679016113,
+      "learning_rate": 0.0002295103114171946,
+      "loss": 1.0966,
+      "step": 656
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.34515380859375,
+      "learning_rate": 0.00022943863335782937,
+      "loss": 1.1398,
+      "step": 657
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.2631828784942627,
+      "learning_rate": 0.00022936684137708806,
+      "loss": 1.3126,
+      "step": 658
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.538662314414978,
+      "learning_rate": 0.00022929493555328117,
+      "loss": 1.7226,
+      "step": 659
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.458886981010437,
+      "learning_rate": 0.00022922291596484337,
+      "loss": 1.5284,
+      "step": 660
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.525400996208191,
+      "learning_rate": 0.00022915078269033346,
+      "loss": 1.3691,
+      "step": 661
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.4852648973464966,
+      "learning_rate": 0.00022907853580843415,
+      "loss": 1.6369,
+      "step": 662
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.523781418800354,
+      "learning_rate": 0.0002290061753979522,
+      "loss": 1.4643,
+      "step": 663
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.3667558431625366,
+      "learning_rate": 0.00022893370153781817,
+      "loss": 1.0672,
+      "step": 664
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8203182220458984,
+      "learning_rate": 0.00022886111430708627,
+      "loss": 2.1031,
+      "step": 665
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.458509922027588,
+      "learning_rate": 0.00022878841378493452,
+      "loss": 1.5407,
+      "step": 666
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.1977659463882446,
+      "learning_rate": 0.0002287156000506644,
+      "loss": 1.5075,
+      "step": 667
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.1007845401763916,
+      "learning_rate": 0.00022864267318370096,
+      "loss": 0.9576,
+      "step": 668
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.6353801488876343,
+      "learning_rate": 0.00022856963326359266,
+      "loss": 1.7101,
+      "step": 669
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.335712194442749,
+      "learning_rate": 0.00022849648037001115,
+      "loss": 1.0678,
+      "step": 670
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.375508189201355,
+      "learning_rate": 0.0002284232145827515,
+      "loss": 1.4913,
+      "step": 671
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.211701512336731,
+      "learning_rate": 0.00022834983598173182,
+      "loss": 1.0438,
+      "step": 672
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.7795249223709106,
+      "learning_rate": 0.00022827634464699323,
+      "loss": 1.9137,
+      "step": 673
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.4882405996322632,
+      "learning_rate": 0.00022820274065869992,
+      "loss": 1.5704,
+      "step": 674
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.3354179859161377,
+      "learning_rate": 0.00022812902409713893,
+      "loss": 1.3485,
+      "step": 675
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.5397255420684814,
+      "learning_rate": 0.00022805519504272005,
+      "loss": 2.0343,
+      "step": 676
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.4044562578201294,
+      "learning_rate": 0.00022798125357597582,
+      "loss": 1.7446,
+      "step": 677
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.284012794494629,
+      "learning_rate": 0.00022790719977756142,
+      "loss": 0.9859,
+      "step": 678
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.879234790802002,
+      "learning_rate": 0.00022783303372825447,
+      "loss": 1.7374,
+      "step": 679
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.278829574584961,
+      "learning_rate": 0.00022775875550895517,
+      "loss": 1.4937,
+      "step": 680
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.3762723207473755,
+      "learning_rate": 0.00022768436520068596,
+      "loss": 1.1091,
+      "step": 681
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.4809486865997314,
+      "learning_rate": 0.00022760986288459156,
+      "loss": 1.6401,
+      "step": 682
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.4666237831115723,
+      "learning_rate": 0.00022753524864193892,
+      "loss": 1.7877,
+      "step": 683
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.3258100748062134,
+      "learning_rate": 0.00022746052255411708,
+      "loss": 1.2407,
+      "step": 684
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.178125023841858,
+      "learning_rate": 0.000227385684702637,
+      "loss": 1.4674,
+      "step": 685
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.3732388019561768,
+      "learning_rate": 0.00022731073516913164,
+      "loss": 1.4729,
+      "step": 686
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.357108235359192,
+      "learning_rate": 0.00022723567403535572,
+      "loss": 1.428,
+      "step": 687
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.293455719947815,
+      "learning_rate": 0.00022716050138318574,
+      "loss": 1.3115,
+      "step": 688
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.1836376190185547,
+      "learning_rate": 0.00022708521729461985,
+      "loss": 1.2031,
+      "step": 689
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.3905551433563232,
+      "learning_rate": 0.00022700982185177765,
+      "loss": 1.8465,
+      "step": 690
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.2469663619995117,
+      "learning_rate": 0.00022693431513690033,
+      "loss": 1.5076,
+      "step": 691
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.2081923484802246,
+      "learning_rate": 0.00022685869723235042,
+      "loss": 1.5242,
+      "step": 692
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.2785173654556274,
+      "learning_rate": 0.00022678296822061173,
+      "loss": 1.3025,
+      "step": 693
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.3872441053390503,
+      "learning_rate": 0.00022670712818428923,
+      "loss": 1.7145,
+      "step": 694
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.1730290651321411,
+      "learning_rate": 0.00022663117720610903,
+      "loss": 1.1832,
+      "step": 695
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.0906507968902588,
+      "learning_rate": 0.00022655511536891825,
+      "loss": 1.378,
+      "step": 696
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.3207216262817383,
+      "learning_rate": 0.00022647894275568492,
+      "loss": 1.2161,
+      "step": 697
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.4555737972259521,
+      "learning_rate": 0.00022640265944949797,
+      "loss": 1.4499,
+      "step": 698
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.3717491626739502,
+      "learning_rate": 0.000226326265533567,
+      "loss": 1.7293,
+      "step": 699
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.2536697387695312,
+      "learning_rate": 0.00022624976109122225,
+      "loss": 1.4178,
+      "step": 700
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.550447940826416,
+      "learning_rate": 0.0002261731462059146,
+      "loss": 1.4674,
+      "step": 701
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.2245453596115112,
+      "learning_rate": 0.0002260964209612153,
+      "loss": 1.2625,
+      "step": 702
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.2388466596603394,
+      "learning_rate": 0.0002260195854408161,
+      "loss": 1.6174,
+      "step": 703
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.4022188186645508,
+      "learning_rate": 0.00022594263972852897,
+      "loss": 1.0605,
+      "step": 704
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.275978922843933,
+      "learning_rate": 0.00022586558390828604,
+      "loss": 1.3885,
+      "step": 705
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.493241786956787,
+      "learning_rate": 0.0002257884180641396,
+      "loss": 1.5634,
+      "step": 706
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.1711384057998657,
+      "learning_rate": 0.00022571114228026198,
+      "loss": 1.2732,
+      "step": 707
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.2927380800247192,
+      "learning_rate": 0.0002256337566409454,
+      "loss": 1.4771,
+      "step": 708
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.2375295162200928,
+      "learning_rate": 0.0002255562612306018,
+      "loss": 0.9908,
+      "step": 709
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.1654285192489624,
+      "learning_rate": 0.00022547865613376308,
+      "loss": 0.9158,
+      "step": 710
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.694615125656128,
+      "learning_rate": 0.0002254009414350806,
+      "loss": 1.517,
+      "step": 711
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.110138177871704,
+      "learning_rate": 0.0002253231172193254,
+      "loss": 1.2596,
+      "step": 712
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.5078182220458984,
+      "learning_rate": 0.00022524518357138788,
+      "loss": 1.6952,
+      "step": 713
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.7366492748260498,
+      "learning_rate": 0.00022516714057627783,
+      "loss": 1.8556,
+      "step": 714
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.4952809810638428,
+      "learning_rate": 0.0002250889883191244,
+      "loss": 1.5153,
+      "step": 715
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.2906392812728882,
+      "learning_rate": 0.00022501072688517576,
+      "loss": 1.0992,
+      "step": 716
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.26261568069458,
+      "learning_rate": 0.0002249323563597994,
+      "loss": 1.1001,
+      "step": 717
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.437803030014038,
+      "learning_rate": 0.00022485387682848155,
+      "loss": 1.519,
+      "step": 718
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.454239845275879,
+      "learning_rate": 0.00022477528837682755,
+      "loss": 1.258,
+      "step": 719
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.403212547302246,
+      "learning_rate": 0.00022469659109056143,
+      "loss": 1.6505,
+      "step": 720
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.2380965948104858,
+      "learning_rate": 0.000224617785055526,
+      "loss": 1.3611,
+      "step": 721
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.41748046875,
+      "learning_rate": 0.00022453887035768266,
+      "loss": 1.8791,
+      "step": 722
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.4731191396713257,
+      "learning_rate": 0.00022445984708311132,
+      "loss": 1.3579,
+      "step": 723
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.688100814819336,
+      "learning_rate": 0.00022438071531801038,
+      "loss": 1.5332,
+      "step": 724
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7024562358856201,
+      "learning_rate": 0.00022430147514869653,
+      "loss": 1.7091,
+      "step": 725
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.3762348890304565,
+      "learning_rate": 0.00022422212666160478,
+      "loss": 1.7395,
+      "step": 726
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.4019901752471924,
+      "learning_rate": 0.0002241426699432882,
+      "loss": 1.0866,
+      "step": 727
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.861675500869751,
+      "learning_rate": 0.00022406310508041796,
+      "loss": 2.1158,
+      "step": 728
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.3458441495895386,
+      "learning_rate": 0.00022398343215978322,
+      "loss": 1.4403,
+      "step": 729
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.5659409761428833,
+      "learning_rate": 0.000223903651268291,
+      "loss": 1.7634,
+      "step": 730
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.2884780168533325,
+      "learning_rate": 0.00022382376249296598,
+      "loss": 1.5543,
+      "step": 731
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.0735148191452026,
+      "learning_rate": 0.00022374376592095076,
+      "loss": 1.2122,
+      "step": 732
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.3894761800765991,
+      "learning_rate": 0.00022366366163950523,
+      "loss": 1.7313,
+      "step": 733
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.186671257019043,
+      "learning_rate": 0.00022358344973600705,
+      "loss": 0.9415,
+      "step": 734
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.2778778076171875,
+      "learning_rate": 0.00022350313029795112,
+      "loss": 1.871,
+      "step": 735
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.7073333263397217,
+      "learning_rate": 0.0002234227034129496,
+      "loss": 1.7262,
+      "step": 736
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.0448979139328003,
+      "learning_rate": 0.00022334216916873196,
+      "loss": 0.8838,
+      "step": 737
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0367178916931152,
+      "learning_rate": 0.00022326152765314472,
+      "loss": 1.5082,
+      "step": 738
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.5965039730072021,
+      "learning_rate": 0.0002231807789541515,
+      "loss": 1.6947,
+      "step": 739
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.2586615085601807,
+      "learning_rate": 0.00022309992315983263,
+      "loss": 1.2724,
+      "step": 740
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.498313546180725,
+      "learning_rate": 0.0002230189603583855,
+      "loss": 1.6083,
+      "step": 741
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.2010595798492432,
+      "learning_rate": 0.00022293789063812403,
+      "loss": 1.0541,
+      "step": 742
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.2845265865325928,
+      "learning_rate": 0.00022285671408747888,
+      "loss": 0.9787,
+      "step": 743
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.5691664218902588,
+      "learning_rate": 0.00022277543079499723,
+      "loss": 1.9072,
+      "step": 744
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8554484844207764,
+      "learning_rate": 0.00022269404084934269,
+      "loss": 1.6881,
+      "step": 745
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.4981218576431274,
+      "learning_rate": 0.00022261254433929514,
+      "loss": 1.2664,
+      "step": 746
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.5050944089889526,
+      "learning_rate": 0.00022253094135375075,
+      "loss": 1.3197,
+      "step": 747
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.4656778573989868,
+      "learning_rate": 0.0002224492319817219,
+      "loss": 1.5641,
+      "step": 748
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.433406114578247,
+      "learning_rate": 0.0002223674163123369,
+      "loss": 1.6133,
+      "step": 749
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.4999152421951294,
+      "learning_rate": 0.00022228549443484004,
+      "loss": 1.1804,
+      "step": 750
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.386148452758789,
+      "learning_rate": 0.00022220346643859153,
+      "loss": 1.4288,
+      "step": 751
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.305163025856018,
+      "learning_rate": 0.00022212133241306722,
+      "loss": 0.9577,
+      "step": 752
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.3146147727966309,
+      "learning_rate": 0.00022203909244785874,
+      "loss": 1.4475,
+      "step": 753
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.5211604833602905,
+      "learning_rate": 0.00022195674663267324,
+      "loss": 1.7606,
+      "step": 754
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.4717007875442505,
+      "learning_rate": 0.0002218742950573332,
+      "loss": 1.638,
+      "step": 755
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.1527776718139648,
+      "learning_rate": 0.0002217917378117767,
+      "loss": 1.3322,
+      "step": 756
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.599092721939087,
+      "learning_rate": 0.00022170907498605686,
+      "loss": 1.6299,
+      "step": 757
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.2002819776535034,
+      "learning_rate": 0.0002216263066703421,
+      "loss": 1.3424,
+      "step": 758
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.3727798461914062,
+      "learning_rate": 0.0002215434329549159,
+      "loss": 1.7718,
+      "step": 759
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.1956920623779297,
+      "learning_rate": 0.0002214604539301767,
+      "loss": 0.7773,
+      "step": 760
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.3804597854614258,
+      "learning_rate": 0.0002213773696866377,
+      "loss": 1.3221,
+      "step": 761
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.134540319442749,
+      "learning_rate": 0.00022129418031492705,
+      "loss": 1.1406,
+      "step": 762
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.599867582321167,
+      "learning_rate": 0.00022121088590578743,
+      "loss": 1.5812,
+      "step": 763
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.1916415691375732,
+      "learning_rate": 0.0002211274865500762,
+      "loss": 1.1639,
+      "step": 764
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.4034931659698486,
+      "learning_rate": 0.00022104398233876516,
+      "loss": 1.6485,
+      "step": 765
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.217153549194336,
+      "learning_rate": 0.0002209603733629404,
+      "loss": 1.2622,
+      "step": 766
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.2423417568206787,
+      "learning_rate": 0.00022087665971380244,
+      "loss": 1.3384,
+      "step": 767
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.2787408828735352,
+      "learning_rate": 0.00022079284148266587,
+      "loss": 1.3278,
+      "step": 768
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.0672158002853394,
+      "learning_rate": 0.0002207089187609594,
+      "loss": 1.4183,
+      "step": 769
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.4464133977890015,
+      "learning_rate": 0.00022062489164022565,
+      "loss": 2.0707,
+      "step": 770
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.4786409139633179,
+      "learning_rate": 0.00022054076021212128,
+      "loss": 1.3817,
+      "step": 771
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.325783133506775,
+      "learning_rate": 0.00022045652456841651,
+      "loss": 1.2509,
+      "step": 772
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.3716827630996704,
+      "learning_rate": 0.0002203721848009954,
+      "loss": 1.5183,
+      "step": 773
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.4279142618179321,
+      "learning_rate": 0.00022028774100185555,
+      "loss": 2.0254,
+      "step": 774
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.3607648611068726,
+      "learning_rate": 0.00022020319326310805,
+      "loss": 1.5698,
+      "step": 775
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.3437821865081787,
+      "learning_rate": 0.00022011854167697725,
+      "loss": 1.5756,
+      "step": 776
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.1076477766036987,
+      "learning_rate": 0.00022003378633580098,
+      "loss": 1.0518,
+      "step": 777
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.1801143884658813,
+      "learning_rate": 0.00021994892733203007,
+      "loss": 1.3424,
+      "step": 778
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.5795953273773193,
+      "learning_rate": 0.00021986396475822848,
+      "loss": 1.5052,
+      "step": 779
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.3943787813186646,
+      "learning_rate": 0.00021977889870707318,
+      "loss": 1.4324,
+      "step": 780
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.5282405614852905,
+      "learning_rate": 0.00021969372927135397,
+      "loss": 1.3772,
+      "step": 781
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.6440399885177612,
+      "learning_rate": 0.00021960845654397349,
+      "loss": 1.1698,
+      "step": 782
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.5308705568313599,
+      "learning_rate": 0.0002195230806179469,
+      "loss": 1.4512,
+      "step": 783
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.3149093389511108,
+      "learning_rate": 0.00021943760158640212,
+      "loss": 1.1414,
+      "step": 784
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.2626311779022217,
+      "learning_rate": 0.00021935201954257938,
+      "loss": 2.0695,
+      "step": 785
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.4505892992019653,
+      "learning_rate": 0.0002192663345798314,
+      "loss": 1.6039,
+      "step": 786
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.3184144496917725,
+      "learning_rate": 0.00021918054679162302,
+      "loss": 1.2249,
+      "step": 787
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.313861608505249,
+      "learning_rate": 0.00021909465627153142,
+      "loss": 1.5637,
+      "step": 788
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.7990531921386719,
+      "learning_rate": 0.00021900866311324568,
+      "loss": 1.7634,
+      "step": 789
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.6349966526031494,
+      "learning_rate": 0.00021892256741056694,
+      "loss": 1.9673,
+      "step": 790
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.3929771184921265,
+      "learning_rate": 0.00021883636925740813,
+      "loss": 1.3263,
+      "step": 791
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.237173318862915,
+      "learning_rate": 0.00021875006874779397,
+      "loss": 1.5712,
+      "step": 792
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.4733974933624268,
+      "learning_rate": 0.00021866366597586082,
+      "loss": 1.3899,
+      "step": 793
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.4081130027770996,
+      "learning_rate": 0.0002185771610358566,
+      "loss": 1.6359,
+      "step": 794
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.3193763494491577,
+      "learning_rate": 0.00021849055402214064,
+      "loss": 1.3589,
+      "step": 795
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.4265916347503662,
+      "learning_rate": 0.00021840384502918363,
+      "loss": 1.4326,
+      "step": 796
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.3795539140701294,
+      "learning_rate": 0.00021831703415156752,
+      "loss": 1.2139,
+      "step": 797
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.7029786109924316,
+      "learning_rate": 0.00021823012148398542,
+      "loss": 1.5998,
+      "step": 798
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.5365678071975708,
+      "learning_rate": 0.00021814310712124134,
+      "loss": 1.9567,
+      "step": 799
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.8077536821365356,
+      "learning_rate": 0.00021805599115825038,
+      "loss": 1.8023,
+      "step": 800
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.4349457025527954,
+      "learning_rate": 0.0002179687736900383,
+      "loss": 1.7043,
+      "step": 801
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.450710654258728,
+      "learning_rate": 0.00021788145481174183,
+      "loss": 1.2796,
+      "step": 802
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.2185072898864746,
+      "learning_rate": 0.00021779403461860805,
+      "loss": 1.1272,
+      "step": 803
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.6366219520568848,
+      "learning_rate": 0.00021770651320599473,
+      "loss": 1.5251,
+      "step": 804
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.445094347000122,
+      "learning_rate": 0.0002176188906693699,
+      "loss": 1.3265,
+      "step": 805
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.5272247791290283,
+      "learning_rate": 0.00021753116710431205,
+      "loss": 0.9833,
+      "step": 806
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.4574639797210693,
+      "learning_rate": 0.00021744334260650984,
+      "loss": 1.6412,
+      "step": 807
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.1877381801605225,
+      "learning_rate": 0.0002173554172717619,
+      "loss": 1.3086,
+      "step": 808
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.4031157493591309,
+      "learning_rate": 0.000217267391195977,
+      "loss": 1.587,
+      "step": 809
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.2422850131988525,
+      "learning_rate": 0.0002171792644751737,
+      "loss": 1.334,
+      "step": 810
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.3261579275131226,
+      "learning_rate": 0.00021709103720548052,
+      "loss": 1.2248,
+      "step": 811
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.0835561752319336,
+      "learning_rate": 0.00021700270948313532,
+      "loss": 0.9224,
+      "step": 812
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.3430688381195068,
+      "learning_rate": 0.00021691428140448594,
+      "loss": 1.2989,
+      "step": 813
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.2712783813476562,
+      "learning_rate": 0.00021682575306598934,
+      "loss": 1.4027,
+      "step": 814
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.2724041938781738,
+      "learning_rate": 0.00021673712456421202,
+      "loss": 1.2837,
+      "step": 815
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.2153583765029907,
+      "learning_rate": 0.00021664839599582974,
+      "loss": 1.3666,
+      "step": 816
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.2074302434921265,
+      "learning_rate": 0.00021655956745762742,
+      "loss": 1.4386,
+      "step": 817
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.372488021850586,
+      "learning_rate": 0.0002164706390464989,
+      "loss": 1.626,
+      "step": 818
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.1615793704986572,
+      "learning_rate": 0.00021638161085944706,
+      "loss": 1.1384,
+      "step": 819
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.1077085733413696,
+      "learning_rate": 0.00021629248299358367,
+      "loss": 0.9159,
+      "step": 820
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.2498594522476196,
+      "learning_rate": 0.00021620325554612912,
+      "loss": 1.2139,
+      "step": 821
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.6090998649597168,
+      "learning_rate": 0.00021611392861441243,
+      "loss": 1.9134,
+      "step": 822
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.2484623193740845,
+      "learning_rate": 0.00021602450229587123,
+      "loss": 1.2969,
+      "step": 823
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.4213321208953857,
+      "learning_rate": 0.0002159349766880515,
+      "loss": 1.717,
+      "step": 824
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.0547410249710083,
+      "learning_rate": 0.0002158453518886075,
+      "loss": 0.9849,
+      "step": 825
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.2111304998397827,
+      "learning_rate": 0.00021575562799530169,
+      "loss": 1.2678,
+      "step": 826
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.66452956199646,
+      "learning_rate": 0.0002156658051060047,
+      "loss": 1.4332,
+      "step": 827
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.3838409185409546,
+      "learning_rate": 0.0002155758833186951,
+      "loss": 1.1844,
+      "step": 828
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.5047062635421753,
+      "learning_rate": 0.0002154858627314593,
+      "loss": 1.7889,
+      "step": 829
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.286152720451355,
+      "learning_rate": 0.00021539574344249152,
+      "loss": 1.1978,
+      "step": 830
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.1742347478866577,
+      "learning_rate": 0.00021530552555009365,
+      "loss": 1.0202,
+      "step": 831
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.7456132173538208,
+      "learning_rate": 0.0002152152091526751,
+      "loss": 1.0362,
+      "step": 832
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.381914496421814,
+      "learning_rate": 0.00021512479434875276,
+      "loss": 1.4382,
+      "step": 833
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.2699590921401978,
+      "learning_rate": 0.00021503428123695082,
+      "loss": 1.512,
+      "step": 834
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.1738026142120361,
+      "learning_rate": 0.00021494366991600076,
+      "loss": 1.2607,
+      "step": 835
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.9550883769989014,
+      "learning_rate": 0.0002148529604847412,
+      "loss": 1.8124,
+      "step": 836
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.504042148590088,
+      "learning_rate": 0.00021476215304211765,
+      "loss": 1.4629,
+      "step": 837
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.0610342025756836,
+      "learning_rate": 0.00021467124768718269,
+      "loss": 0.9717,
+      "step": 838
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.478193759918213,
+      "learning_rate": 0.00021458024451909564,
+      "loss": 1.802,
+      "step": 839
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.218469262123108,
+      "learning_rate": 0.00021448914363712246,
+      "loss": 1.3854,
+      "step": 840
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.2758036851882935,
+      "learning_rate": 0.00021439794514063577,
+      "loss": 1.2681,
+      "step": 841
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.3164936304092407,
+      "learning_rate": 0.00021430664912911463,
+      "loss": 1.4467,
+      "step": 842
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.414792537689209,
+      "learning_rate": 0.0002142152557021445,
+      "loss": 1.5323,
+      "step": 843
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.1610770225524902,
+      "learning_rate": 0.00021412376495941705,
+      "loss": 1.4057,
+      "step": 844
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.2606096267700195,
+      "learning_rate": 0.00021403217700073017,
+      "loss": 1.19,
+      "step": 845
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.1834993362426758,
+      "learning_rate": 0.0002139404919259877,
+      "loss": 1.5106,
+      "step": 846
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.2146964073181152,
+      "learning_rate": 0.00021384870983519956,
+      "loss": 1.4995,
+      "step": 847
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.208569049835205,
+      "learning_rate": 0.0002137568308284813,
+      "loss": 1.0893,
+      "step": 848
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.429599642753601,
+      "learning_rate": 0.00021366485500605437,
+      "loss": 1.1302,
+      "step": 849
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.868918538093567,
+      "learning_rate": 0.0002135727824682457,
+      "loss": 2.0047,
+      "step": 850
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.1158243417739868,
+      "learning_rate": 0.00021348061331548777,
+      "loss": 1.1251,
+      "step": 851
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.1816633939743042,
+      "learning_rate": 0.00021338834764831843,
+      "loss": 1.0859,
+      "step": 852
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.3078497648239136,
+      "learning_rate": 0.0002132959855673809,
+      "loss": 1.3654,
+      "step": 853
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.5226352214813232,
+      "learning_rate": 0.00021320352717342335,
+      "loss": 1.3744,
+      "step": 854
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.194767713546753,
+      "learning_rate": 0.0002131109725672992,
+      "loss": 1.1332,
+      "step": 855
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.292400598526001,
+      "learning_rate": 0.00021301832184996686,
+      "loss": 1.1676,
+      "step": 856
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.307769536972046,
+      "learning_rate": 0.00021292557512248933,
+      "loss": 1.3764,
+      "step": 857
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.5784074068069458,
+      "learning_rate": 0.00021283273248603457,
+      "loss": 1.8682,
+      "step": 858
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.4524656534194946,
+      "learning_rate": 0.00021273979404187507,
+      "loss": 1.355,
+      "step": 859
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.2774903774261475,
+      "learning_rate": 0.0002126467598913878,
+      "loss": 1.1275,
+      "step": 860
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.364617943763733,
+      "learning_rate": 0.00021255363013605419,
+      "loss": 1.7862,
+      "step": 861
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.259527325630188,
+      "learning_rate": 0.0002124604048774599,
+      "loss": 1.3798,
+      "step": 862
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.6640812158584595,
+      "learning_rate": 0.0002123670842172948,
+      "loss": 1.8239,
+      "step": 863
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.2190797328948975,
+      "learning_rate": 0.0002122736682573528,
+      "loss": 1.0124,
+      "step": 864
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.341090202331543,
+      "learning_rate": 0.00021218015709953175,
+      "loss": 0.985,
+      "step": 865
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.3625142574310303,
+      "learning_rate": 0.00021208655084583344,
+      "loss": 1.0811,
+      "step": 866
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.861794352531433,
+      "learning_rate": 0.0002119928495983633,
+      "loss": 1.5199,
+      "step": 867
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.5730104446411133,
+      "learning_rate": 0.00021189905345933032,
+      "loss": 1.4189,
+      "step": 868
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.224785566329956,
+      "learning_rate": 0.0002118051625310471,
+      "loss": 1.2948,
+      "step": 869
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.4107848405838013,
+      "learning_rate": 0.00021171117691592964,
+      "loss": 1.3929,
+      "step": 870
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.16511070728302,
+      "learning_rate": 0.00021161709671649721,
+      "loss": 1.0999,
+      "step": 871
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.30039119720459,
+      "learning_rate": 0.00021152292203537216,
+      "loss": 1.6013,
+      "step": 872
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.5170789957046509,
+      "learning_rate": 0.00021142865297528004,
+      "loss": 1.8839,
+      "step": 873
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.2411909103393555,
+      "learning_rate": 0.00021133428963904927,
+      "loss": 1.4265,
+      "step": 874
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.0655996799468994,
+      "learning_rate": 0.00021123983212961112,
+      "loss": 1.2643,
+      "step": 875
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.1014069318771362,
+      "learning_rate": 0.00021114528054999953,
+      "loss": 1.0326,
+      "step": 876
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.3042875528335571,
+      "learning_rate": 0.0002110506350033512,
+      "loss": 1.5919,
+      "step": 877
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9417317509651184,
+      "learning_rate": 0.00021095589559290518,
+      "loss": 0.8426,
+      "step": 878
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.341861605644226,
+      "learning_rate": 0.00021086106242200296,
+      "loss": 1.2377,
+      "step": 879
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.2200474739074707,
+      "learning_rate": 0.0002107661355940883,
+      "loss": 1.4793,
+      "step": 880
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.4158042669296265,
+      "learning_rate": 0.00021067111521270713,
+      "loss": 1.752,
+      "step": 881
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.4061779975891113,
+      "learning_rate": 0.0002105760013815074,
+      "loss": 1.089,
+      "step": 882
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.3693805932998657,
+      "learning_rate": 0.00021048079420423908,
+      "loss": 1.4861,
+      "step": 883
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.4483033418655396,
+      "learning_rate": 0.00021038549378475386,
+      "loss": 1.4553,
+      "step": 884
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.283494234085083,
+      "learning_rate": 0.00021029010022700518,
+      "loss": 1.2053,
+      "step": 885
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.5050575733184814,
+      "learning_rate": 0.00021019461363504805,
+      "loss": 1.2142,
+      "step": 886
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.2599867582321167,
+      "learning_rate": 0.00021009903411303902,
+      "loss": 1.3242,
+      "step": 887
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.2808127403259277,
+      "learning_rate": 0.00021000336176523593,
+      "loss": 1.1527,
+      "step": 888
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.145213007926941,
+      "learning_rate": 0.00020990759669599799,
+      "loss": 1.1483,
+      "step": 889
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.1295148134231567,
+      "learning_rate": 0.0002098117390097854,
+      "loss": 1.0159,
+      "step": 890
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.159382939338684,
+      "learning_rate": 0.00020971578881115948,
+      "loss": 1.1626,
+      "step": 891
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.6490403413772583,
+      "learning_rate": 0.00020961974620478248,
+      "loss": 1.7539,
+      "step": 892
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.517959475517273,
+      "learning_rate": 0.0002095236112954174,
+      "loss": 1.6324,
+      "step": 893
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.4953640699386597,
+      "learning_rate": 0.0002094273841879279,
+      "loss": 1.4384,
+      "step": 894
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.3297251462936401,
+      "learning_rate": 0.00020933106498727825,
+      "loss": 1.1462,
+      "step": 895
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.2726671695709229,
+      "learning_rate": 0.00020923465379853323,
+      "loss": 1.0989,
+      "step": 896
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.1799503564834595,
+      "learning_rate": 0.00020913815072685785,
+      "loss": 0.9603,
+      "step": 897
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.2199748754501343,
+      "learning_rate": 0.00020904155587751742,
+      "loss": 1.149,
+      "step": 898
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.8773846626281738,
+      "learning_rate": 0.00020894486935587732,
+      "loss": 0.8965,
+      "step": 899
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.669113039970398,
+      "learning_rate": 0.00020884809126740295,
+      "loss": 1.7323,
+      "step": 900
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.3780359029769897,
+      "learning_rate": 0.00020875122171765958,
+      "loss": 1.47,
+      "step": 901
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.1259912252426147,
+      "learning_rate": 0.00020865426081231227,
+      "loss": 1.3505,
+      "step": 902
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.4640130996704102,
+      "learning_rate": 0.00020855720865712568,
+      "loss": 1.6051,
+      "step": 903
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.0640982389450073,
+      "learning_rate": 0.00020846006535796407,
+      "loss": 1.109,
+      "step": 904
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.1131609678268433,
+      "learning_rate": 0.00020836283102079108,
+      "loss": 1.1407,
+      "step": 905
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.1612396240234375,
+      "learning_rate": 0.00020826550575166964,
+      "loss": 1.2,
+      "step": 906
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.6626241207122803,
+      "learning_rate": 0.00020816808965676186,
+      "loss": 1.8475,
+      "step": 907
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.438838005065918,
+      "learning_rate": 0.000208070582842329,
+      "loss": 1.4186,
+      "step": 908
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.4505665302276611,
+      "learning_rate": 0.00020797298541473127,
+      "loss": 1.6378,
+      "step": 909
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.3546316623687744,
+      "learning_rate": 0.00020787529748042758,
+      "loss": 1.3801,
+      "step": 910
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.5958383083343506,
+      "learning_rate": 0.0002077775191459757,
+      "loss": 1.7361,
+      "step": 911
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.6373250484466553,
+      "learning_rate": 0.000207679650518032,
+      "loss": 1.8638,
+      "step": 912
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.126513957977295,
+      "learning_rate": 0.00020758169170335123,
+      "loss": 1.2306,
+      "step": 913
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.2469696998596191,
+      "learning_rate": 0.0002074836428087867,
+      "loss": 1.4286,
+      "step": 914
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.3433685302734375,
+      "learning_rate": 0.00020738550394128983,
+      "loss": 1.1645,
+      "step": 915
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.274887204170227,
+      "learning_rate": 0.0002072872752079102,
+      "loss": 1.4313,
+      "step": 916
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.2697557210922241,
+      "learning_rate": 0.0002071889567157955,
+      "loss": 1.1121,
+      "step": 917
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.6256011724472046,
+      "learning_rate": 0.00020709054857219122,
+      "loss": 2.3053,
+      "step": 918
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.4982554912567139,
+      "learning_rate": 0.00020699205088444073,
+      "loss": 1.5737,
+      "step": 919
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.3380775451660156,
+      "learning_rate": 0.00020689346375998504,
+      "loss": 1.3681,
+      "step": 920
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.1807115077972412,
+      "learning_rate": 0.0002067947873063627,
+      "loss": 1.121,
+      "step": 921
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.6416897773742676,
+      "learning_rate": 0.0002066960216312097,
+      "loss": 1.585,
+      "step": 922
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.310311198234558,
+      "learning_rate": 0.00020659716684225942,
+      "loss": 1.0295,
+      "step": 923
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.283509373664856,
+      "learning_rate": 0.00020649822304734234,
+      "loss": 0.8625,
+      "step": 924
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.6311004161834717,
+      "learning_rate": 0.00020639919035438616,
+      "loss": 1.5964,
+      "step": 925
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.4839829206466675,
+      "learning_rate": 0.00020630006887141533,
+      "loss": 1.3802,
+      "step": 926
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.6016814708709717,
+      "learning_rate": 0.00020620085870655142,
+      "loss": 1.3962,
+      "step": 927
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.3555843830108643,
+      "learning_rate": 0.00020610155996801256,
+      "loss": 1.3931,
+      "step": 928
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.5287679433822632,
+      "learning_rate": 0.00020600217276411353,
+      "loss": 1.0829,
+      "step": 929
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.1010750532150269,
+      "learning_rate": 0.00020590269720326568,
+      "loss": 0.9906,
+      "step": 930
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.2286770343780518,
+      "learning_rate": 0.00020580313339397662,
+      "loss": 1.099,
+      "step": 931
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.4633378982543945,
+      "learning_rate": 0.0002057034814448503,
+      "loss": 1.2768,
+      "step": 932
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.0793206691741943,
+      "learning_rate": 0.0002056037414645868,
+      "loss": 1.5472,
+      "step": 933
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.041421890258789,
+      "learning_rate": 0.00020550391356198222,
+      "loss": 1.0693,
+      "step": 934
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.7149021625518799,
+      "learning_rate": 0.00020540399784592852,
+      "loss": 1.6992,
+      "step": 935
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.1274030208587646,
+      "learning_rate": 0.00020530399442541356,
+      "loss": 1.1349,
+      "step": 936
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.6209064722061157,
+      "learning_rate": 0.00020520390340952074,
+      "loss": 1.6395,
+      "step": 937
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.0727425813674927,
+      "learning_rate": 0.00020510372490742913,
+      "loss": 1.0555,
+      "step": 938
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.1689037084579468,
+      "learning_rate": 0.00020500345902841311,
+      "loss": 1.4163,
+      "step": 939
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.1197943687438965,
+      "learning_rate": 0.00020490310588184242,
+      "loss": 1.669,
+      "step": 940
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.0973910093307495,
+      "learning_rate": 0.00020480266557718201,
+      "loss": 1.2043,
+      "step": 941
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.459172010421753,
+      "learning_rate": 0.00020470213822399198,
+      "loss": 1.6599,
+      "step": 942
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.4092894792556763,
+      "learning_rate": 0.00020460152393192717,
+      "loss": 1.5244,
+      "step": 943
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.0109457969665527,
+      "learning_rate": 0.0002045008228107374,
+      "loss": 1.0698,
+      "step": 944
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.1727893352508545,
+      "learning_rate": 0.00020440003497026722,
+      "loss": 0.8874,
+      "step": 945
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.4926542043685913,
+      "learning_rate": 0.0002042991605204557,
+      "loss": 1.1908,
+      "step": 946
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.2154324054718018,
+      "learning_rate": 0.00020419819957133642,
+      "loss": 1.0055,
+      "step": 947
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.1640907526016235,
+      "learning_rate": 0.00020409715223303733,
+      "loss": 1.0244,
+      "step": 948
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.094460129737854,
+      "learning_rate": 0.00020399601861578054,
+      "loss": 1.086,
+      "step": 949
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.0807240009307861,
+      "learning_rate": 0.0002038947988298824,
+      "loss": 1.0952,
+      "step": 950
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.2182865142822266,
+      "learning_rate": 0.00020379349298575312,
+      "loss": 0.8286,
+      "step": 951
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.2438262701034546,
+      "learning_rate": 0.0002036921011938968,
+      "loss": 1.449,
+      "step": 952
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.0871261358261108,
+      "learning_rate": 0.00020359062356491142,
+      "loss": 1.005,
+      "step": 953
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.3494447469711304,
+      "learning_rate": 0.0002034890602094885,
+      "loss": 1.599,
+      "step": 954
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.3108654022216797,
+      "learning_rate": 0.00020338741123841297,
+      "loss": 1.3039,
+      "step": 955
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.4159644842147827,
+      "learning_rate": 0.00020328567676256335,
+      "loss": 1.3015,
+      "step": 956
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.267722249031067,
+      "learning_rate": 0.00020318385689291126,
+      "loss": 1.1788,
+      "step": 957
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.2124974727630615,
+      "learning_rate": 0.00020308195174052157,
+      "loss": 1.3477,
+      "step": 958
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.7106714248657227,
+      "learning_rate": 0.00020297996141655216,
+      "loss": 1.7461,
+      "step": 959
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.3902013301849365,
+      "learning_rate": 0.00020287788603225377,
+      "loss": 1.3323,
+      "step": 960
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.9589385390281677,
+      "learning_rate": 0.00020277572569896995,
+      "loss": 0.7713,
+      "step": 961
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.223151683807373,
+      "learning_rate": 0.00020267348052813692,
+      "loss": 1.2357,
+      "step": 962
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.9997661709785461,
+      "learning_rate": 0.0002025711506312834,
+      "loss": 0.9083,
+      "step": 963
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.2039114236831665,
+      "learning_rate": 0.0002024687361200306,
+      "loss": 1.0277,
+      "step": 964
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.2620604038238525,
+      "learning_rate": 0.00020236623710609202,
+      "loss": 0.9147,
+      "step": 965
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.603683352470398,
+      "learning_rate": 0.00020226365370127323,
+      "loss": 1.3692,
+      "step": 966
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.1889077425003052,
+      "learning_rate": 0.00020216098601747197,
+      "loss": 1.3586,
+      "step": 967
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.9763696193695068,
+      "learning_rate": 0.00020205823416667788,
+      "loss": 1.0737,
+      "step": 968
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.9170948266983032,
+      "learning_rate": 0.00020195539826097238,
+      "loss": 1.7105,
+      "step": 969
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.3122986555099487,
+      "learning_rate": 0.00020185247841252856,
+      "loss": 1.6161,
+      "step": 970
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.4224249124526978,
+      "learning_rate": 0.00020174947473361118,
+      "loss": 1.5831,
+      "step": 971
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.293426752090454,
+      "learning_rate": 0.00020164638733657637,
+      "loss": 1.3102,
+      "step": 972
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.07170569896698,
+      "learning_rate": 0.0002015432163338715,
+      "loss": 0.8819,
+      "step": 973
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.181535005569458,
+      "learning_rate": 0.00020143996183803525,
+      "loss": 1.1775,
+      "step": 974
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.3837966918945312,
+      "learning_rate": 0.00020133662396169743,
+      "loss": 1.3791,
+      "step": 975
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.1325147151947021,
+      "learning_rate": 0.00020123320281757856,
+      "loss": 0.9244,
+      "step": 976
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.8120399713516235,
+      "learning_rate": 0.0002011296985184903,
+      "loss": 1.3342,
+      "step": 977
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.260125994682312,
+      "learning_rate": 0.0002010261111773347,
+      "loss": 1.3665,
+      "step": 978
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.342794418334961,
+      "learning_rate": 0.00020092244090710472,
+      "loss": 1.2215,
+      "step": 979
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.5904815196990967,
+      "learning_rate": 0.0002008186878208834,
+      "loss": 1.071,
+      "step": 980
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.2493515014648438,
+      "learning_rate": 0.00020071485203184455,
+      "loss": 1.5185,
+      "step": 981
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.2379603385925293,
+      "learning_rate": 0.00020061093365325184,
+      "loss": 1.239,
+      "step": 982
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.4626051187515259,
+      "learning_rate": 0.00020050693279845914,
+      "loss": 1.5534,
+      "step": 983
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.3357350826263428,
+      "learning_rate": 0.0002004028495809103,
+      "loss": 1.8324,
+      "step": 984
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.4206643104553223,
+      "learning_rate": 0.0002002986841141391,
+      "loss": 1.251,
+      "step": 985
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.436003565788269,
+      "learning_rate": 0.00020019443651176884,
+      "loss": 1.7747,
+      "step": 986
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.350991129875183,
+      "learning_rate": 0.0002000901068875126,
+      "loss": 1.5258,
+      "step": 987
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.1044528484344482,
+      "learning_rate": 0.00019998569535517276,
+      "loss": 1.269,
+      "step": 988
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.4368743896484375,
+      "learning_rate": 0.00019988120202864125,
+      "loss": 1.7985,
+      "step": 989
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.080803394317627,
+      "learning_rate": 0.00019977662702189904,
+      "loss": 0.8469,
+      "step": 990
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.4379607439041138,
+      "learning_rate": 0.00019967197044901618,
+      "loss": 1.48,
+      "step": 991
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.1783312559127808,
+      "learning_rate": 0.00019956723242415194,
+      "loss": 1.4203,
+      "step": 992
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.2120602130889893,
+      "learning_rate": 0.0001994624130615542,
+      "loss": 1.2925,
+      "step": 993
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.0752936601638794,
+      "learning_rate": 0.00019935751247555954,
+      "loss": 1.4643,
+      "step": 994
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.1408238410949707,
+      "learning_rate": 0.0001992525307805934,
+      "loss": 1.3049,
+      "step": 995
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.1272218227386475,
+      "learning_rate": 0.0001991474680911694,
+      "loss": 1.0908,
+      "step": 996
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.525564432144165,
+      "learning_rate": 0.0001990423245218897,
+      "loss": 1.2493,
+      "step": 997
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.3659435510635376,
+      "learning_rate": 0.0001989371001874446,
+      "loss": 1.402,
+      "step": 998
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.344510793685913,
+      "learning_rate": 0.00019883179520261245,
+      "loss": 1.519,
+      "step": 999
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.455416202545166,
+      "learning_rate": 0.0001987264096822598,
+      "loss": 1.6428,
+      "step": 1000
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.206539273262024,
+      "learning_rate": 0.00019862094374134074,
+      "loss": 1.2198,
+      "step": 1001
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.2967417240142822,
+      "learning_rate": 0.0001985153974948973,
+      "loss": 1.4467,
+      "step": 1002
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.3688677549362183,
+      "learning_rate": 0.00019840977105805896,
+      "loss": 1.5175,
+      "step": 1003
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.4224023818969727,
+      "learning_rate": 0.00019830406454604289,
+      "loss": 1.3388,
+      "step": 1004
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.1016898155212402,
+      "learning_rate": 0.00019819827807415332,
+      "loss": 1.256,
+      "step": 1005
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.9830572605133057,
+      "learning_rate": 0.00019809241175778195,
+      "loss": 1.0293,
+      "step": 1006
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.3320682048797607,
+      "learning_rate": 0.00019798646571240742,
+      "loss": 1.6151,
+      "step": 1007
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.203979730606079,
+      "learning_rate": 0.00019788044005359534,
+      "loss": 1.0101,
+      "step": 1008
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.121233344078064,
+      "learning_rate": 0.0001977743348969983,
+      "loss": 0.9686,
+      "step": 1009
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.2444918155670166,
+      "learning_rate": 0.0001976681503583555,
+      "loss": 1.5227,
+      "step": 1010
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.322102665901184,
+      "learning_rate": 0.0001975618865534927,
+      "loss": 1.6012,
+      "step": 1011
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.2791893482208252,
+      "learning_rate": 0.0001974555435983222,
+      "loss": 1.3951,
+      "step": 1012
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.2599291801452637,
+      "learning_rate": 0.00019734912160884255,
+      "loss": 1.5789,
+      "step": 1013
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.2614623308181763,
+      "learning_rate": 0.0001972426207011387,
+      "loss": 1.2646,
+      "step": 1014
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.9194993376731873,
+      "learning_rate": 0.0001971360409913814,
+      "loss": 1.1464,
+      "step": 1015
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.739387035369873,
+      "learning_rate": 0.00019702938259582754,
+      "loss": 1.6735,
+      "step": 1016
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.1382012367248535,
+      "learning_rate": 0.00019692264563081984,
+      "loss": 1.2768,
+      "step": 1017
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.9404625296592712,
+      "learning_rate": 0.00019681583021278673,
+      "loss": 1.1835,
+      "step": 1018
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.2209988832473755,
+      "learning_rate": 0.00019670893645824212,
+      "loss": 0.9123,
+      "step": 1019
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.1911715269088745,
+      "learning_rate": 0.00019660196448378542,
+      "loss": 1.4704,
+      "step": 1020
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.1453022956848145,
+      "learning_rate": 0.00019649491440610137,
+      "loss": 1.1527,
+      "step": 1021
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.9455304145812988,
+      "learning_rate": 0.0001963877863419599,
+      "loss": 0.9554,
+      "step": 1022
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.5460333824157715,
+      "learning_rate": 0.00019628058040821607,
+      "loss": 1.5023,
+      "step": 1023
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.0515046119689941,
+      "learning_rate": 0.00019617329672180975,
+      "loss": 1.3552,
+      "step": 1024
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.4096556901931763,
+      "learning_rate": 0.00019606593539976574,
+      "loss": 1.3264,
+      "step": 1025
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.2442529201507568,
+      "learning_rate": 0.00019595849655919343,
+      "loss": 0.8898,
+      "step": 1026
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8789858818054199,
+      "learning_rate": 0.00019585098031728684,
+      "loss": 0.9477,
+      "step": 1027
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.2422945499420166,
+      "learning_rate": 0.0001957433867913244,
+      "loss": 1.1831,
+      "step": 1028
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.2705669403076172,
+      "learning_rate": 0.00019563571609866883,
+      "loss": 1.4243,
+      "step": 1029
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.4680469036102295,
+      "learning_rate": 0.000195527968356767,
+      "loss": 1.2629,
+      "step": 1030
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.369444727897644,
+      "learning_rate": 0.0001954201436831499,
+      "loss": 1.6357,
+      "step": 1031
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.4526840448379517,
+      "learning_rate": 0.00019531224219543233,
+      "loss": 1.3395,
+      "step": 1032
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.1711682081222534,
+      "learning_rate": 0.00019520426401131304,
+      "loss": 1.0378,
+      "step": 1033
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.085681676864624,
+      "learning_rate": 0.0001950962092485742,
+      "loss": 1.3059,
+      "step": 1034
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.2222062349319458,
+      "learning_rate": 0.00019498807802508173,
+      "loss": 1.1306,
+      "step": 1035
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.879671573638916,
+      "learning_rate": 0.00019487987045878485,
+      "loss": 0.7041,
+      "step": 1036
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.091941475868225,
+      "learning_rate": 0.00019477158666771617,
+      "loss": 0.9351,
+      "step": 1037
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.0203659534454346,
+      "learning_rate": 0.00019466322676999122,
+      "loss": 1.1873,
+      "step": 1038
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.1126773357391357,
+      "learning_rate": 0.00019455479088380874,
+      "loss": 0.9875,
+      "step": 1039
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.186734676361084,
+      "learning_rate": 0.00019444627912745027,
+      "loss": 1.146,
+      "step": 1040
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.2858699560165405,
+      "learning_rate": 0.00019433769161928023,
+      "loss": 0.932,
+      "step": 1041
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.0115717649459839,
+      "learning_rate": 0.00019422902847774544,
+      "loss": 0.9152,
+      "step": 1042
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.0166630744934082,
+      "learning_rate": 0.00019412028982137547,
+      "loss": 0.981,
+      "step": 1043
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.0794763565063477,
+      "learning_rate": 0.00019401147576878207,
+      "loss": 0.8717,
+      "step": 1044
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.314604640007019,
+      "learning_rate": 0.0001939025864386594,
+      "loss": 0.8732,
+      "step": 1045
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.1519172191619873,
+      "learning_rate": 0.00019379362194978355,
+      "loss": 0.9723,
+      "step": 1046
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.5442712306976318,
+      "learning_rate": 0.00019368458242101274,
+      "loss": 1.3932,
+      "step": 1047
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.2915006875991821,
+      "learning_rate": 0.000193575467971287,
+      "loss": 0.9708,
+      "step": 1048
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.2446788549423218,
+      "learning_rate": 0.00019346627871962805,
+      "loss": 0.969,
+      "step": 1049
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.3067604303359985,
+      "learning_rate": 0.00019335701478513922,
+      "loss": 1.1397,
+      "step": 1050
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.340515375137329,
+      "learning_rate": 0.00019324767628700533,
+      "loss": 1.2468,
+      "step": 1051
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.191666841506958,
+      "learning_rate": 0.0001931382633444925,
+      "loss": 0.9814,
+      "step": 1052
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.2348589897155762,
+      "learning_rate": 0.00019302877607694808,
+      "loss": 0.8009,
+      "step": 1053
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.4749141931533813,
+      "learning_rate": 0.00019291921460380047,
+      "loss": 0.9804,
+      "step": 1054
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.1957335472106934,
+      "learning_rate": 0.00019280957904455908,
+      "loss": 0.7987,
+      "step": 1055
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.232792854309082,
+      "learning_rate": 0.00019269986951881397,
+      "loss": 0.7048,
+      "step": 1056
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.184746503829956,
+      "learning_rate": 0.00019259008614623608,
+      "loss": 1.0233,
+      "step": 1057
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.4626327753067017,
+      "learning_rate": 0.00019248022904657682,
+      "loss": 1.2599,
+      "step": 1058
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.4971569776535034,
+      "learning_rate": 0.00019237029833966791,
+      "loss": 1.1236,
+      "step": 1059
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.2860374450683594,
+      "learning_rate": 0.0001922602941454216,
+      "loss": 0.9537,
+      "step": 1060
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.3197345733642578,
+      "learning_rate": 0.0001921502165838301,
+      "loss": 0.6415,
+      "step": 1061
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.3860671520233154,
+      "learning_rate": 0.0001920400657749657,
+      "loss": 1.11,
+      "step": 1062
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.2164676189422607,
+      "learning_rate": 0.0001919298418389806,
+      "loss": 0.9324,
+      "step": 1063
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.0971100330352783,
+      "learning_rate": 0.00019181954489610675,
+      "loss": 0.7691,
+      "step": 1064
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.2012615203857422,
+      "learning_rate": 0.00019170917506665586,
+      "loss": 1.0179,
+      "step": 1065
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.4872874021530151,
+      "learning_rate": 0.00019159873247101894,
+      "loss": 1.2214,
+      "step": 1066
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.1903997659683228,
+      "learning_rate": 0.00019148821722966654,
+      "loss": 0.7667,
+      "step": 1067
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.1436363458633423,
+      "learning_rate": 0.0001913776294631483,
+      "loss": 0.9839,
+      "step": 1068
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.2749754190444946,
+      "learning_rate": 0.0001912669692920931,
+      "loss": 0.7828,
+      "step": 1069
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.1588584184646606,
+      "learning_rate": 0.00019115623683720872,
+      "loss": 0.8214,
+      "step": 1070
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.3065457344055176,
+      "learning_rate": 0.0001910454322192819,
+      "loss": 0.7492,
+      "step": 1071
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.2145508527755737,
+      "learning_rate": 0.00019093455555917792,
+      "loss": 1.2346,
+      "step": 1072
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.2115508317947388,
+      "learning_rate": 0.00019082360697784086,
+      "loss": 1.1687,
+      "step": 1073
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.1368943452835083,
+      "learning_rate": 0.00019071258659629297,
+      "loss": 0.9316,
+      "step": 1074
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.0516806840896606,
+      "learning_rate": 0.00019060149453563509,
+      "loss": 0.6583,
+      "step": 1075
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.2155439853668213,
+      "learning_rate": 0.0001904903309170461,
+      "loss": 1.109,
+      "step": 1076
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.0046262741088867,
+      "learning_rate": 0.000190379095861783,
+      "loss": 0.5603,
+      "step": 1077
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.075215458869934,
+      "learning_rate": 0.00019026778949118061,
+      "loss": 0.7349,
+      "step": 1078
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.7739472389221191,
+      "learning_rate": 0.00019015641192665172,
+      "loss": 1.2746,
+      "step": 1079
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.1507693529129028,
+      "learning_rate": 0.00019004496328968662,
+      "loss": 0.9482,
+      "step": 1080
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.2549113035202026,
+      "learning_rate": 0.00018993344370185316,
+      "loss": 1.0353,
+      "step": 1081
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.3268182277679443,
+      "learning_rate": 0.00018982185328479662,
+      "loss": 1.2067,
+      "step": 1082
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.0886125564575195,
+      "learning_rate": 0.00018971019216023954,
+      "loss": 0.6659,
+      "step": 1083
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.0336356163024902,
+      "learning_rate": 0.00018959846044998152,
+      "loss": 0.7435,
+      "step": 1084
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.3987981081008911,
+      "learning_rate": 0.00018948665827589928,
+      "loss": 1.3236,
+      "step": 1085
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.208726167678833,
+      "learning_rate": 0.00018937478575994628,
+      "loss": 0.8633,
+      "step": 1086
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.9976479411125183,
+      "learning_rate": 0.0001892628430241527,
+      "loss": 0.71,
+      "step": 1087
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.2253488302230835,
+      "learning_rate": 0.00018915083019062552,
+      "loss": 0.8444,
+      "step": 1088
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.0337365865707397,
+      "learning_rate": 0.00018903874738154786,
+      "loss": 0.8667,
+      "step": 1089
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.9623100757598877,
+      "learning_rate": 0.00018892659471917946,
+      "loss": 0.631,
+      "step": 1090
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.048235297203064,
+      "learning_rate": 0.00018881437232585614,
+      "loss": 0.8731,
+      "step": 1091
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.1531683206558228,
+      "learning_rate": 0.00018870208032398968,
+      "loss": 0.965,
+      "step": 1092
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.4350179433822632,
+      "learning_rate": 0.00018858971883606795,
+      "loss": 1.3628,
+      "step": 1093
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.2467992305755615,
+      "learning_rate": 0.00018847728798465462,
+      "loss": 1.0464,
+      "step": 1094
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.1029807329177856,
+      "learning_rate": 0.00018836478789238887,
+      "loss": 0.5425,
+      "step": 1095
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.1991395950317383,
+      "learning_rate": 0.0001882522186819855,
+      "loss": 0.7859,
+      "step": 1096
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.2336832284927368,
+      "learning_rate": 0.0001881395804762347,
+      "loss": 0.9693,
+      "step": 1097
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.3640801906585693,
+      "learning_rate": 0.00018802687339800194,
+      "loss": 1.1183,
+      "step": 1098
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.9197866320610046,
+      "learning_rate": 0.0001879140975702278,
+      "loss": 0.5181,
+      "step": 1099
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.2878599166870117,
+      "learning_rate": 0.00018780125311592782,
+      "loss": 1.1359,
+      "step": 1100
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.1678839921951294,
+      "learning_rate": 0.00018768834015819242,
+      "loss": 0.8813,
+      "step": 1101
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.182902216911316,
+      "learning_rate": 0.00018757535882018677,
+      "loss": 0.9381,
+      "step": 1102
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.4679259061813354,
+      "learning_rate": 0.00018746230922515056,
+      "loss": 1.3104,
+      "step": 1103
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.3574234247207642,
+      "learning_rate": 0.00018734919149639804,
+      "loss": 1.3659,
+      "step": 1104
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.0465551614761353,
+      "learning_rate": 0.00018723600575731766,
+      "loss": 0.7132,
+      "step": 1105
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.2153966426849365,
+      "learning_rate": 0.0001871227521313722,
+      "loss": 1.1371,
+      "step": 1106
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.1352930068969727,
+      "learning_rate": 0.00018700943074209832,
+      "loss": 0.8152,
+      "step": 1107
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.13296639919281,
+      "learning_rate": 0.0001868960417131067,
+      "loss": 0.7719,
+      "step": 1108
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.8700066804885864,
+      "learning_rate": 0.00018678258516808183,
+      "loss": 0.4736,
+      "step": 1109
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.1155449151992798,
+      "learning_rate": 0.00018666906123078177,
+      "loss": 0.8818,
+      "step": 1110
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.0150140523910522,
+      "learning_rate": 0.00018655547002503807,
+      "loss": 0.6064,
+      "step": 1111
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.755456805229187,
+      "learning_rate": 0.00018644181167475578,
+      "loss": 1.3616,
+      "step": 1112
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.0811755657196045,
+      "learning_rate": 0.0001863280863039131,
+      "loss": 0.8551,
+      "step": 1113
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.2699004411697388,
+      "learning_rate": 0.00018621429403656135,
+      "loss": 0.9028,
+      "step": 1114
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.1369472742080688,
+      "learning_rate": 0.0001861004349968248,
+      "loss": 0.8264,
+      "step": 1115
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.8735272884368896,
+      "learning_rate": 0.00018598650930890063,
+      "loss": 0.4994,
+      "step": 1116
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.5912067890167236,
+      "learning_rate": 0.00018587251709705857,
+      "loss": 1.245,
+      "step": 1117
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.310185194015503,
+      "learning_rate": 0.00018575845848564117,
+      "loss": 1.0203,
+      "step": 1118
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.2570037841796875,
+      "learning_rate": 0.00018564433359906311,
+      "loss": 0.8932,
+      "step": 1119
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.3394275903701782,
+      "learning_rate": 0.00018553014256181158,
+      "loss": 1.4023,
+      "step": 1120
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.382513403892517,
+      "learning_rate": 0.00018541588549844584,
+      "loss": 0.6844,
+      "step": 1121
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.244809865951538,
+      "learning_rate": 0.00018530156253359714,
+      "loss": 0.9417,
+      "step": 1122
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.0259147882461548,
+      "learning_rate": 0.00018518717379196873,
+      "loss": 0.8997,
+      "step": 1123
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.8853318691253662,
+      "learning_rate": 0.00018507271939833546,
+      "loss": 0.4958,
+      "step": 1124
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.3527164459228516,
+      "learning_rate": 0.00018495819947754398,
+      "loss": 1.1756,
+      "step": 1125
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.355082392692566,
+      "learning_rate": 0.00018484361415451218,
+      "loss": 1.2114,
+      "step": 1126
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.2699202299118042,
+      "learning_rate": 0.00018472896355422946,
+      "loss": 0.6697,
+      "step": 1127
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.4729478359222412,
+      "learning_rate": 0.00018461424780175645,
+      "loss": 1.4576,
+      "step": 1128
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.1743948459625244,
+      "learning_rate": 0.00018449946702222465,
+      "loss": 0.8025,
+      "step": 1129
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.4571646451950073,
+      "learning_rate": 0.00018438462134083675,
+      "loss": 1.0594,
+      "step": 1130
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.3907976150512695,
+      "learning_rate": 0.00018426971088286604,
+      "loss": 1.0993,
+      "step": 1131
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.4448930025100708,
+      "learning_rate": 0.00018415473577365653,
+      "loss": 1.171,
+      "step": 1132
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.0253756046295166,
+      "learning_rate": 0.00018403969613862278,
+      "loss": 0.592,
+      "step": 1133
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.2286980152130127,
+      "learning_rate": 0.00018392459210324975,
+      "loss": 1.043,
+      "step": 1134
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.1266531944274902,
+      "learning_rate": 0.00018380942379309247,
+      "loss": 0.6981,
+      "step": 1135
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.3554151058197021,
+      "learning_rate": 0.00018369419133377636,
+      "loss": 0.6397,
+      "step": 1136
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.3931509256362915,
+      "learning_rate": 0.0001835788948509966,
+      "loss": 0.9987,
+      "step": 1137
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.070232629776001,
+      "learning_rate": 0.00018346353447051827,
+      "loss": 0.8607,
+      "step": 1138
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.1246999502182007,
+      "learning_rate": 0.00018334811031817626,
+      "loss": 0.8323,
+      "step": 1139
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.3503284454345703,
+      "learning_rate": 0.00018323262251987476,
+      "loss": 0.9194,
+      "step": 1140
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.0961241722106934,
+      "learning_rate": 0.00018311707120158768,
+      "loss": 0.7868,
+      "step": 1141
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.3863087892532349,
+      "learning_rate": 0.000183001456489358,
+      "loss": 1.2253,
+      "step": 1142
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.0947837829589844,
+      "learning_rate": 0.000182885778509298,
+      "loss": 0.932,
+      "step": 1143
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.5243417024612427,
+      "learning_rate": 0.00018277003738758886,
+      "loss": 1.1453,
+      "step": 1144
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.2207045555114746,
+      "learning_rate": 0.00018265423325048067,
+      "loss": 1.0409,
+      "step": 1145
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.0962190628051758,
+      "learning_rate": 0.00018253836622429229,
+      "loss": 0.981,
+      "step": 1146
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.893530547618866,
+      "learning_rate": 0.00018242243643541115,
+      "loss": 0.6334,
+      "step": 1147
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.153269648551941,
+      "learning_rate": 0.00018230644401029313,
+      "loss": 0.9353,
+      "step": 1148
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.3481030464172363,
+      "learning_rate": 0.00018219038907546254,
+      "loss": 1.11,
+      "step": 1149
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.2734599113464355,
+      "learning_rate": 0.00018207427175751165,
+      "loss": 1.037,
+      "step": 1150
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.8931764364242554,
+      "learning_rate": 0.000181958092183101,
+      "loss": 0.5017,
+      "step": 1151
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.3497036695480347,
+      "learning_rate": 0.0001818418504789589,
+      "loss": 0.7842,
+      "step": 1152
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.232225775718689,
+      "learning_rate": 0.00018172554677188155,
+      "loss": 0.9868,
+      "step": 1153
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.5071431398391724,
+      "learning_rate": 0.00018160918118873264,
+      "loss": 1.1483,
+      "step": 1154
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.155139684677124,
+      "learning_rate": 0.00018149275385644353,
+      "loss": 0.9804,
+      "step": 1155
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.130774736404419,
+      "learning_rate": 0.00018137626490201273,
+      "loss": 0.9467,
+      "step": 1156
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.2794990539550781,
+      "learning_rate": 0.00018125971445250613,
+      "loss": 1.1415,
+      "step": 1157
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.308632731437683,
+      "learning_rate": 0.00018114310263505666,
+      "loss": 0.9147,
+      "step": 1158
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.013698935508728,
+      "learning_rate": 0.0001810264295768641,
+      "loss": 0.5198,
+      "step": 1159
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.1336946487426758,
+      "learning_rate": 0.0001809096954051952,
+      "loss": 0.7719,
+      "step": 1160
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.2344475984573364,
+      "learning_rate": 0.00018079290024738316,
+      "loss": 0.9903,
+      "step": 1161
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.2032207250595093,
+      "learning_rate": 0.0001806760442308279,
+      "loss": 0.8217,
+      "step": 1162
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.0712535381317139,
+      "learning_rate": 0.00018055912748299565,
+      "loss": 0.9274,
+      "step": 1163
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.2632014751434326,
+      "learning_rate": 0.00018044215013141886,
+      "loss": 0.842,
+      "step": 1164
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.4885759353637695,
+      "learning_rate": 0.00018032511230369605,
+      "loss": 0.9768,
+      "step": 1165
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.3400936126708984,
+      "learning_rate": 0.00018020801412749185,
+      "loss": 0.7842,
+      "step": 1166
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.155679702758789,
+      "learning_rate": 0.0001800908557305366,
+      "loss": 0.4815,
+      "step": 1167
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.4821432828903198,
+      "learning_rate": 0.00017997363724062633,
+      "loss": 0.9798,
+      "step": 1168
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.346875786781311,
+      "learning_rate": 0.0001798563587856227,
+      "loss": 0.9786,
+      "step": 1169
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.5733568668365479,
+      "learning_rate": 0.0001797390204934527,
+      "loss": 1.1719,
+      "step": 1170
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.9875895380973816,
+      "learning_rate": 0.0001796216224921086,
+      "loss": 0.4865,
+      "step": 1171
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.3760789632797241,
+      "learning_rate": 0.00017950416490964785,
+      "loss": 0.9958,
+      "step": 1172
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.535853624343872,
+      "learning_rate": 0.00017938664787419285,
+      "loss": 1.6119,
+      "step": 1173
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.409841537475586,
+      "learning_rate": 0.00017926907151393087,
+      "loss": 1.1741,
+      "step": 1174
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.3689299821853638,
+      "learning_rate": 0.00017915143595711387,
+      "loss": 0.7744,
+      "step": 1175
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.3054869174957275,
+      "learning_rate": 0.00017903374133205834,
+      "loss": 0.901,
+      "step": 1176
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.3351161479949951,
+      "learning_rate": 0.00017891598776714537,
+      "loss": 1.3365,
+      "step": 1177
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.4273672103881836,
+      "learning_rate": 0.0001787981753908202,
+      "loss": 1.0755,
+      "step": 1178
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.6055691242218018,
+      "learning_rate": 0.00017868030433159217,
+      "loss": 1.131,
+      "step": 1179
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.9369842410087585,
+      "learning_rate": 0.00017856237471803473,
+      "loss": 0.6455,
+      "step": 1180
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.1839696168899536,
+      "learning_rate": 0.00017844438667878525,
+      "loss": 1.3519,
+      "step": 1181
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.9531559348106384,
+      "learning_rate": 0.00017832634034254476,
+      "loss": 0.6091,
+      "step": 1182
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.396750569343567,
+      "learning_rate": 0.00017820823583807778,
+      "loss": 0.6756,
+      "step": 1183
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.2895655632019043,
+      "learning_rate": 0.00017809007329421244,
+      "loss": 1.3657,
+      "step": 1184
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.5052480697631836,
+      "learning_rate": 0.00017797185283984017,
+      "loss": 1.3727,
+      "step": 1185
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.4311729669570923,
+      "learning_rate": 0.00017785357460391543,
+      "loss": 1.0337,
+      "step": 1186
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.0396034717559814,
+      "learning_rate": 0.00017773523871545581,
+      "loss": 0.6779,
+      "step": 1187
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.4612584114074707,
+      "learning_rate": 0.0001776168453035419,
+      "loss": 1.2791,
+      "step": 1188
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.1547785997390747,
+      "learning_rate": 0.00017749839449731673,
+      "loss": 0.8077,
+      "step": 1189
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.1935621500015259,
+      "learning_rate": 0.00017737988642598624,
+      "loss": 0.6824,
+      "step": 1190
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.043798565864563,
+      "learning_rate": 0.00017726132121881864,
+      "loss": 0.7545,
+      "step": 1191
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.3549226522445679,
+      "learning_rate": 0.00017714269900514457,
+      "loss": 1.1167,
+      "step": 1192
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.1858012676239014,
+      "learning_rate": 0.00017702401991435683,
+      "loss": 0.9483,
+      "step": 1193
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.0237493515014648,
+      "learning_rate": 0.0001769052840759102,
+      "loss": 0.7064,
+      "step": 1194
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.4379832744598389,
+      "learning_rate": 0.00017678649161932146,
+      "loss": 1.1197,
+      "step": 1195
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.8990068435668945,
+      "learning_rate": 0.00017666764267416904,
+      "loss": 0.4801,
+      "step": 1196
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.2909594774246216,
+      "learning_rate": 0.0001765487373700931,
+      "loss": 0.9925,
+      "step": 1197
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.4378669261932373,
+      "learning_rate": 0.0001764297758367952,
+      "loss": 1.1316,
+      "step": 1198
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.493698239326477,
+      "learning_rate": 0.00017631075820403823,
+      "loss": 0.9835,
+      "step": 1199
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.2861640453338623,
+      "learning_rate": 0.00017619168460164636,
+      "loss": 1.2319,
+      "step": 1200
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.1576422452926636,
+      "learning_rate": 0.00017607255515950468,
+      "loss": 0.75,
+      "step": 1201
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.2239806652069092,
+      "learning_rate": 0.00017595337000755936,
+      "loss": 0.9199,
+      "step": 1202
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.2435945272445679,
+      "learning_rate": 0.0001758341292758171,
+      "loss": 1.1669,
+      "step": 1203
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.130832552909851,
+      "learning_rate": 0.00017571483309434545,
+      "loss": 0.5273,
+      "step": 1204
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.2331628799438477,
+      "learning_rate": 0.00017559548159327238,
+      "loss": 0.8775,
+      "step": 1205
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.422377586364746,
+      "learning_rate": 0.0001754760749027861,
+      "loss": 0.9985,
+      "step": 1206
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.3476437330245972,
+      "learning_rate": 0.0001753566131531352,
+      "loss": 0.9553,
+      "step": 1207
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.1213027238845825,
+      "learning_rate": 0.00017523709647462816,
+      "loss": 0.9864,
+      "step": 1208
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.28190016746521,
+      "learning_rate": 0.00017511752499763347,
+      "loss": 0.9803,
+      "step": 1209
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.8669500350952148,
+      "learning_rate": 0.00017499789885257937,
+      "loss": 1.5848,
+      "step": 1210
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.5373106002807617,
+      "learning_rate": 0.0001748782181699537,
+      "loss": 1.3516,
+      "step": 1211
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.0895917415618896,
+      "learning_rate": 0.00017475848308030385,
+      "loss": 0.8032,
+      "step": 1212
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.0846320390701294,
+      "learning_rate": 0.00017463869371423657,
+      "loss": 0.7299,
+      "step": 1213
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.2971018552780151,
+      "learning_rate": 0.0001745188502024177,
+      "loss": 0.8337,
+      "step": 1214
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.5091488361358643,
+      "learning_rate": 0.00017439895267557225,
+      "loss": 1.2648,
+      "step": 1215
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.170003056526184,
+      "learning_rate": 0.00017427900126448415,
+      "loss": 0.8008,
+      "step": 1216
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.247881293296814,
+      "learning_rate": 0.000174158996099996,
+      "loss": 0.8748,
+      "step": 1217
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.2198271751403809,
+      "learning_rate": 0.00017403893731300917,
+      "loss": 0.8583,
+      "step": 1218
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.173490285873413,
+      "learning_rate": 0.0001739188250344834,
+      "loss": 0.8296,
+      "step": 1219
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.0720738172531128,
+      "learning_rate": 0.00017379865939543684,
+      "loss": 0.56,
+      "step": 1220
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.481160044670105,
+      "learning_rate": 0.00017367844052694592,
+      "loss": 0.8178,
+      "step": 1221
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.0757296085357666,
+      "learning_rate": 0.00017355816856014496,
+      "loss": 0.7001,
+      "step": 1222
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.9257085919380188,
+      "learning_rate": 0.00017343784362622636,
+      "loss": 0.6719,
+      "step": 1223
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.2127162218093872,
+      "learning_rate": 0.00017331746585644018,
+      "loss": 0.8813,
+      "step": 1224
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.4545499086380005,
+      "learning_rate": 0.0001731970353820942,
+      "loss": 0.7659,
+      "step": 1225
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.1807411909103394,
+      "learning_rate": 0.00017307655233455362,
+      "loss": 0.632,
+      "step": 1226
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.146098256111145,
+      "learning_rate": 0.00017295601684524104,
+      "loss": 0.9484,
+      "step": 1227
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.102598786354065,
+      "learning_rate": 0.00017283542904563624,
+      "loss": 0.6946,
+      "step": 1228
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.276293396949768,
+      "learning_rate": 0.00017271478906727604,
+      "loss": 1.104,
+      "step": 1229
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.6087700128555298,
+      "learning_rate": 0.00017259409704175425,
+      "loss": 0.9789,
+      "step": 1230
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.503319263458252,
+      "learning_rate": 0.00017247335310072135,
+      "loss": 1.3899,
+      "step": 1231
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.1361923217773438,
+      "learning_rate": 0.00017235255737588453,
+      "loss": 0.663,
+      "step": 1232
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.0255531072616577,
+      "learning_rate": 0.0001722317099990074,
+      "loss": 0.8311,
+      "step": 1233
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.3245635032653809,
+      "learning_rate": 0.00017211081110191,
+      "loss": 0.8968,
+      "step": 1234
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.347150444984436,
+      "learning_rate": 0.0001719898608164685,
+      "loss": 0.99,
+      "step": 1235
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.0764960050582886,
+      "learning_rate": 0.00017186885927461515,
+      "loss": 0.8349,
+      "step": 1236
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.0420310497283936,
+      "learning_rate": 0.00017174780660833805,
+      "loss": 0.8254,
+      "step": 1237
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.7763484716415405,
+      "learning_rate": 0.00017162670294968116,
+      "loss": 1.8731,
+      "step": 1238
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.8756821751594543,
+      "learning_rate": 0.00017150554843074406,
+      "loss": 0.4927,
+      "step": 1239
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.204406499862671,
+      "learning_rate": 0.00017138434318368168,
+      "loss": 0.955,
+      "step": 1240
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.2387770414352417,
+      "learning_rate": 0.00017126308734070444,
+      "loss": 1.0619,
+      "step": 1241
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.0382620096206665,
+      "learning_rate": 0.00017114178103407783,
+      "loss": 0.7705,
+      "step": 1242
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.9521413445472717,
+      "learning_rate": 0.00017102042439612246,
+      "loss": 0.5972,
+      "step": 1243
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.4533095359802246,
+      "learning_rate": 0.00017089901755921387,
+      "loss": 1.268,
+      "step": 1244
+    },
+    {
+      "epoch": 1.2,
+      "eval_loss": 1.4754925966262817,
+      "eval_runtime": 28.2048,
+      "eval_samples_per_second": 3.014,
+      "eval_steps_per_second": 1.525,
+      "step": 1244
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.1730684041976929,
+      "learning_rate": 0.00017077756065578224,
+      "loss": 0.838,
+      "step": 1245
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.598128080368042,
+      "learning_rate": 0.00017065605381831242,
+      "loss": 0.9254,
+      "step": 1246
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.2155981063842773,
+      "learning_rate": 0.00017053449717934377,
+      "loss": 0.9429,
+      "step": 1247
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.1935361623764038,
+      "learning_rate": 0.00017041289087146999,
+      "loss": 0.8267,
+      "step": 1248
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.027895450592041,
+      "learning_rate": 0.0001702912350273388,
+      "loss": 0.8369,
+      "step": 1249
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.0867793560028076,
+      "learning_rate": 0.00017016952977965216,
+      "loss": 0.701,
+      "step": 1250
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.261580467224121,
+      "learning_rate": 0.00017004777526116578,
+      "loss": 1.3502,
+      "step": 1251
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.3365131616592407,
+      "learning_rate": 0.0001699259716046891,
+      "loss": 1.1648,
+      "step": 1252
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.0136491060256958,
+      "learning_rate": 0.00016980411894308535,
+      "loss": 0.7513,
+      "step": 1253
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.0561655759811401,
+      "learning_rate": 0.00016968221740927092,
+      "loss": 0.6481,
+      "step": 1254
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.2038730382919312,
+      "learning_rate": 0.0001695602671362158,
+      "loss": 1.1284,
+      "step": 1255
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.089046835899353,
+      "learning_rate": 0.00016943826825694297,
+      "loss": 0.6818,
+      "step": 1256
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.2403234243392944,
+      "learning_rate": 0.00016931622090452845,
+      "loss": 1.0335,
+      "step": 1257
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.31325101852417,
+      "learning_rate": 0.00016919412521210117,
+      "loss": 0.8025,
+      "step": 1258
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.9705163836479187,
+      "learning_rate": 0.0001690719813128428,
+      "loss": 0.5008,
+      "step": 1259
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.4634290933609009,
+      "learning_rate": 0.00016894978933998758,
+      "loss": 1.2704,
+      "step": 1260
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.4494082927703857,
+      "learning_rate": 0.00016882754942682216,
+      "loss": 0.9463,
+      "step": 1261
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.2695739269256592,
+      "learning_rate": 0.0001687052617066855,
+      "loss": 0.8996,
+      "step": 1262
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.9623706936836243,
+      "learning_rate": 0.0001685829263129687,
+      "loss": 0.8756,
+      "step": 1263
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.99055016040802,
+      "learning_rate": 0.00016846054337911487,
+      "loss": 0.6137,
+      "step": 1264
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.3150849342346191,
+      "learning_rate": 0.00016833811303861907,
+      "loss": 0.7671,
+      "step": 1265
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.3409266471862793,
+      "learning_rate": 0.00016821563542502782,
+      "loss": 0.9181,
+      "step": 1266
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.1556544303894043,
+      "learning_rate": 0.0001680931106719395,
+      "loss": 0.8652,
+      "step": 1267
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.262967824935913,
+      "learning_rate": 0.00016797053891300377,
+      "loss": 0.9559,
+      "step": 1268
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.3712326288223267,
+      "learning_rate": 0.0001678479202819215,
+      "loss": 1.1403,
+      "step": 1269
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.3038856983184814,
+      "learning_rate": 0.00016772525491244472,
+      "loss": 1.1641,
+      "step": 1270
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.2619274854660034,
+      "learning_rate": 0.00016760254293837664,
+      "loss": 0.8464,
+      "step": 1271
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.2340606451034546,
+      "learning_rate": 0.00016747978449357104,
+      "loss": 1.0419,
+      "step": 1272
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.9198154211044312,
+      "learning_rate": 0.00016735697971193247,
+      "loss": 0.7761,
+      "step": 1273
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.0390851497650146,
+      "learning_rate": 0.00016723412872741617,
+      "loss": 0.989,
+      "step": 1274
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.4333105087280273,
+      "learning_rate": 0.00016711123167402752,
+      "loss": 1.2331,
+      "step": 1275
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.4277148246765137,
+      "learning_rate": 0.0001669882886858224,
+      "loss": 0.7869,
+      "step": 1276
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.2195308208465576,
+      "learning_rate": 0.00016686529989690658,
+      "loss": 1.0324,
+      "step": 1277
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.05024254322052,
+      "learning_rate": 0.00016674226544143595,
+      "loss": 0.7656,
+      "step": 1278
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.4788092374801636,
+      "learning_rate": 0.0001666191854536162,
+      "loss": 1.0497,
+      "step": 1279
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.185929536819458,
+      "learning_rate": 0.0001664960600677026,
+      "loss": 0.8825,
+      "step": 1280
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.2931797504425049,
+      "learning_rate": 0.00016637288941799997,
+      "loss": 1.2804,
+      "step": 1281
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.1615878343582153,
+      "learning_rate": 0.00016624967363886253,
+      "loss": 0.7602,
+      "step": 1282
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.217430830001831,
+      "learning_rate": 0.00016612641286469377,
+      "loss": 0.9886,
+      "step": 1283
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.4690346717834473,
+      "learning_rate": 0.00016600310722994613,
+      "loss": 1.1719,
+      "step": 1284
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.2001279592514038,
+      "learning_rate": 0.00016587975686912113,
+      "loss": 0.877,
+      "step": 1285
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.923620879650116,
+      "learning_rate": 0.0001657563619167689,
+      "loss": 0.6299,
+      "step": 1286
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.2251518964767456,
+      "learning_rate": 0.0001656329225074884,
+      "loss": 0.7887,
+      "step": 1287
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.6504628658294678,
+      "learning_rate": 0.00016550943877592704,
+      "loss": 0.9195,
+      "step": 1288
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.341532826423645,
+      "learning_rate": 0.00016538591085678043,
+      "loss": 0.8872,
+      "step": 1289
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.5258677005767822,
+      "learning_rate": 0.00016526233888479255,
+      "loss": 1.1348,
+      "step": 1290
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.1335400342941284,
+      "learning_rate": 0.00016513872299475535,
+      "loss": 0.7941,
+      "step": 1291
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.2744227647781372,
+      "learning_rate": 0.00016501506332150873,
+      "loss": 0.9789,
+      "step": 1292
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.2145026922225952,
+      "learning_rate": 0.00016489135999994025,
+      "loss": 0.9446,
+      "step": 1293
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.9690343141555786,
+      "learning_rate": 0.00016476761316498527,
+      "loss": 0.8523,
+      "step": 1294
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.2858216762542725,
+      "learning_rate": 0.00016464382295162643,
+      "loss": 1.2376,
+      "step": 1295
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.5635852813720703,
+      "learning_rate": 0.00016451998949489378,
+      "loss": 1.3802,
+      "step": 1296
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.374159812927246,
+      "learning_rate": 0.00016439611292986452,
+      "loss": 1.1782,
+      "step": 1297
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.4581077098846436,
+      "learning_rate": 0.0001642721933916628,
+      "loss": 1.1469,
+      "step": 1298
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.4149705171585083,
+      "learning_rate": 0.0001641482310154598,
+      "loss": 1.1342,
+      "step": 1299
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.981275737285614,
+      "learning_rate": 0.0001640242259364733,
+      "loss": 0.6698,
+      "step": 1300
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.568472146987915,
+      "learning_rate": 0.0001639001782899677,
+      "loss": 0.8356,
+      "step": 1301
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.5767405033111572,
+      "learning_rate": 0.0001637760882112539,
+      "loss": 1.1833,
+      "step": 1302
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.1946420669555664,
+      "learning_rate": 0.0001636519558356889,
+      "loss": 0.938,
+      "step": 1303
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.0555033683776855,
+      "learning_rate": 0.00016352778129867607,
+      "loss": 0.7367,
+      "step": 1304
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.7251577377319336,
+      "learning_rate": 0.00016340356473566454,
+      "loss": 1.2856,
+      "step": 1305
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.1670684814453125,
+      "learning_rate": 0.00016327930628214952,
+      "loss": 1.1319,
+      "step": 1306
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.2817931175231934,
+      "learning_rate": 0.0001631550060736717,
+      "loss": 1.0645,
+      "step": 1307
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.2379206418991089,
+      "learning_rate": 0.00016303066424581746,
+      "loss": 0.6547,
+      "step": 1308
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.0804026126861572,
+      "learning_rate": 0.0001629062809342185,
+      "loss": 0.7313,
+      "step": 1309
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.3285449743270874,
+      "learning_rate": 0.00016278185627455183,
+      "loss": 0.9509,
+      "step": 1310
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.3058103322982788,
+      "learning_rate": 0.00016265739040253948,
+      "loss": 0.974,
+      "step": 1311
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.261012077331543,
+      "learning_rate": 0.00016253288345394854,
+      "loss": 0.9932,
+      "step": 1312
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.5090525150299072,
+      "learning_rate": 0.0001624083355645908,
+      "loss": 1.2195,
+      "step": 1313
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.3207299709320068,
+      "learning_rate": 0.00016228374687032277,
+      "loss": 0.9553,
+      "step": 1314
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.3063914775848389,
+      "learning_rate": 0.00016215911750704545,
+      "loss": 0.9287,
+      "step": 1315
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.2555960416793823,
+      "learning_rate": 0.00016203444761070425,
+      "loss": 0.9512,
+      "step": 1316
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.2886016368865967,
+      "learning_rate": 0.0001619097373172887,
+      "loss": 1.2255,
+      "step": 1317
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.507990837097168,
+      "learning_rate": 0.00016178498676283252,
+      "loss": 1.2742,
+      "step": 1318
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.4528714418411255,
+      "learning_rate": 0.00016166019608341317,
+      "loss": 1.0882,
+      "step": 1319
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.2334988117218018,
+      "learning_rate": 0.00016153536541515208,
+      "loss": 1.0011,
+      "step": 1320
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.2907617092132568,
+      "learning_rate": 0.00016141049489421414,
+      "loss": 1.0559,
+      "step": 1321
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.0541855096817017,
+      "learning_rate": 0.00016128558465680777,
+      "loss": 0.996,
+      "step": 1322
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.9502456188201904,
+      "learning_rate": 0.0001611606348391848,
+      "loss": 0.6858,
+      "step": 1323
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.081721305847168,
+      "learning_rate": 0.00016103564557764006,
+      "loss": 0.9702,
+      "step": 1324
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.2104740142822266,
+      "learning_rate": 0.00016091061700851153,
+      "loss": 0.8464,
+      "step": 1325
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.061679720878601,
+      "learning_rate": 0.00016078554926818,
+      "loss": 0.7161,
+      "step": 1326
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.1888184547424316,
+      "learning_rate": 0.0001606604424930691,
+      "loss": 0.9908,
+      "step": 1327
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.3492275476455688,
+      "learning_rate": 0.00016053529681964485,
+      "loss": 1.1636,
+      "step": 1328
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.8432413339614868,
+      "learning_rate": 0.00016041011238441592,
+      "loss": 1.0478,
+      "step": 1329
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.2208960056304932,
+      "learning_rate": 0.00016028488932393303,
+      "loss": 0.9052,
+      "step": 1330
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.5409330129623413,
+      "learning_rate": 0.00016015962777478926,
+      "loss": 0.8948,
+      "step": 1331
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.4419323205947876,
+      "learning_rate": 0.0001600343278736195,
+      "loss": 1.148,
+      "step": 1332
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.1415199041366577,
+      "learning_rate": 0.00015990898975710058,
+      "loss": 1.129,
+      "step": 1333
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.625235676765442,
+      "learning_rate": 0.000159783613561951,
+      "loss": 1.2489,
+      "step": 1334
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.1700189113616943,
+      "learning_rate": 0.00015965819942493064,
+      "loss": 0.7232,
+      "step": 1335
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.1410647630691528,
+      "learning_rate": 0.00015953274748284102,
+      "loss": 0.8594,
+      "step": 1336
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.0741961002349854,
+      "learning_rate": 0.00015940725787252477,
+      "loss": 0.8474,
+      "step": 1337
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.0182451009750366,
+      "learning_rate": 0.00015928173073086557,
+      "loss": 0.7573,
+      "step": 1338
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.0768117904663086,
+      "learning_rate": 0.00015915616619478813,
+      "loss": 1.2029,
+      "step": 1339
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.025019884109497,
+      "learning_rate": 0.0001590305644012578,
+      "loss": 0.9151,
+      "step": 1340
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.3672230243682861,
+      "learning_rate": 0.0001589049254872808,
+      "loss": 1.3459,
+      "step": 1341
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.2047271728515625,
+      "learning_rate": 0.00015877924958990363,
+      "loss": 1.0726,
+      "step": 1342
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 4.216643810272217,
+      "learning_rate": 0.00015865353684621332,
+      "loss": 0.902,
+      "step": 1343
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.0406891107559204,
+      "learning_rate": 0.00015852778739333683,
+      "loss": 0.7834,
+      "step": 1344
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.3921639919281006,
+      "learning_rate": 0.00015840200136844144,
+      "loss": 1.0515,
+      "step": 1345
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.4456580877304077,
+      "learning_rate": 0.00015827617890873418,
+      "loss": 0.9935,
+      "step": 1346
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.5223323106765747,
+      "learning_rate": 0.00015815032015146181,
+      "loss": 1.2268,
+      "step": 1347
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.01329505443573,
+      "learning_rate": 0.0001580244252339108,
+      "loss": 0.6671,
+      "step": 1348
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.2426472902297974,
+      "learning_rate": 0.00015789849429340684,
+      "loss": 0.8958,
+      "step": 1349
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.4329215288162231,
+      "learning_rate": 0.00015777252746731517,
+      "loss": 0.8278,
+      "step": 1350
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.1606448888778687,
+      "learning_rate": 0.00015764652489304003,
+      "loss": 0.9097,
+      "step": 1351
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.0638396739959717,
+      "learning_rate": 0.00015752048670802468,
+      "loss": 0.8128,
+      "step": 1352
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.1829980611801147,
+      "learning_rate": 0.00015739441304975124,
+      "loss": 0.7878,
+      "step": 1353
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.1757057905197144,
+      "learning_rate": 0.0001572683040557405,
+      "loss": 0.6479,
+      "step": 1354
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.0985661745071411,
+      "learning_rate": 0.00015714215986355182,
+      "loss": 0.8554,
+      "step": 1355
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.2881649732589722,
+      "learning_rate": 0.00015701598061078294,
+      "loss": 1.0739,
+      "step": 1356
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.6648950576782227,
+      "learning_rate": 0.0001568897664350698,
+      "loss": 1.4414,
+      "step": 1357
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.171886682510376,
+      "learning_rate": 0.00015676351747408653,
+      "loss": 0.6933,
+      "step": 1358
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.1470980644226074,
+      "learning_rate": 0.00015663723386554512,
+      "loss": 0.8455,
+      "step": 1359
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.2526499032974243,
+      "learning_rate": 0.00015651091574719544,
+      "loss": 1.1787,
+      "step": 1360
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.0531766414642334,
+      "learning_rate": 0.00015638456325682486,
+      "loss": 0.615,
+      "step": 1361
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.813473641872406,
+      "learning_rate": 0.00015625817653225842,
+      "loss": 0.4417,
+      "step": 1362
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.1922723054885864,
+      "learning_rate": 0.00015613175571135837,
+      "loss": 1.0914,
+      "step": 1363
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.3487348556518555,
+      "learning_rate": 0.00015600530093202417,
+      "loss": 1.0338,
+      "step": 1364
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.015734076499939,
+      "learning_rate": 0.0001558788123321924,
+      "loss": 0.6342,
+      "step": 1365
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.4268220663070679,
+      "learning_rate": 0.00015575229004983647,
+      "loss": 1.3462,
+      "step": 1366
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.2125145196914673,
+      "learning_rate": 0.0001556257342229665,
+      "loss": 1.0775,
+      "step": 1367
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.6474987268447876,
+      "learning_rate": 0.00015549914498962927,
+      "loss": 0.8819,
+      "step": 1368
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.220054030418396,
+      "learning_rate": 0.000155372522487908,
+      "loss": 1.0804,
+      "step": 1369
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.1821403503417969,
+      "learning_rate": 0.00015524586685592212,
+      "loss": 0.7528,
+      "step": 1370
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.4041728973388672,
+      "learning_rate": 0.00015511917823182728,
+      "loss": 1.0064,
+      "step": 1371
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.2141079902648926,
+      "learning_rate": 0.00015499245675381504,
+      "loss": 0.8372,
+      "step": 1372
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.4008363485336304,
+      "learning_rate": 0.00015486570256011286,
+      "loss": 0.8686,
+      "step": 1373
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.357132911682129,
+      "learning_rate": 0.00015473891578898395,
+      "loss": 1.2691,
+      "step": 1374
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.5801461935043335,
+      "learning_rate": 0.0001546120965787268,
+      "loss": 1.0234,
+      "step": 1375
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.1867705583572388,
+      "learning_rate": 0.00015448524506767565,
+      "loss": 0.9288,
+      "step": 1376
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.2222601175308228,
+      "learning_rate": 0.00015435836139419964,
+      "loss": 1.1417,
+      "step": 1377
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.1129976511001587,
+      "learning_rate": 0.00015423144569670324,
+      "loss": 0.994,
+      "step": 1378
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.2277920246124268,
+      "learning_rate": 0.00015410449811362566,
+      "loss": 0.6984,
+      "step": 1379
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.030815601348877,
+      "learning_rate": 0.00015397751878344106,
+      "loss": 0.7813,
+      "step": 1380
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.9685385227203369,
+      "learning_rate": 0.00015385050784465804,
+      "loss": 0.5461,
+      "step": 1381
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.3186579942703247,
+      "learning_rate": 0.00015372346543581985,
+      "loss": 0.7531,
+      "step": 1382
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.6090222597122192,
+      "learning_rate": 0.00015359639169550403,
+      "loss": 1.5783,
+      "step": 1383
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.7105401754379272,
+      "learning_rate": 0.0001534692867623222,
+      "loss": 1.302,
+      "step": 1384
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.6060900688171387,
+      "learning_rate": 0.00015334215077492016,
+      "loss": 1.0983,
+      "step": 1385
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.1998249292373657,
+      "learning_rate": 0.00015321498387197745,
+      "loss": 0.892,
+      "step": 1386
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.340060830116272,
+      "learning_rate": 0.0001530877861922073,
+      "loss": 0.8928,
+      "step": 1387
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.3283123970031738,
+      "learning_rate": 0.00015296055787435674,
+      "loss": 0.8075,
+      "step": 1388
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.1961750984191895,
+      "learning_rate": 0.00015283329905720595,
+      "loss": 0.6859,
+      "step": 1389
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.2345889806747437,
+      "learning_rate": 0.00015270600987956858,
+      "loss": 1.0289,
+      "step": 1390
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.4048421382904053,
+      "learning_rate": 0.0001525786904802913,
+      "loss": 1.1179,
+      "step": 1391
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.215480089187622,
+      "learning_rate": 0.00015245134099825376,
+      "loss": 0.9839,
+      "step": 1392
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.142815113067627,
+      "learning_rate": 0.00015232396157236839,
+      "loss": 0.9429,
+      "step": 1393
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.5882869958877563,
+      "learning_rate": 0.00015219655234158033,
+      "loss": 1.4284,
+      "step": 1394
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.3742319345474243,
+      "learning_rate": 0.00015206911344486728,
+      "loss": 1.5371,
+      "step": 1395
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.8513232469558716,
+      "learning_rate": 0.0001519416450212392,
+      "loss": 0.5392,
+      "step": 1396
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.068568468093872,
+      "learning_rate": 0.00015181414720973833,
+      "loss": 0.7804,
+      "step": 1397
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.9229900240898132,
+      "learning_rate": 0.00015168662014943885,
+      "loss": 0.6506,
+      "step": 1398
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.099913477897644,
+      "learning_rate": 0.000151559063979447,
+      "loss": 1.0733,
+      "step": 1399
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.2708427906036377,
+      "learning_rate": 0.00015143147883890065,
+      "loss": 0.89,
+      "step": 1400
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.1063934564590454,
+      "learning_rate": 0.00015130386486696942,
+      "loss": 0.8854,
+      "step": 1401
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.1168344020843506,
+      "learning_rate": 0.00015117622220285417,
+      "loss": 0.885,
+      "step": 1402
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.444875955581665,
+      "learning_rate": 0.00015104855098578725,
+      "loss": 0.9915,
+      "step": 1403
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.1813578605651855,
+      "learning_rate": 0.00015092085135503207,
+      "loss": 0.837,
+      "step": 1404
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.827250599861145,
+      "learning_rate": 0.00015079312344988294,
+      "loss": 0.6524,
+      "step": 1405
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.2115777730941772,
+      "learning_rate": 0.00015066536740966524,
+      "loss": 1.1392,
+      "step": 1406
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.4499708414077759,
+      "learning_rate": 0.00015053758337373483,
+      "loss": 1.3624,
+      "step": 1407
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.160974383354187,
+      "learning_rate": 0.00015040977148147823,
+      "loss": 0.8182,
+      "step": 1408
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.2181323766708374,
+      "learning_rate": 0.00015028193187231227,
+      "loss": 1.2563,
+      "step": 1409
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.2282092571258545,
+      "learning_rate": 0.00015015406468568405,
+      "loss": 0.7993,
+      "step": 1410
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.5285307168960571,
+      "learning_rate": 0.00015002617006107074,
+      "loss": 1.247,
+      "step": 1411
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.6421436071395874,
+      "learning_rate": 0.0001498982481379795,
+      "loss": 1.4423,
+      "step": 1412
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.2297186851501465,
+      "learning_rate": 0.00014977029905594719,
+      "loss": 1.1478,
+      "step": 1413
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.445056676864624,
+      "learning_rate": 0.00014964232295454027,
+      "loss": 0.9675,
+      "step": 1414
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.2162317037582397,
+      "learning_rate": 0.0001495143199733548,
+      "loss": 0.7747,
+      "step": 1415
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.1276546716690063,
+      "learning_rate": 0.00014938629025201605,
+      "loss": 0.7539,
+      "step": 1416
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.2960880994796753,
+      "learning_rate": 0.00014925823393017846,
+      "loss": 1.0838,
+      "step": 1417
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.4467681646347046,
+      "learning_rate": 0.00014913015114752554,
+      "loss": 1.2902,
+      "step": 1418
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.0221946239471436,
+      "learning_rate": 0.00014900204204376964,
+      "loss": 0.7521,
+      "step": 1419
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.7419352531433105,
+      "learning_rate": 0.00014887390675865183,
+      "loss": 1.3662,
+      "step": 1420
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.406962275505066,
+      "learning_rate": 0.00014874574543194172,
+      "loss": 0.8323,
+      "step": 1421
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.3675340414047241,
+      "learning_rate": 0.0001486175582034373,
+      "loss": 1.1238,
+      "step": 1422
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.2011470794677734,
+      "learning_rate": 0.00014848934521296492,
+      "loss": 0.6788,
+      "step": 1423
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.2760342359542847,
+      "learning_rate": 0.0001483611066003789,
+      "loss": 1.1523,
+      "step": 1424
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.3156168460845947,
+      "learning_rate": 0.00014823284250556164,
+      "loss": 1.3448,
+      "step": 1425
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.3862630128860474,
+      "learning_rate": 0.00014810455306842312,
+      "loss": 0.928,
+      "step": 1426
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.1149640083312988,
+      "learning_rate": 0.00014797623842890132,
+      "loss": 1.0698,
+      "step": 1427
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.6064804792404175,
+      "learning_rate": 0.0001478478987269613,
+      "loss": 1.1212,
+      "step": 1428
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.2255253791809082,
+      "learning_rate": 0.00014771953410259577,
+      "loss": 1.0053,
+      "step": 1429
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.1613675355911255,
+      "learning_rate": 0.0001475911446958245,
+      "loss": 0.8965,
+      "step": 1430
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.42286217212677,
+      "learning_rate": 0.00014746273064669425,
+      "loss": 1.2165,
+      "step": 1431
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.0494049787521362,
+      "learning_rate": 0.00014733429209527876,
+      "loss": 0.7192,
+      "step": 1432
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.020506739616394,
+      "learning_rate": 0.00014720582918167845,
+      "loss": 0.7034,
+      "step": 1433
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.3031187057495117,
+      "learning_rate": 0.00014707734204602027,
+      "loss": 0.7535,
+      "step": 1434
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.0941170454025269,
+      "learning_rate": 0.0001469488308284577,
+      "loss": 0.8199,
+      "step": 1435
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.1094365119934082,
+      "learning_rate": 0.00014682029566917042,
+      "loss": 0.6207,
+      "step": 1436
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.0460867881774902,
+      "learning_rate": 0.00014669173670836416,
+      "loss": 0.8711,
+      "step": 1437
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.3944430351257324,
+      "learning_rate": 0.00014656315408627076,
+      "loss": 0.9551,
+      "step": 1438
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.3652623891830444,
+      "learning_rate": 0.00014643454794314775,
+      "loss": 0.9262,
+      "step": 1439
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.2555691003799438,
+      "learning_rate": 0.00014630591841927838,
+      "loss": 1.0118,
+      "step": 1440
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.467557668685913,
+      "learning_rate": 0.00014617726565497138,
+      "loss": 1.202,
+      "step": 1441
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.7499139308929443,
+      "learning_rate": 0.00014604858979056084,
+      "loss": 0.3857,
+      "step": 1442
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.6192408800125122,
+      "learning_rate": 0.00014591989096640604,
+      "loss": 1.1809,
+      "step": 1443
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.4283156394958496,
+      "learning_rate": 0.00014579116932289128,
+      "loss": 0.9565,
+      "step": 1444
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.0685770511627197,
+      "learning_rate": 0.0001456624250004258,
+      "loss": 0.7835,
+      "step": 1445
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.3966482877731323,
+      "learning_rate": 0.00014553365813944351,
+      "loss": 1.1632,
+      "step": 1446
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.2985618114471436,
+      "learning_rate": 0.00014540486888040306,
+      "loss": 0.8087,
+      "step": 1447
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.616455316543579,
+      "learning_rate": 0.00014527605736378731,
+      "loss": 0.9968,
+      "step": 1448
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.1916112899780273,
+      "learning_rate": 0.00014514722373010357,
+      "loss": 1.1061,
+      "step": 1449
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.2969380617141724,
+      "learning_rate": 0.00014501836811988322,
+      "loss": 0.825,
+      "step": 1450
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.3342739343643188,
+      "learning_rate": 0.00014488949067368158,
+      "loss": 1.0068,
+      "step": 1451
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.2201863527297974,
+      "learning_rate": 0.00014476059153207779,
+      "loss": 0.8792,
+      "step": 1452
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.213496446609497,
+      "learning_rate": 0.00014463167083567476,
+      "loss": 0.843,
+      "step": 1453
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.3563382625579834,
+      "learning_rate": 0.00014450272872509878,
+      "loss": 1.0632,
+      "step": 1454
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.2063534259796143,
+      "learning_rate": 0.00014437376534099958,
+      "loss": 0.7342,
+      "step": 1455
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.9122269749641418,
+      "learning_rate": 0.00014424478082405003,
+      "loss": 0.4485,
+      "step": 1456
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.608677625656128,
+      "learning_rate": 0.0001441157753149461,
+      "loss": 1.0279,
+      "step": 1457
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.0568768978118896,
+      "learning_rate": 0.0001439867489544066,
+      "loss": 0.6611,
+      "step": 1458
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.5482009649276733,
+      "learning_rate": 0.0001438577018831733,
+      "loss": 1.3599,
+      "step": 1459
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.0343140363693237,
+      "learning_rate": 0.00014372863424201018,
+      "loss": 1.0214,
+      "step": 1460
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.2698018550872803,
+      "learning_rate": 0.000143599546171704,
+      "loss": 0.9969,
+      "step": 1461
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.0547878742218018,
+      "learning_rate": 0.0001434704378130637,
+      "loss": 0.9606,
+      "step": 1462
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.2461342811584473,
+      "learning_rate": 0.00014334130930692023,
+      "loss": 1.1568,
+      "step": 1463
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.1943764686584473,
+      "learning_rate": 0.00014321216079412667,
+      "loss": 0.8852,
+      "step": 1464
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.0265756845474243,
+      "learning_rate": 0.0001430829924155578,
+      "loss": 0.8125,
+      "step": 1465
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.0927410125732422,
+      "learning_rate": 0.0001429538043121102,
+      "loss": 0.8627,
+      "step": 1466
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.9594815969467163,
+      "learning_rate": 0.00014282459662470193,
+      "loss": 0.7131,
+      "step": 1467
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.190256118774414,
+      "learning_rate": 0.00014269536949427228,
+      "loss": 0.8461,
+      "step": 1468
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.3381212949752808,
+      "learning_rate": 0.00014256612306178193,
+      "loss": 1.261,
+      "step": 1469
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.52683687210083,
+      "learning_rate": 0.0001424368574682125,
+      "loss": 1.0221,
+      "step": 1470
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.3034223318099976,
+      "learning_rate": 0.0001423075728545666,
+      "loss": 0.9496,
+      "step": 1471
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.6940810680389404,
+      "learning_rate": 0.0001421782693618674,
+      "loss": 1.3889,
+      "step": 1472
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.9740185737609863,
+      "learning_rate": 0.000142048947131159,
+      "loss": 0.8512,
+      "step": 1473
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.2997483015060425,
+      "learning_rate": 0.00014191960630350554,
+      "loss": 0.9987,
+      "step": 1474
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.3484057188034058,
+      "learning_rate": 0.00014179024701999174,
+      "loss": 0.814,
+      "step": 1475
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.665818691253662,
+      "learning_rate": 0.00014166086942172238,
+      "loss": 0.9987,
+      "step": 1476
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.8933560252189636,
+      "learning_rate": 0.0001415314736498221,
+      "loss": 0.6541,
+      "step": 1477
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.2219523191452026,
+      "learning_rate": 0.00014140205984543555,
+      "loss": 0.8712,
+      "step": 1478
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.3110696077346802,
+      "learning_rate": 0.00014127262814972688,
+      "loss": 0.7635,
+      "step": 1479
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.2418899536132812,
+      "learning_rate": 0.00014114317870387986,
+      "loss": 0.7825,
+      "step": 1480
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.1317052841186523,
+      "learning_rate": 0.0001410137116490976,
+      "loss": 0.9781,
+      "step": 1481
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.1634619235992432,
+      "learning_rate": 0.0001408842271266024,
+      "loss": 0.706,
+      "step": 1482
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.0847632884979248,
+      "learning_rate": 0.0001407547252776356,
+      "loss": 0.9616,
+      "step": 1483
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.4543782472610474,
+      "learning_rate": 0.00014062520624345752,
+      "loss": 0.8445,
+      "step": 1484
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.081655502319336,
+      "learning_rate": 0.00014049567016534716,
+      "loss": 0.7163,
+      "step": 1485
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.9920649528503418,
+      "learning_rate": 0.00014036611718460203,
+      "loss": 0.5961,
+      "step": 1486
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.379042387008667,
+      "learning_rate": 0.0001402365474425383,
+      "loss": 1.1791,
+      "step": 1487
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.5765132904052734,
+      "learning_rate": 0.0001401069610804902,
+      "loss": 1.2701,
+      "step": 1488
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.1928505897521973,
+      "learning_rate": 0.00013997735823981016,
+      "loss": 0.8722,
+      "step": 1489
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.1438642740249634,
+      "learning_rate": 0.00013984773906186874,
+      "loss": 0.711,
+      "step": 1490
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.0099387168884277,
+      "learning_rate": 0.00013971810368805404,
+      "loss": 0.6834,
+      "step": 1491
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.1511831283569336,
+      "learning_rate": 0.00013958845225977204,
+      "loss": 0.8102,
+      "step": 1492
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.2723909616470337,
+      "learning_rate": 0.0001394587849184461,
+      "loss": 0.8983,
+      "step": 1493
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.0919370651245117,
+      "learning_rate": 0.00013932910180551714,
+      "loss": 0.5822,
+      "step": 1494
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.42845618724823,
+      "learning_rate": 0.000139199403062443,
+      "loss": 0.8466,
+      "step": 1495
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.8288830518722534,
+      "learning_rate": 0.00013906968883069883,
+      "loss": 1.0043,
+      "step": 1496
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.3897336721420288,
+      "learning_rate": 0.00013893995925177643,
+      "loss": 0.8318,
+      "step": 1497
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.1646203994750977,
+      "learning_rate": 0.00013881021446718458,
+      "loss": 0.8964,
+      "step": 1498
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.2308577299118042,
+      "learning_rate": 0.0001386804546184485,
+      "loss": 0.7555,
+      "step": 1499
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.551567554473877,
+      "learning_rate": 0.00013855067984710988,
+      "loss": 1.4192,
+      "step": 1500
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.1755435466766357,
+      "learning_rate": 0.00013842089029472671,
+      "loss": 0.8243,
+      "step": 1501
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.1090277433395386,
+      "learning_rate": 0.000138291086102873,
+      "loss": 0.7283,
+      "step": 1502
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.368507981300354,
+      "learning_rate": 0.0001381612674131389,
+      "loss": 1.298,
+      "step": 1503
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.3204469680786133,
+      "learning_rate": 0.00013803143436713023,
+      "loss": 0.9995,
+      "step": 1504
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.4416377544403076,
+      "learning_rate": 0.0001379015871064685,
+      "loss": 0.9646,
+      "step": 1505
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.3736774921417236,
+      "learning_rate": 0.00013777172577279084,
+      "loss": 1.0291,
+      "step": 1506
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.1656876802444458,
+      "learning_rate": 0.00013764185050774957,
+      "loss": 0.9058,
+      "step": 1507
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.9170559644699097,
+      "learning_rate": 0.0001375119614530123,
+      "loss": 0.686,
+      "step": 1508
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.2503794431686401,
+      "learning_rate": 0.00013738205875026165,
+      "loss": 0.6152,
+      "step": 1509
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.9299453496932983,
+      "learning_rate": 0.0001372521425411951,
+      "loss": 0.5456,
+      "step": 1510
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.043879747390747,
+      "learning_rate": 0.00013712221296752493,
+      "loss": 0.6329,
+      "step": 1511
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.9172692894935608,
+      "learning_rate": 0.00013699227017097793,
+      "loss": 0.6599,
+      "step": 1512
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.0393551588058472,
+      "learning_rate": 0.00013686231429329543,
+      "loss": 0.7241,
+      "step": 1513
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.1035585403442383,
+      "learning_rate": 0.00013673234547623283,
+      "loss": 0.7771,
+      "step": 1514
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.099898099899292,
+      "learning_rate": 0.00013660236386155988,
+      "loss": 0.7152,
+      "step": 1515
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.1429855823516846,
+      "learning_rate": 0.00013647236959106004,
+      "loss": 0.8427,
+      "step": 1516
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.9771751761436462,
+      "learning_rate": 0.00013634236280653085,
+      "loss": 0.6092,
+      "step": 1517
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.1621931791305542,
+      "learning_rate": 0.00013621234364978325,
+      "loss": 0.9588,
+      "step": 1518
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.0901761054992676,
+      "learning_rate": 0.00013608231226264179,
+      "loss": 0.8274,
+      "step": 1519
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.3528239727020264,
+      "learning_rate": 0.00013595226878694442,
+      "loss": 1.2063,
+      "step": 1520
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.6294797658920288,
+      "learning_rate": 0.00013582221336454214,
+      "loss": 1.2005,
+      "step": 1521
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.2546576261520386,
+      "learning_rate": 0.0001356921461372991,
+      "loss": 0.8308,
+      "step": 1522
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.2436532974243164,
+      "learning_rate": 0.0001355620672470922,
+      "loss": 0.9026,
+      "step": 1523
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.2715680599212646,
+      "learning_rate": 0.00013543197683581123,
+      "loss": 0.8447,
+      "step": 1524
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.0670286417007446,
+      "learning_rate": 0.00013530187504535844,
+      "loss": 0.7272,
+      "step": 1525
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.9957221746444702,
+      "learning_rate": 0.0001351717620176484,
+      "loss": 0.726,
+      "step": 1526
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.3824573755264282,
+      "learning_rate": 0.00013504163789460823,
+      "loss": 0.9593,
+      "step": 1527
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.2613593339920044,
+      "learning_rate": 0.00013491150281817675,
+      "loss": 1.1716,
+      "step": 1528
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.1571170091629028,
+      "learning_rate": 0.00013478135693030513,
+      "loss": 0.8439,
+      "step": 1529
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.3491780757904053,
+      "learning_rate": 0.00013465120037295606,
+      "loss": 1.152,
+      "step": 1530
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.3383604288101196,
+      "learning_rate": 0.000134521033288104,
+      "loss": 0.9002,
+      "step": 1531
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.0674033164978027,
+      "learning_rate": 0.00013439085581773481,
+      "loss": 0.7625,
+      "step": 1532
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.0441478490829468,
+      "learning_rate": 0.00013426066810384572,
+      "loss": 0.6468,
+      "step": 1533
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.3126263618469238,
+      "learning_rate": 0.0001341304702884452,
+      "loss": 1.0375,
+      "step": 1534
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.1851307153701782,
+      "learning_rate": 0.00013400026251355257,
+      "loss": 1.0054,
+      "step": 1535
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.0638484954833984,
+      "learning_rate": 0.0001338700449211982,
+      "loss": 0.6203,
+      "step": 1536
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.396504521369934,
+      "learning_rate": 0.00013373981765342304,
+      "loss": 1.4193,
+      "step": 1537
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.1101491451263428,
+      "learning_rate": 0.00013360958085227866,
+      "loss": 0.6708,
+      "step": 1538
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.9452186226844788,
+      "learning_rate": 0.000133479334659827,
+      "loss": 0.8028,
+      "step": 1539
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.9814640283584595,
+      "learning_rate": 0.0001333490792181402,
+      "loss": 0.8605,
+      "step": 1540
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.123543620109558,
+      "learning_rate": 0.0001332188146693006,
+      "loss": 0.738,
+      "step": 1541
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.3086490631103516,
+      "learning_rate": 0.00013308854115540042,
+      "loss": 1.0046,
+      "step": 1542
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.9039118885993958,
+      "learning_rate": 0.00013295825881854168,
+      "loss": 0.7189,
+      "step": 1543
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.2540236711502075,
+      "learning_rate": 0.00013282796780083588,
+      "loss": 1.0421,
+      "step": 1544
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.2566171884536743,
+      "learning_rate": 0.00013269766824440424,
+      "loss": 0.9669,
+      "step": 1545
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.304291844367981,
+      "learning_rate": 0.00013256736029137705,
+      "loss": 0.8689,
+      "step": 1546
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.2309794425964355,
+      "learning_rate": 0.00013243704408389393,
+      "loss": 1.1371,
+      "step": 1547
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.1405993700027466,
+      "learning_rate": 0.00013230671976410344,
+      "loss": 1.0609,
+      "step": 1548
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.1470849514007568,
+      "learning_rate": 0.00013217638747416296,
+      "loss": 0.949,
+      "step": 1549
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.3759626150131226,
+      "learning_rate": 0.00013204604735623867,
+      "loss": 1.3161,
+      "step": 1550
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.4023290872573853,
+      "learning_rate": 0.00013191569955250512,
+      "loss": 1.1112,
+      "step": 1551
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.2357980012893677,
+      "learning_rate": 0.0001317853442051454,
+      "loss": 1.0148,
+      "step": 1552
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.2000268697738647,
+      "learning_rate": 0.00013165498145635072,
+      "loss": 0.8196,
+      "step": 1553
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.956527829170227,
+      "learning_rate": 0.0001315246114483205,
+      "loss": 0.5612,
+      "step": 1554
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.0706440210342407,
+      "learning_rate": 0.00013139423432326187,
+      "loss": 1.0052,
+      "step": 1555
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.8552688956260681,
+      "learning_rate": 0.0001312638502233899,
+      "loss": 0.6562,
+      "step": 1556
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.3808249235153198,
+      "learning_rate": 0.00013113345929092728,
+      "loss": 0.9808,
+      "step": 1557
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.2328780889511108,
+      "learning_rate": 0.00013100306166810395,
+      "loss": 0.996,
+      "step": 1558
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.611670732498169,
+      "learning_rate": 0.0001308726574971574,
+      "loss": 1.3974,
+      "step": 1559
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.2272011041641235,
+      "learning_rate": 0.00013074224692033202,
+      "loss": 0.7273,
+      "step": 1560
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.3213979005813599,
+      "learning_rate": 0.00013061183007987939,
+      "loss": 1.0876,
+      "step": 1561
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.980205237865448,
+      "learning_rate": 0.00013048140711805788,
+      "loss": 0.6217,
+      "step": 1562
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.5037871599197388,
+      "learning_rate": 0.0001303509781771324,
+      "loss": 1.4245,
+      "step": 1563
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.2028981447219849,
+      "learning_rate": 0.00013022054339937458,
+      "loss": 0.9513,
+      "step": 1564
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.123295545578003,
+      "learning_rate": 0.00013009010292706224,
+      "loss": 0.8559,
+      "step": 1565
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.2569561004638672,
+      "learning_rate": 0.00012995965690247956,
+      "loss": 1.1427,
+      "step": 1566
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.2987000942230225,
+      "learning_rate": 0.00012982920546791667,
+      "loss": 0.8334,
+      "step": 1567
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.2294447422027588,
+      "learning_rate": 0.00012969874876566964,
+      "loss": 1.0278,
+      "step": 1568
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.5399640798568726,
+      "learning_rate": 0.00012956828693804027,
+      "loss": 1.4469,
+      "step": 1569
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.1422955989837646,
+      "learning_rate": 0.00012943782012733598,
+      "loss": 0.682,
+      "step": 1570
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.2182711362838745,
+      "learning_rate": 0.00012930734847586963,
+      "loss": 1.0153,
+      "step": 1571
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.2179765701293945,
+      "learning_rate": 0.00012917687212595934,
+      "loss": 0.5329,
+      "step": 1572
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.2386904954910278,
+      "learning_rate": 0.00012904639121992834,
+      "loss": 0.9261,
+      "step": 1573
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.0264219045639038,
+      "learning_rate": 0.00012891590590010483,
+      "loss": 0.7033,
+      "step": 1574
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.187770962715149,
+      "learning_rate": 0.00012878541630882183,
+      "loss": 1.0614,
+      "step": 1575
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.918658971786499,
+      "learning_rate": 0.00012865492258841709,
+      "loss": 0.6751,
+      "step": 1576
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.899031400680542,
+      "learning_rate": 0.0001285244248812327,
+      "loss": 0.7438,
+      "step": 1577
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.3633794784545898,
+      "learning_rate": 0.00012839392332961538,
+      "loss": 0.8423,
+      "step": 1578
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.436004400253296,
+      "learning_rate": 0.0001282634180759157,
+      "loss": 0.9681,
+      "step": 1579
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.1468559503555298,
+      "learning_rate": 0.00012813290926248856,
+      "loss": 0.6503,
+      "step": 1580
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.2126736640930176,
+      "learning_rate": 0.0001280023970316925,
+      "loss": 0.9377,
+      "step": 1581
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.6733336448669434,
+      "learning_rate": 0.0001278718815258899,
+      "loss": 0.9995,
+      "step": 1582
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.2680801153182983,
+      "learning_rate": 0.00012774136288744685,
+      "loss": 0.9284,
+      "step": 1583
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.2771387100219727,
+      "learning_rate": 0.00012761084125873266,
+      "loss": 0.9957,
+      "step": 1584
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.2861063480377197,
+      "learning_rate": 0.0001274803167821199,
+      "loss": 0.9192,
+      "step": 1585
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.2119808197021484,
+      "learning_rate": 0.0001273497895999844,
+      "loss": 0.8964,
+      "step": 1586
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.9807702302932739,
+      "learning_rate": 0.00012721925985470481,
+      "loss": 0.6691,
+      "step": 1587
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.4058620929718018,
+      "learning_rate": 0.0001270887276886626,
+      "loss": 1.1064,
+      "step": 1588
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.1498188972473145,
+      "learning_rate": 0.000126958193244242,
+      "loss": 0.9539,
+      "step": 1589
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.8641445636749268,
+      "learning_rate": 0.0001268276566638295,
+      "loss": 0.7446,
+      "step": 1590
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.9338680505752563,
+      "learning_rate": 0.00012669711808981413,
+      "loss": 0.6265,
+      "step": 1591
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.3480240106582642,
+      "learning_rate": 0.00012656657766458698,
+      "loss": 1.1703,
+      "step": 1592
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.0655312538146973,
+      "learning_rate": 0.00012643603553054116,
+      "loss": 1.2219,
+      "step": 1593
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.325788974761963,
+      "learning_rate": 0.00012630549183007176,
+      "loss": 0.8702,
+      "step": 1594
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.2315642833709717,
+      "learning_rate": 0.0001261749467055754,
+      "loss": 0.6193,
+      "step": 1595
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.9755103588104248,
+      "learning_rate": 0.0001260444002994504,
+      "loss": 0.9851,
+      "step": 1596
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.2712702751159668,
+      "learning_rate": 0.00012591385275409638,
+      "loss": 1.2029,
+      "step": 1597
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.4285528659820557,
+      "learning_rate": 0.00012578330421191428,
+      "loss": 1.1168,
+      "step": 1598
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.7838413715362549,
+      "learning_rate": 0.0001256527548153061,
+      "loss": 1.12,
+      "step": 1599
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.9507400393486023,
+      "learning_rate": 0.00012552220470667472,
+      "loss": 0.6351,
+      "step": 1600
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.8768706917762756,
+      "learning_rate": 0.0001253916540284239,
+      "loss": 0.5172,
+      "step": 1601
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.1305594444274902,
+      "learning_rate": 0.0001252611029229579,
+      "loss": 0.7985,
+      "step": 1602
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.086591124534607,
+      "learning_rate": 0.0001251305515326816,
+      "loss": 0.7133,
+      "step": 1603
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.212422251701355,
+      "learning_rate": 0.000125,
+      "loss": 0.8705,
+      "step": 1604
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.0888148546218872,
+      "learning_rate": 0.00012486944846731843,
+      "loss": 0.8609,
+      "step": 1605
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.300306797027588,
+      "learning_rate": 0.0001247388970770421,
+      "loss": 0.8719,
+      "step": 1606
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.213152289390564,
+      "learning_rate": 0.00012460834597157613,
+      "loss": 0.7651,
+      "step": 1607
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.1947771310806274,
+      "learning_rate": 0.00012447779529332532,
+      "loss": 0.7965,
+      "step": 1608
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.9396220445632935,
+      "learning_rate": 0.00012434724518469394,
+      "loss": 0.7223,
+      "step": 1609
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.363637924194336,
+      "learning_rate": 0.00012421669578808575,
+      "loss": 0.8611,
+      "step": 1610
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.1343002319335938,
+      "learning_rate": 0.00012408614724590365,
+      "loss": 0.7998,
+      "step": 1611
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.739519476890564,
+      "learning_rate": 0.00012395559970054964,
+      "loss": 1.5514,
+      "step": 1612
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.0173429250717163,
+      "learning_rate": 0.00012382505329442463,
+      "loss": 0.6983,
+      "step": 1613
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.0030152797698975,
+      "learning_rate": 0.0001236945081699283,
+      "loss": 0.9811,
+      "step": 1614
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.199392318725586,
+      "learning_rate": 0.00012356396446945887,
+      "loss": 1.2252,
+      "step": 1615
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.9313424229621887,
+      "learning_rate": 0.000123433422335413,
+      "loss": 0.6646,
+      "step": 1616
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.1196904182434082,
+      "learning_rate": 0.00012330288191018588,
+      "loss": 0.8581,
+      "step": 1617
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.0649549961090088,
+      "learning_rate": 0.00012317234333617048,
+      "loss": 0.5758,
+      "step": 1618
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.8960950970649719,
+      "learning_rate": 0.000123041806755758,
+      "loss": 0.6896,
+      "step": 1619
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.4193485975265503,
+      "learning_rate": 0.00012291127231133737,
+      "loss": 1.1335,
+      "step": 1620
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.1770983934402466,
+      "learning_rate": 0.0001227807401452952,
+      "loss": 0.9083,
+      "step": 1621
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.001149296760559,
+      "learning_rate": 0.0001226502104000156,
+      "loss": 0.8143,
+      "step": 1622
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.3807704448699951,
+      "learning_rate": 0.0001225196832178801,
+      "loss": 1.4339,
+      "step": 1623
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.19136381149292,
+      "learning_rate": 0.00012238915874126737,
+      "loss": 0.8244,
+      "step": 1624
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.25921630859375,
+      "learning_rate": 0.00012225863711255315,
+      "loss": 1.0063,
+      "step": 1625
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.991645097732544,
+      "learning_rate": 0.0001221281184741101,
+      "loss": 0.7513,
+      "step": 1626
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.4019756317138672,
+      "learning_rate": 0.00012199760296830754,
+      "loss": 1.0369,
+      "step": 1627
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.0501247644424438,
+      "learning_rate": 0.00012186709073751149,
+      "loss": 0.9839,
+      "step": 1628
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.8534177541732788,
+      "learning_rate": 0.0001217365819240843,
+      "loss": 0.5141,
+      "step": 1629
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.4940394163131714,
+      "learning_rate": 0.00012160607667038463,
+      "loss": 1.0276,
+      "step": 1630
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.013058066368103,
+      "learning_rate": 0.00012147557511876729,
+      "loss": 0.985,
+      "step": 1631
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.1882599592208862,
+      "learning_rate": 0.00012134507741158296,
+      "loss": 1.0577,
+      "step": 1632
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.571408748626709,
+      "learning_rate": 0.0001212145836911782,
+      "loss": 1.0311,
+      "step": 1633
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.175654411315918,
+      "learning_rate": 0.00012108409409989524,
+      "loss": 0.7299,
+      "step": 1634
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.0675251483917236,
+      "learning_rate": 0.00012095360878007173,
+      "loss": 0.7819,
+      "step": 1635
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.181757926940918,
+      "learning_rate": 0.00012082312787404072,
+      "loss": 0.8871,
+      "step": 1636
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.2116355895996094,
+      "learning_rate": 0.0001206926515241304,
+      "loss": 0.9785,
+      "step": 1637
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.3060861825942993,
+      "learning_rate": 0.00012056217987266403,
+      "loss": 0.9901,
+      "step": 1638
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.3970521688461304,
+      "learning_rate": 0.00012043171306195977,
+      "loss": 0.8015,
+      "step": 1639
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.1799612045288086,
+      "learning_rate": 0.00012030125123433038,
+      "loss": 1.1066,
+      "step": 1640
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.2180298566818237,
+      "learning_rate": 0.00012017079453208332,
+      "loss": 0.8753,
+      "step": 1641
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.4600939750671387,
+      "learning_rate": 0.00012004034309752044,
+      "loss": 0.8927,
+      "step": 1642
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.1986831426620483,
+      "learning_rate": 0.00011990989707293774,
+      "loss": 1.1189,
+      "step": 1643
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.06000816822052,
+      "learning_rate": 0.00011977945660062543,
+      "loss": 0.7865,
+      "step": 1644
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.4312909841537476,
+      "learning_rate": 0.00011964902182286759,
+      "loss": 1.0943,
+      "step": 1645
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.35890531539917,
+      "learning_rate": 0.00011951859288194214,
+      "loss": 0.8092,
+      "step": 1646
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.47928786277771,
+      "learning_rate": 0.0001193881699201206,
+      "loss": 1.1237,
+      "step": 1647
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.0819437503814697,
+      "learning_rate": 0.00011925775307966799,
+      "loss": 0.6155,
+      "step": 1648
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.1378880739212036,
+      "learning_rate": 0.00011912734250284264,
+      "loss": 0.9366,
+      "step": 1649
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.1234723329544067,
+      "learning_rate": 0.00011899693833189607,
+      "loss": 0.7372,
+      "step": 1650
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.1745365858078003,
+      "learning_rate": 0.00011886654070907275,
+      "loss": 0.626,
+      "step": 1651
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.4009250402450562,
+      "learning_rate": 0.00011873614977661011,
+      "loss": 1.2027,
+      "step": 1652
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.167280673980713,
+      "learning_rate": 0.00011860576567673814,
+      "loss": 0.8992,
+      "step": 1653
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.1178878545761108,
+      "learning_rate": 0.00011847538855167954,
+      "loss": 0.6062,
+      "step": 1654
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.3119362592697144,
+      "learning_rate": 0.00011834501854364929,
+      "loss": 1.0858,
+      "step": 1655
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.1016749143600464,
+      "learning_rate": 0.00011821465579485464,
+      "loss": 0.8645,
+      "step": 1656
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.315070390701294,
+      "learning_rate": 0.00011808430044749492,
+      "loss": 0.9478,
+      "step": 1657
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.3839716911315918,
+      "learning_rate": 0.00011795395264376138,
+      "loss": 0.8104,
+      "step": 1658
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.1264443397521973,
+      "learning_rate": 0.00011782361252583709,
+      "loss": 0.6846,
+      "step": 1659
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.3495508432388306,
+      "learning_rate": 0.00011769328023589661,
+      "loss": 1.0685,
+      "step": 1660
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.1462335586547852,
+      "learning_rate": 0.0001175629559161061,
+      "loss": 0.7131,
+      "step": 1661
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.0215023756027222,
+      "learning_rate": 0.000117432639708623,
+      "loss": 0.8112,
+      "step": 1662
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.1370227336883545,
+      "learning_rate": 0.0001173023317555958,
+      "loss": 0.7725,
+      "step": 1663
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.4763069152832031,
+      "learning_rate": 0.00011717203219916411,
+      "loss": 0.9204,
+      "step": 1664
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.1759175062179565,
+      "learning_rate": 0.00011704174118145832,
+      "loss": 1.1234,
+      "step": 1665
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.1167455911636353,
+      "learning_rate": 0.00011691145884459956,
+      "loss": 0.7173,
+      "step": 1666
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.054011583328247,
+      "learning_rate": 0.00011678118533069937,
+      "loss": 0.6072,
+      "step": 1667
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.2380262613296509,
+      "learning_rate": 0.0001166509207818598,
+      "loss": 0.8174,
+      "step": 1668
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.1825151443481445,
+      "learning_rate": 0.00011652066534017304,
+      "loss": 0.5941,
+      "step": 1669
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.345412015914917,
+      "learning_rate": 0.00011639041914772138,
+      "loss": 0.8084,
+      "step": 1670
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.1241743564605713,
+      "learning_rate": 0.00011626018234657698,
+      "loss": 0.8581,
+      "step": 1671
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.2227263450622559,
+      "learning_rate": 0.00011612995507880181,
+      "loss": 0.9555,
+      "step": 1672
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.9729316234588623,
+      "learning_rate": 0.00011599973748644745,
+      "loss": 1.0784,
+      "step": 1673
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.5396397113800049,
+      "learning_rate": 0.00011586952971155484,
+      "loss": 1.3129,
+      "step": 1674
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.2963906526565552,
+      "learning_rate": 0.00011573933189615429,
+      "loss": 0.9409,
+      "step": 1675
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.539739966392517,
+      "learning_rate": 0.00011560914418226522,
+      "loss": 1.3643,
+      "step": 1676
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.143496036529541,
+      "learning_rate": 0.00011547896671189602,
+      "loss": 0.8545,
+      "step": 1677
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.3482871055603027,
+      "learning_rate": 0.00011534879962704396,
+      "loss": 0.7935,
+      "step": 1678
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.1831111907958984,
+      "learning_rate": 0.00011521864306969489,
+      "loss": 0.8544,
+      "step": 1679
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.034184217453003,
+      "learning_rate": 0.00011508849718182328,
+      "loss": 0.831,
+      "step": 1680
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.2171016931533813,
+      "learning_rate": 0.00011495836210539185,
+      "loss": 0.9327,
+      "step": 1681
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.1705622673034668,
+      "learning_rate": 0.0001148282379823516,
+      "loss": 0.7978,
+      "step": 1682
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.2037105560302734,
+      "learning_rate": 0.00011469812495464164,
+      "loss": 0.9249,
+      "step": 1683
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.4301562309265137,
+      "learning_rate": 0.00011456802316418879,
+      "loss": 1.2376,
+      "step": 1684
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.8832158446311951,
+      "learning_rate": 0.00011443793275290783,
+      "loss": 0.5311,
+      "step": 1685
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.06523597240448,
+      "learning_rate": 0.00011430785386270095,
+      "loss": 0.74,
+      "step": 1686
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.0592454671859741,
+      "learning_rate": 0.00011417778663545786,
+      "loss": 0.9425,
+      "step": 1687
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.1511181592941284,
+      "learning_rate": 0.00011404773121305557,
+      "loss": 0.991,
+      "step": 1688
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.0210840702056885,
+      "learning_rate": 0.0001139176877373582,
+      "loss": 0.7791,
+      "step": 1689
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.1962254047393799,
+      "learning_rate": 0.00011378765635021676,
+      "loss": 1.0616,
+      "step": 1690
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.127206563949585,
+      "learning_rate": 0.00011365763719346918,
+      "loss": 0.9061,
+      "step": 1691
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.847680926322937,
+      "learning_rate": 0.00011352763040893996,
+      "loss": 0.5002,
+      "step": 1692
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.2024614810943604,
+      "learning_rate": 0.00011339763613844014,
+      "loss": 0.7685,
+      "step": 1693
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.1078310012817383,
+      "learning_rate": 0.00011326765452376717,
+      "loss": 0.8315,
+      "step": 1694
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.2252180576324463,
+      "learning_rate": 0.0001131376857067046,
+      "loss": 0.8286,
+      "step": 1695
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.832125186920166,
+      "learning_rate": 0.00011300772982902208,
+      "loss": 0.4677,
+      "step": 1696
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.0959038734436035,
+      "learning_rate": 0.0001128777870324751,
+      "loss": 0.8417,
+      "step": 1697
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.1298578977584839,
+      "learning_rate": 0.00011274785745880492,
+      "loss": 0.927,
+      "step": 1698
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.3126767873764038,
+      "learning_rate": 0.0001126179412497384,
+      "loss": 0.9112,
+      "step": 1699
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.009222388267517,
+      "learning_rate": 0.0001124880385469877,
+      "loss": 1.0107,
+      "step": 1700
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.049607515335083,
+      "learning_rate": 0.00011235814949225044,
+      "loss": 0.7388,
+      "step": 1701
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.1917518377304077,
+      "learning_rate": 0.00011222827422720916,
+      "loss": 0.8073,
+      "step": 1702
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.2827389240264893,
+      "learning_rate": 0.00011209841289353151,
+      "loss": 0.7094,
+      "step": 1703
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.8789745569229126,
+      "learning_rate": 0.0001119685656328698,
+      "loss": 0.5815,
+      "step": 1704
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.9101778268814087,
+      "learning_rate": 0.00011183873258686112,
+      "loss": 0.5959,
+      "step": 1705
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.9749009013175964,
+      "learning_rate": 0.00011170891389712704,
+      "loss": 0.636,
+      "step": 1706
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.2081713676452637,
+      "learning_rate": 0.00011157910970527335,
+      "loss": 0.9361,
+      "step": 1707
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.3563265800476074,
+      "learning_rate": 0.00011144932015289017,
+      "loss": 0.9394,
+      "step": 1708
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.036575436592102,
+      "learning_rate": 0.00011131954538155153,
+      "loss": 0.8256,
+      "step": 1709
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.026749610900879,
+      "learning_rate": 0.00011118978553281541,
+      "loss": 0.7384,
+      "step": 1710
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.8870358467102051,
+      "learning_rate": 0.00011106004074822356,
+      "loss": 0.5476,
+      "step": 1711
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.9484975337982178,
+      "learning_rate": 0.00011093031116930122,
+      "loss": 0.655,
+      "step": 1712
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.34621000289917,
+      "learning_rate": 0.000110800596937557,
+      "loss": 1.0488,
+      "step": 1713
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.0034739971160889,
+      "learning_rate": 0.0001106708981944829,
+      "loss": 0.6783,
+      "step": 1714
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.5905283689498901,
+      "learning_rate": 0.00011054121508155387,
+      "loss": 1.4052,
+      "step": 1715
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.4879212379455566,
+      "learning_rate": 0.00011041154774022798,
+      "loss": 1.1154,
+      "step": 1716
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.1741530895233154,
+      "learning_rate": 0.00011028189631194599,
+      "loss": 0.8851,
+      "step": 1717
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.1095561981201172,
+      "learning_rate": 0.00011015226093813129,
+      "loss": 0.8269,
+      "step": 1718
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.4847856760025024,
+      "learning_rate": 0.00011002264176018983,
+      "loss": 1.2673,
+      "step": 1719
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.4990216493606567,
+      "learning_rate": 0.00010989303891950983,
+      "loss": 1.5933,
+      "step": 1720
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.3621212244033813,
+      "learning_rate": 0.00010976345255746172,
+      "loss": 0.7648,
+      "step": 1721
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.175217866897583,
+      "learning_rate": 0.00010963388281539797,
+      "loss": 0.9908,
+      "step": 1722
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.1372090578079224,
+      "learning_rate": 0.00010950432983465288,
+      "loss": 0.887,
+      "step": 1723
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.3557263612747192,
+      "learning_rate": 0.0001093747937565425,
+      "loss": 1.0648,
+      "step": 1724
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.1040289402008057,
+      "learning_rate": 0.00010924527472236439,
+      "loss": 0.5231,
+      "step": 1725
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.4421457052230835,
+      "learning_rate": 0.00010911577287339762,
+      "loss": 1.5833,
+      "step": 1726
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.1788628101348877,
+      "learning_rate": 0.00010898628835090243,
+      "loss": 0.8498,
+      "step": 1727
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.0729405879974365,
+      "learning_rate": 0.00010885682129612016,
+      "loss": 0.857,
+      "step": 1728
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.8381689786911011,
+      "learning_rate": 0.00010872737185027318,
+      "loss": 0.5624,
+      "step": 1729
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.0547059774398804,
+      "learning_rate": 0.00010859794015456449,
+      "loss": 0.9163,
+      "step": 1730
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.547420859336853,
+      "learning_rate": 0.00010846852635017794,
+      "loss": 1.1406,
+      "step": 1731
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.2853301763534546,
+      "learning_rate": 0.00010833913057827769,
+      "loss": 0.9673,
+      "step": 1732
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.172054648399353,
+      "learning_rate": 0.00010820975298000828,
+      "loss": 0.8101,
+      "step": 1733
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.3405346870422363,
+      "learning_rate": 0.00010808039369649445,
+      "loss": 0.8742,
+      "step": 1734
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.9736365675926208,
+      "learning_rate": 0.00010795105286884104,
+      "loss": 0.6063,
+      "step": 1735
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.025546908378601,
+      "learning_rate": 0.00010782173063813258,
+      "loss": 0.7712,
+      "step": 1736
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.3535629510879517,
+      "learning_rate": 0.00010769242714543343,
+      "loss": 1.039,
+      "step": 1737
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.2922720909118652,
+      "learning_rate": 0.0001075631425317875,
+      "loss": 1.0313,
+      "step": 1738
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.1379432678222656,
+      "learning_rate": 0.00010743387693821807,
+      "loss": 1.0168,
+      "step": 1739
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.063762903213501,
+      "learning_rate": 0.00010730463050572773,
+      "loss": 0.6007,
+      "step": 1740
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.3603583574295044,
+      "learning_rate": 0.00010717540337529809,
+      "loss": 0.9233,
+      "step": 1741
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.1882977485656738,
+      "learning_rate": 0.00010704619568788979,
+      "loss": 0.8773,
+      "step": 1742
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.9956715106964111,
+      "learning_rate": 0.0001069170075844422,
+      "loss": 0.6026,
+      "step": 1743
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.4287210702896118,
+      "learning_rate": 0.00010678783920587335,
+      "loss": 0.8701,
+      "step": 1744
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.173557996749878,
+      "learning_rate": 0.00010665869069307979,
+      "loss": 0.7308,
+      "step": 1745
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.12995183467865,
+      "learning_rate": 0.00010652956218693632,
+      "loss": 0.9228,
+      "step": 1746
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.1359423398971558,
+      "learning_rate": 0.000106400453828296,
+      "loss": 0.9147,
+      "step": 1747
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.2205817699432373,
+      "learning_rate": 0.00010627136575798981,
+      "loss": 0.9102,
+      "step": 1748
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.2005549669265747,
+      "learning_rate": 0.00010614229811682674,
+      "loss": 0.9014,
+      "step": 1749
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.3272981643676758,
+      "learning_rate": 0.0001060132510455934,
+      "loss": 0.992,
+      "step": 1750
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.143885612487793,
+      "learning_rate": 0.00010588422468505396,
+      "loss": 0.7217,
+      "step": 1751
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.626641869544983,
+      "learning_rate": 0.00010575521917595005,
+      "loss": 1.2493,
+      "step": 1752
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.1745210886001587,
+      "learning_rate": 0.00010562623465900048,
+      "loss": 1.0751,
+      "step": 1753
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.2065612077713013,
+      "learning_rate": 0.00010549727127490125,
+      "loss": 0.7826,
+      "step": 1754
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.4140163660049438,
+      "learning_rate": 0.00010536832916432527,
+      "loss": 0.9993,
+      "step": 1755
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.2656209468841553,
+      "learning_rate": 0.00010523940846792222,
+      "loss": 0.7522,
+      "step": 1756
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.4236201047897339,
+      "learning_rate": 0.00010511050932631841,
+      "loss": 1.0854,
+      "step": 1757
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.815237820148468,
+      "learning_rate": 0.0001049816318801168,
+      "loss": 0.3725,
+      "step": 1758
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.0349665880203247,
+      "learning_rate": 0.0001048527762698964,
+      "loss": 0.8778,
+      "step": 1759
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.2187834978103638,
+      "learning_rate": 0.00010472394263621266,
+      "loss": 0.9184,
+      "step": 1760
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.215148687362671,
+      "learning_rate": 0.00010459513111959695,
+      "loss": 1.087,
+      "step": 1761
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.1747918128967285,
+      "learning_rate": 0.00010446634186055646,
+      "loss": 0.7544,
+      "step": 1762
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.4393612146377563,
+      "learning_rate": 0.00010433757499957423,
+      "loss": 1.2608,
+      "step": 1763
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.0888879299163818,
+      "learning_rate": 0.00010420883067710875,
+      "loss": 0.8795,
+      "step": 1764
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.2865713834762573,
+      "learning_rate": 0.00010408010903359398,
+      "loss": 0.9863,
+      "step": 1765
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.033387303352356,
+      "learning_rate": 0.00010395141020943918,
+      "loss": 0.6285,
+      "step": 1766
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.0320261716842651,
+      "learning_rate": 0.00010382273434502862,
+      "loss": 0.6868,
+      "step": 1767
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.0416369438171387,
+      "learning_rate": 0.00010369408158072164,
+      "loss": 0.8719,
+      "step": 1768
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.0624701976776123,
+      "learning_rate": 0.00010356545205685227,
+      "loss": 0.8254,
+      "step": 1769
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.2965495586395264,
+      "learning_rate": 0.00010343684591372926,
+      "loss": 0.9601,
+      "step": 1770
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.0764302015304565,
+      "learning_rate": 0.00010330826329163587,
+      "loss": 0.5743,
+      "step": 1771
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.0314794778823853,
+      "learning_rate": 0.00010317970433082963,
+      "loss": 0.5785,
+      "step": 1772
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.2629597187042236,
+      "learning_rate": 0.00010305116917154233,
+      "loss": 0.5249,
+      "step": 1773
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.4034559726715088,
+      "learning_rate": 0.00010292265795397974,
+      "loss": 0.8213,
+      "step": 1774
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.548949122428894,
+      "learning_rate": 0.00010279417081832161,
+      "loss": 1.2624,
+      "step": 1775
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.2316782474517822,
+      "learning_rate": 0.00010266570790472129,
+      "loss": 0.8921,
+      "step": 1776
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.1409668922424316,
+      "learning_rate": 0.00010253726935330578,
+      "loss": 0.8301,
+      "step": 1777
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.2139928340911865,
+      "learning_rate": 0.00010240885530417557,
+      "loss": 0.842,
+      "step": 1778
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.1755350828170776,
+      "learning_rate": 0.00010228046589740425,
+      "loss": 1.2071,
+      "step": 1779
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.0287277698516846,
+      "learning_rate": 0.00010215210127303874,
+      "loss": 0.6262,
+      "step": 1780
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.7859232425689697,
+      "learning_rate": 0.00010202376157109869,
+      "loss": 1.2228,
+      "step": 1781
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.0035349130630493,
+      "learning_rate": 0.00010189544693157684,
+      "loss": 0.7594,
+      "step": 1782
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.4260820150375366,
+      "learning_rate": 0.00010176715749443838,
+      "loss": 0.8775,
+      "step": 1783
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.3202534914016724,
+      "learning_rate": 0.00010163889339962109,
+      "loss": 0.9169,
+      "step": 1784
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.4712368249893188,
+      "learning_rate": 0.00010151065478703508,
+      "loss": 1.197,
+      "step": 1785
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.1128965616226196,
+      "learning_rate": 0.00010138244179656271,
+      "loss": 0.7421,
+      "step": 1786
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.981045663356781,
+      "learning_rate": 0.00010125425456805831,
+      "loss": 0.5326,
+      "step": 1787
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.068075180053711,
+      "learning_rate": 0.00010112609324134818,
+      "loss": 0.8425,
+      "step": 1788
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.9965289235115051,
+      "learning_rate": 0.00010099795795623037,
+      "loss": 0.6533,
+      "step": 1789
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.4239498376846313,
+      "learning_rate": 0.00010086984885247446,
+      "loss": 0.9943,
+      "step": 1790
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.2689576148986816,
+      "learning_rate": 0.00010074176606982158,
+      "loss": 0.9002,
+      "step": 1791
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.8137116432189941,
+      "learning_rate": 0.00010061370974798398,
+      "loss": 0.6084,
+      "step": 1792
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.1245088577270508,
+      "learning_rate": 0.0001004856800266452,
+      "loss": 0.6368,
+      "step": 1793
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.4730548858642578,
+      "learning_rate": 0.00010035767704545972,
+      "loss": 1.2741,
+      "step": 1794
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.6720154285430908,
+      "learning_rate": 0.00010022970094405282,
+      "loss": 1.2503,
+      "step": 1795
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.2065510749816895,
+      "learning_rate": 0.00010010175186202051,
+      "loss": 0.8932,
+      "step": 1796
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.3618618249893188,
+      "learning_rate": 9.997382993892925e-05,
+      "loss": 0.8034,
+      "step": 1797
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.9418081641197205,
+      "learning_rate": 9.984593531431596e-05,
+      "loss": 0.6693,
+      "step": 1798
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.377590298652649,
+      "learning_rate": 9.971806812768777e-05,
+      "loss": 1.045,
+      "step": 1799
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.1018116474151611,
+      "learning_rate": 9.959022851852181e-05,
+      "loss": 0.9014,
+      "step": 1800
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.0155657529830933,
+      "learning_rate": 9.94624166262652e-05,
+      "loss": 0.8305,
+      "step": 1801
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.9664143323898315,
+      "learning_rate": 9.93346325903348e-05,
+      "loss": 0.9144,
+      "step": 1802
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.164772868156433,
+      "learning_rate": 9.92068765501171e-05,
+      "loss": 0.6573,
+      "step": 1803
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.186354160308838,
+      "learning_rate": 9.907914864496795e-05,
+      "loss": 0.795,
+      "step": 1804
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.0216922760009766,
+      "learning_rate": 9.895144901421276e-05,
+      "loss": 0.9515,
+      "step": 1805
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.3452105522155762,
+      "learning_rate": 9.882377779714581e-05,
+      "loss": 0.9363,
+      "step": 1806
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.1607165336608887,
+      "learning_rate": 9.869613513303061e-05,
+      "loss": 0.8842,
+      "step": 1807
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.6268951892852783,
+      "learning_rate": 9.856852116109934e-05,
+      "loss": 0.9655,
+      "step": 1808
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.1395448446273804,
+      "learning_rate": 9.844093602055302e-05,
+      "loss": 0.7502,
+      "step": 1809
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.886443018913269,
+      "learning_rate": 9.831337985056118e-05,
+      "loss": 0.5854,
+      "step": 1810
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.210634469985962,
+      "learning_rate": 9.818585279026172e-05,
+      "loss": 0.5795,
+      "step": 1811
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.1434961557388306,
+      "learning_rate": 9.805835497876081e-05,
+      "loss": 0.8192,
+      "step": 1812
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.2203586101531982,
+      "learning_rate": 9.793088655513271e-05,
+      "loss": 0.9838,
+      "step": 1813
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.5141627788543701,
+      "learning_rate": 9.780344765841966e-05,
+      "loss": 0.9707,
+      "step": 1814
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.3851544857025146,
+      "learning_rate": 9.767603842763163e-05,
+      "loss": 1.0544,
+      "step": 1815
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.5114184617996216,
+      "learning_rate": 9.754865900174626e-05,
+      "loss": 1.0471,
+      "step": 1816
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.1894009113311768,
+      "learning_rate": 9.742130951970872e-05,
+      "loss": 0.8344,
+      "step": 1817
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.153277039527893,
+      "learning_rate": 9.729399012043142e-05,
+      "loss": 0.8441,
+      "step": 1818
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.0140668153762817,
+      "learning_rate": 9.716670094279408e-05,
+      "loss": 0.5894,
+      "step": 1819
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.1728570461273193,
+      "learning_rate": 9.703944212564331e-05,
+      "loss": 0.9653,
+      "step": 1820
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.7548686265945435,
+      "learning_rate": 9.69122138077927e-05,
+      "loss": 1.028,
+      "step": 1821
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.3905855417251587,
+      "learning_rate": 9.678501612802263e-05,
+      "loss": 1.0491,
+      "step": 1822
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.0206429958343506,
+      "learning_rate": 9.665784922507986e-05,
+      "loss": 0.7978,
+      "step": 1823
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.292028546333313,
+      "learning_rate": 9.653071323767782e-05,
+      "loss": 1.0386,
+      "step": 1824
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.0786666870117188,
+      "learning_rate": 9.6403608304496e-05,
+      "loss": 0.9896,
+      "step": 1825
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.233822226524353,
+      "learning_rate": 9.627653456418015e-05,
+      "loss": 0.8662,
+      "step": 1826
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.1821707487106323,
+      "learning_rate": 9.6149492155342e-05,
+      "loss": 0.9316,
+      "step": 1827
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.213242769241333,
+      "learning_rate": 9.602248121655896e-05,
+      "loss": 0.8224,
+      "step": 1828
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.3918932676315308,
+      "learning_rate": 9.589550188637431e-05,
+      "loss": 1.3013,
+      "step": 1829
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.8927656412124634,
+      "learning_rate": 9.576855430329677e-05,
+      "loss": 0.4922,
+      "step": 1830
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.371404767036438,
+      "learning_rate": 9.564163860580034e-05,
+      "loss": 1.4036,
+      "step": 1831
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.2355639934539795,
+      "learning_rate": 9.551475493232434e-05,
+      "loss": 1.0875,
+      "step": 1832
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.2743234634399414,
+      "learning_rate": 9.538790342127317e-05,
+      "loss": 0.8392,
+      "step": 1833
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.3681198358535767,
+      "learning_rate": 9.526108421101608e-05,
+      "loss": 1.1149,
+      "step": 1834
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.9238698482513428,
+      "learning_rate": 9.513429743988715e-05,
+      "loss": 0.7203,
+      "step": 1835
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.1493499279022217,
+      "learning_rate": 9.5007543246185e-05,
+      "loss": 0.6815,
+      "step": 1836
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.1091593503952026,
+      "learning_rate": 9.488082176817276e-05,
+      "loss": 1.0553,
+      "step": 1837
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.177230954170227,
+      "learning_rate": 9.475413314407791e-05,
+      "loss": 1.1332,
+      "step": 1838
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.0102418661117554,
+      "learning_rate": 9.462747751209203e-05,
+      "loss": 0.6488,
+      "step": 1839
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.9865624308586121,
+      "learning_rate": 9.450085501037074e-05,
+      "loss": 0.7,
+      "step": 1840
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.4693363904953003,
+      "learning_rate": 9.437426577703352e-05,
+      "loss": 0.8631,
+      "step": 1841
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.2326639890670776,
+      "learning_rate": 9.424770995016355e-05,
+      "loss": 0.9311,
+      "step": 1842
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.074099063873291,
+      "learning_rate": 9.412118766780762e-05,
+      "loss": 0.9159,
+      "step": 1843
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.105237603187561,
+      "learning_rate": 9.399469906797584e-05,
+      "loss": 0.9116,
+      "step": 1844
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.9910746812820435,
+      "learning_rate": 9.386824428864169e-05,
+      "loss": 0.6469,
+      "step": 1845
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.4138925075531006,
+      "learning_rate": 9.37418234677416e-05,
+      "loss": 1.0355,
+      "step": 1846
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.0311132669448853,
+      "learning_rate": 9.361543674317517e-05,
+      "loss": 0.6162,
+      "step": 1847
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.1188533306121826,
+      "learning_rate": 9.348908425280462e-05,
+      "loss": 1.0292,
+      "step": 1848
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.9886927604675293,
+      "learning_rate": 9.33627661344549e-05,
+      "loss": 0.5008,
+      "step": 1849
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.3715204000473022,
+      "learning_rate": 9.323648252591351e-05,
+      "loss": 0.8163,
+      "step": 1850
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.8927873373031616,
+      "learning_rate": 9.311023356493021e-05,
+      "loss": 0.4377,
+      "step": 1851
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.2950105667114258,
+      "learning_rate": 9.298401938921708e-05,
+      "loss": 0.8877,
+      "step": 1852
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.3683269023895264,
+      "learning_rate": 9.285784013644817e-05,
+      "loss": 1.2326,
+      "step": 1853
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.0928535461425781,
+      "learning_rate": 9.273169594425951e-05,
+      "loss": 0.7462,
+      "step": 1854
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.2238028049468994,
+      "learning_rate": 9.260558695024877e-05,
+      "loss": 0.7335,
+      "step": 1855
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.110521674156189,
+      "learning_rate": 9.247951329197532e-05,
+      "loss": 0.7661,
+      "step": 1856
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.1721173524856567,
+      "learning_rate": 9.235347510695997e-05,
+      "loss": 0.9289,
+      "step": 1857
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.3499724864959717,
+      "learning_rate": 9.222747253268485e-05,
+      "loss": 1.1241,
+      "step": 1858
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.076082706451416,
+      "learning_rate": 9.210150570659317e-05,
+      "loss": 0.7515,
+      "step": 1859
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.311646819114685,
+      "learning_rate": 9.197557476608926e-05,
+      "loss": 0.9019,
+      "step": 1860
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.0273613929748535,
+      "learning_rate": 9.18496798485382e-05,
+      "loss": 0.8496,
+      "step": 1861
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.1790790557861328,
+      "learning_rate": 9.172382109126584e-05,
+      "loss": 0.8454,
+      "step": 1862
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.3008348941802979,
+      "learning_rate": 9.159799863155857e-05,
+      "loss": 1.0305,
+      "step": 1863
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.2865149974822998,
+      "learning_rate": 9.147221260666317e-05,
+      "loss": 0.7998,
+      "step": 1864
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.3089659214019775,
+      "learning_rate": 9.134646315378673e-05,
+      "loss": 0.7147,
+      "step": 1865
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.3149784803390503,
+      "learning_rate": 9.122075041009636e-05,
+      "loss": 0.8714,
+      "step": 1866
+    },
+    {
+      "epoch": 1.8,
+      "eval_loss": 1.2798618078231812,
+      "eval_runtime": 28.3827,
+      "eval_samples_per_second": 2.995,
+      "eval_steps_per_second": 1.515,
+      "step": 1866
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.132293462753296,
+      "learning_rate": 9.109507451271922e-05,
+      "loss": 0.7688,
+      "step": 1867
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.052720308303833,
+      "learning_rate": 9.096943559874222e-05,
+      "loss": 0.8002,
+      "step": 1868
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.0711005926132202,
+      "learning_rate": 9.084383380521194e-05,
+      "loss": 0.6079,
+      "step": 1869
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.3810640573501587,
+      "learning_rate": 9.071826926913446e-05,
+      "loss": 1.1552,
+      "step": 1870
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.2143968343734741,
+      "learning_rate": 9.059274212747525e-05,
+      "loss": 0.8241,
+      "step": 1871
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.5594843626022339,
+      "learning_rate": 9.046725251715898e-05,
+      "loss": 1.6285,
+      "step": 1872
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.9207326173782349,
+      "learning_rate": 9.03418005750694e-05,
+      "loss": 0.6927,
+      "step": 1873
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.8353189826011658,
+      "learning_rate": 9.021638643804907e-05,
+      "loss": 0.4263,
+      "step": 1874
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.267177939414978,
+      "learning_rate": 9.009101024289941e-05,
+      "loss": 0.948,
+      "step": 1875
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.1151788234710693,
+      "learning_rate": 8.996567212638047e-05,
+      "loss": 0.9628,
+      "step": 1876
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.1289645433425903,
+      "learning_rate": 8.984037222521074e-05,
+      "loss": 1.014,
+      "step": 1877
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.169303297996521,
+      "learning_rate": 8.971511067606696e-05,
+      "loss": 0.8052,
+      "step": 1878
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.068725347518921,
+      "learning_rate": 8.958988761558411e-05,
+      "loss": 0.734,
+      "step": 1879
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.1281707286834717,
+      "learning_rate": 8.946470318035514e-05,
+      "loss": 0.9301,
+      "step": 1880
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.137937068939209,
+      "learning_rate": 8.93395575069309e-05,
+      "loss": 0.638,
+      "step": 1881
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.3555268049240112,
+      "learning_rate": 8.921445073182e-05,
+      "loss": 0.9903,
+      "step": 1882
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.433186411857605,
+      "learning_rate": 8.908938299148847e-05,
+      "loss": 0.9861,
+      "step": 1883
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.9451619982719421,
+      "learning_rate": 8.896435442235995e-05,
+      "loss": 0.5024,
+      "step": 1884
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.4170985221862793,
+      "learning_rate": 8.883936516081521e-05,
+      "loss": 1.1437,
+      "step": 1885
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.1750236749649048,
+      "learning_rate": 8.871441534319223e-05,
+      "loss": 0.7278,
+      "step": 1886
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.9916924238204956,
+      "learning_rate": 8.858950510578588e-05,
+      "loss": 0.5394,
+      "step": 1887
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.1440494060516357,
+      "learning_rate": 8.846463458484794e-05,
+      "loss": 0.7198,
+      "step": 1888
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.0721584558486938,
+      "learning_rate": 8.833980391658685e-05,
+      "loss": 0.7725,
+      "step": 1889
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.2798032760620117,
+      "learning_rate": 8.821501323716751e-05,
+      "loss": 1.0844,
+      "step": 1890
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.939553439617157,
+      "learning_rate": 8.809026268271131e-05,
+      "loss": 0.6705,
+      "step": 1891
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.2908331155776978,
+      "learning_rate": 8.796555238929577e-05,
+      "loss": 1.0146,
+      "step": 1892
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.5145471096038818,
+      "learning_rate": 8.784088249295455e-05,
+      "loss": 1.2522,
+      "step": 1893
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.8970005512237549,
+      "learning_rate": 8.771625312967727e-05,
+      "loss": 0.6222,
+      "step": 1894
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.233276605606079,
+      "learning_rate": 8.759166443540923e-05,
+      "loss": 0.8866,
+      "step": 1895
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.0272518396377563,
+      "learning_rate": 8.746711654605152e-05,
+      "loss": 0.8453,
+      "step": 1896
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.503488302230835,
+      "learning_rate": 8.734260959746054e-05,
+      "loss": 1.0577,
+      "step": 1897
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.0901001691818237,
+      "learning_rate": 8.721814372544817e-05,
+      "loss": 0.8395,
+      "step": 1898
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.0978087186813354,
+      "learning_rate": 8.709371906578147e-05,
+      "loss": 0.598,
+      "step": 1899
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.0039117336273193,
+      "learning_rate": 8.696933575418254e-05,
+      "loss": 0.8613,
+      "step": 1900
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.0104575157165527,
+      "learning_rate": 8.684499392632831e-05,
+      "loss": 0.7842,
+      "step": 1901
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.2603685855865479,
+      "learning_rate": 8.672069371785052e-05,
+      "loss": 0.8023,
+      "step": 1902
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.1488606929779053,
+      "learning_rate": 8.659643526433547e-05,
+      "loss": 1.0167,
+      "step": 1903
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.049129843711853,
+      "learning_rate": 8.647221870132396e-05,
+      "loss": 0.7355,
+      "step": 1904
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.4575903415679932,
+      "learning_rate": 8.634804416431113e-05,
+      "loss": 0.9889,
+      "step": 1905
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.9238115549087524,
+      "learning_rate": 8.622391178874614e-05,
+      "loss": 0.5083,
+      "step": 1906
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.1920543909072876,
+      "learning_rate": 8.609982171003231e-05,
+      "loss": 0.9315,
+      "step": 1907
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 2.0908291339874268,
+      "learning_rate": 8.597577406352671e-05,
+      "loss": 0.9874,
+      "step": 1908
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.1943837404251099,
+      "learning_rate": 8.58517689845402e-05,
+      "loss": 0.8991,
+      "step": 1909
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.6717904806137085,
+      "learning_rate": 8.572780660833723e-05,
+      "loss": 1.4169,
+      "step": 1910
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.2120851278305054,
+      "learning_rate": 8.560388707013553e-05,
+      "loss": 0.8864,
+      "step": 1911
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.237656593322754,
+      "learning_rate": 8.548001050510624e-05,
+      "loss": 0.7664,
+      "step": 1912
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.891211986541748,
+      "learning_rate": 8.535617704837358e-05,
+      "loss": 0.4961,
+      "step": 1913
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.9243524074554443,
+      "learning_rate": 8.523238683501472e-05,
+      "loss": 1.4698,
+      "step": 1914
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.1669132709503174,
+      "learning_rate": 8.510864000005974e-05,
+      "loss": 0.858,
+      "step": 1915
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.3465845584869385,
+      "learning_rate": 8.49849366784913e-05,
+      "loss": 1.1071,
+      "step": 1916
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.277543306350708,
+      "learning_rate": 8.486127700524468e-05,
+      "loss": 0.9655,
+      "step": 1917
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.2550066709518433,
+      "learning_rate": 8.473766111520747e-05,
+      "loss": 0.8256,
+      "step": 1918
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.0959876775741577,
+      "learning_rate": 8.461408914321962e-05,
+      "loss": 0.8549,
+      "step": 1919
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.254042387008667,
+      "learning_rate": 8.4490561224073e-05,
+      "loss": 0.9881,
+      "step": 1920
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.036967396736145,
+      "learning_rate": 8.436707749251161e-05,
+      "loss": 0.8791,
+      "step": 1921
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.089248538017273,
+      "learning_rate": 8.424363808323107e-05,
+      "loss": 0.7721,
+      "step": 1922
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.3928107023239136,
+      "learning_rate": 8.41202431308789e-05,
+      "loss": 1.0117,
+      "step": 1923
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.0236631631851196,
+      "learning_rate": 8.399689277005386e-05,
+      "loss": 0.8051,
+      "step": 1924
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.1423760652542114,
+      "learning_rate": 8.387358713530621e-05,
+      "loss": 0.7716,
+      "step": 1925
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.547163724899292,
+      "learning_rate": 8.375032636113744e-05,
+      "loss": 0.8395,
+      "step": 1926
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.176893949508667,
+      "learning_rate": 8.362711058200001e-05,
+      "loss": 1.0601,
+      "step": 1927
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.064943552017212,
+      "learning_rate": 8.350393993229742e-05,
+      "loss": 0.8291,
+      "step": 1928
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.9675112962722778,
+      "learning_rate": 8.33808145463838e-05,
+      "loss": 0.6613,
+      "step": 1929
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.179880142211914,
+      "learning_rate": 8.325773455856406e-05,
+      "loss": 0.6113,
+      "step": 1930
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.3865076303482056,
+      "learning_rate": 8.313470010309345e-05,
+      "loss": 1.164,
+      "step": 1931
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.0845417976379395,
+      "learning_rate": 8.301171131417764e-05,
+      "loss": 0.8402,
+      "step": 1932
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.3481731414794922,
+      "learning_rate": 8.28887683259725e-05,
+      "loss": 1.0072,
+      "step": 1933
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.0890945196151733,
+      "learning_rate": 8.276587127258386e-05,
+      "loss": 0.715,
+      "step": 1934
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.2235881090164185,
+      "learning_rate": 8.264302028806755e-05,
+      "loss": 0.7171,
+      "step": 1935
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.18882155418396,
+      "learning_rate": 8.252021550642899e-05,
+      "loss": 0.8737,
+      "step": 1936
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.4608304500579834,
+      "learning_rate": 8.239745706162336e-05,
+      "loss": 1.1279,
+      "step": 1937
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.1702600717544556,
+      "learning_rate": 8.227474508755529e-05,
+      "loss": 0.8266,
+      "step": 1938
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.3556264638900757,
+      "learning_rate": 8.215207971807856e-05,
+      "loss": 0.9647,
+      "step": 1939
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.1378785371780396,
+      "learning_rate": 8.20294610869963e-05,
+      "loss": 0.734,
+      "step": 1940
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.2239617109298706,
+      "learning_rate": 8.19068893280605e-05,
+      "loss": 0.9285,
+      "step": 1941
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.1962335109710693,
+      "learning_rate": 8.178436457497218e-05,
+      "loss": 0.9248,
+      "step": 1942
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.020392656326294,
+      "learning_rate": 8.1661886961381e-05,
+      "loss": 0.8781,
+      "step": 1943
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.4522514343261719,
+      "learning_rate": 8.153945662088514e-05,
+      "loss": 1.4785,
+      "step": 1944
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.1410571336746216,
+      "learning_rate": 8.14170736870313e-05,
+      "loss": 1.0401,
+      "step": 1945
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.2002620697021484,
+      "learning_rate": 8.129473829331452e-05,
+      "loss": 0.9184,
+      "step": 1946
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.2167078256607056,
+      "learning_rate": 8.117245057317785e-05,
+      "loss": 0.8945,
+      "step": 1947
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.2816885709762573,
+      "learning_rate": 8.10502106600124e-05,
+      "loss": 0.9105,
+      "step": 1948
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.3111622333526611,
+      "learning_rate": 8.092801868715719e-05,
+      "loss": 0.8802,
+      "step": 1949
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 2.199481964111328,
+      "learning_rate": 8.080587478789881e-05,
+      "loss": 1.4535,
+      "step": 1950
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.4825800657272339,
+      "learning_rate": 8.068377909547157e-05,
+      "loss": 1.0266,
+      "step": 1951
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.1725435256958008,
+      "learning_rate": 8.056173174305706e-05,
+      "loss": 1.0433,
+      "step": 1952
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.0390863418579102,
+      "learning_rate": 8.043973286378419e-05,
+      "loss": 0.6835,
+      "step": 1953
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.0793219804763794,
+      "learning_rate": 8.031778259072909e-05,
+      "loss": 0.8351,
+      "step": 1954
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.5510258674621582,
+      "learning_rate": 8.01958810569147e-05,
+      "loss": 1.6831,
+      "step": 1955
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.1949468851089478,
+      "learning_rate": 8.007402839531092e-05,
+      "loss": 0.8632,
+      "step": 1956
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.6904135942459106,
+      "learning_rate": 7.995222473883426e-05,
+      "loss": 1.221,
+      "step": 1957
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.4048141241073608,
+      "learning_rate": 7.983047022034785e-05,
+      "loss": 1.3236,
+      "step": 1958
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.9245966672897339,
+      "learning_rate": 7.97087649726612e-05,
+      "loss": 0.7279,
+      "step": 1959
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.9320887327194214,
+      "learning_rate": 7.958710912853003e-05,
+      "loss": 0.6237,
+      "step": 1960
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.9442594647407532,
+      "learning_rate": 7.946550282065623e-05,
+      "loss": 0.5785,
+      "step": 1961
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.9472630620002747,
+      "learning_rate": 7.93439461816876e-05,
+      "loss": 0.7583,
+      "step": 1962
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.3176424503326416,
+      "learning_rate": 7.922243934421783e-05,
+      "loss": 1.0933,
+      "step": 1963
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.2619996070861816,
+      "learning_rate": 7.910098244078618e-05,
+      "loss": 1.1577,
+      "step": 1964
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.3114757537841797,
+      "learning_rate": 7.897957560387755e-05,
+      "loss": 1.1675,
+      "step": 1965
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.9982705116271973,
+      "learning_rate": 7.885821896592221e-05,
+      "loss": 1.0683,
+      "step": 1966
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.1973057985305786,
+      "learning_rate": 7.873691265929562e-05,
+      "loss": 0.7786,
+      "step": 1967
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.3525385856628418,
+      "learning_rate": 7.861565681631838e-05,
+      "loss": 0.9712,
+      "step": 1968
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.4053092002868652,
+      "learning_rate": 7.849445156925594e-05,
+      "loss": 0.9474,
+      "step": 1969
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.096291184425354,
+      "learning_rate": 7.837329705031882e-05,
+      "loss": 0.6538,
+      "step": 1970
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.0880426168441772,
+      "learning_rate": 7.825219339166193e-05,
+      "loss": 0.7716,
+      "step": 1971
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.013417363166809,
+      "learning_rate": 7.813114072538488e-05,
+      "loss": 0.7396,
+      "step": 1972
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.8553948402404785,
+      "learning_rate": 7.801013918353149e-05,
+      "loss": 0.412,
+      "step": 1973
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.061767578125,
+      "learning_rate": 7.788918889809e-05,
+      "loss": 0.6492,
+      "step": 1974
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.1720582246780396,
+      "learning_rate": 7.77682900009926e-05,
+      "loss": 0.9722,
+      "step": 1975
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.6846498250961304,
+      "learning_rate": 7.764744262411548e-05,
+      "loss": 1.0323,
+      "step": 1976
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.277524709701538,
+      "learning_rate": 7.752664689927868e-05,
+      "loss": 0.9301,
+      "step": 1977
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.1959736347198486,
+      "learning_rate": 7.740590295824578e-05,
+      "loss": 0.9538,
+      "step": 1978
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.221312403678894,
+      "learning_rate": 7.728521093272398e-05,
+      "loss": 0.8935,
+      "step": 1979
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.4521623849868774,
+      "learning_rate": 7.716457095436378e-05,
+      "loss": 1.4089,
+      "step": 1980
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.021066665649414,
+      "learning_rate": 7.704398315475897e-05,
+      "loss": 0.6881,
+      "step": 1981
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.46727454662323,
+      "learning_rate": 7.692344766544642e-05,
+      "loss": 0.9396,
+      "step": 1982
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.8718202114105225,
+      "learning_rate": 7.680296461790583e-05,
+      "loss": 0.673,
+      "step": 1983
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.5483649969100952,
+      "learning_rate": 7.668253414355986e-05,
+      "loss": 1.0771,
+      "step": 1984
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.7343392968177795,
+      "learning_rate": 7.656215637377367e-05,
+      "loss": 0.5231,
+      "step": 1985
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.5243200063705444,
+      "learning_rate": 7.644183143985504e-05,
+      "loss": 1.1263,
+      "step": 1986
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.0501224994659424,
+      "learning_rate": 7.632155947305411e-05,
+      "loss": 0.7451,
+      "step": 1987
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.9759675860404968,
+      "learning_rate": 7.620134060456316e-05,
+      "loss": 0.5377,
+      "step": 1988
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.44871187210083,
+      "learning_rate": 7.608117496551665e-05,
+      "loss": 1.0826,
+      "step": 1989
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.0177377462387085,
+      "learning_rate": 7.596106268699088e-05,
+      "loss": 0.8488,
+      "step": 1990
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.3080190420150757,
+      "learning_rate": 7.584100390000405e-05,
+      "loss": 0.8976,
+      "step": 1991
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.238023042678833,
+      "learning_rate": 7.572099873551585e-05,
+      "loss": 0.863,
+      "step": 1992
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.9784834980964661,
+      "learning_rate": 7.560104732442774e-05,
+      "loss": 0.5743,
+      "step": 1993
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.2834821939468384,
+      "learning_rate": 7.54811497975823e-05,
+      "loss": 1.0481,
+      "step": 1994
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.9290242195129395,
+      "learning_rate": 7.536130628576346e-05,
+      "loss": 0.5789,
+      "step": 1995
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.137740969657898,
+      "learning_rate": 7.524151691969615e-05,
+      "loss": 0.74,
+      "step": 1996
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.9530397653579712,
+      "learning_rate": 7.512178183004632e-05,
+      "loss": 0.3774,
+      "step": 1997
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.8863974809646606,
+      "learning_rate": 7.500210114742068e-05,
+      "loss": 0.7082,
+      "step": 1998
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.524322748184204,
+      "learning_rate": 7.488247500236656e-05,
+      "loss": 0.891,
+      "step": 1999
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.9274705648422241,
+      "learning_rate": 7.476290352537188e-05,
+      "loss": 0.5485,
+      "step": 2000
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.0648436546325684,
+      "learning_rate": 7.464338684686481e-05,
+      "loss": 0.6977,
+      "step": 2001
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.333897590637207,
+      "learning_rate": 7.452392509721391e-05,
+      "loss": 0.7772,
+      "step": 2002
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.1420466899871826,
+      "learning_rate": 7.440451840672766e-05,
+      "loss": 0.8032,
+      "step": 2003
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.3571137189865112,
+      "learning_rate": 7.428516690565455e-05,
+      "loss": 0.8893,
+      "step": 2004
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.165768027305603,
+      "learning_rate": 7.416587072418293e-05,
+      "loss": 0.647,
+      "step": 2005
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.4117252826690674,
+      "learning_rate": 7.404662999244069e-05,
+      "loss": 0.9203,
+      "step": 2006
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.206149935722351,
+      "learning_rate": 7.392744484049533e-05,
+      "loss": 0.8442,
+      "step": 2007
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.2842216491699219,
+      "learning_rate": 7.380831539835367e-05,
+      "loss": 1.046,
+      "step": 2008
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.2140345573425293,
+      "learning_rate": 7.368924179596177e-05,
+      "loss": 0.5501,
+      "step": 2009
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.0220657587051392,
+      "learning_rate": 7.357022416320485e-05,
+      "loss": 0.7256,
+      "step": 2010
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.4720944166183472,
+      "learning_rate": 7.345126262990693e-05,
+      "loss": 0.977,
+      "step": 2011
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.208706021308899,
+      "learning_rate": 7.3332357325831e-05,
+      "loss": 0.7604,
+      "step": 2012
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.3358442783355713,
+      "learning_rate": 7.32135083806786e-05,
+      "loss": 1.0311,
+      "step": 2013
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.1216422319412231,
+      "learning_rate": 7.309471592408984e-05,
+      "loss": 0.5732,
+      "step": 2014
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.0371495485305786,
+      "learning_rate": 7.297598008564324e-05,
+      "loss": 0.5215,
+      "step": 2015
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 2.6368014812469482,
+      "learning_rate": 7.285730099485543e-05,
+      "loss": 1.1245,
+      "step": 2016
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.0725258588790894,
+      "learning_rate": 7.273867878118139e-05,
+      "loss": 0.5464,
+      "step": 2017
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.368296504020691,
+      "learning_rate": 7.262011357401377e-05,
+      "loss": 1.0568,
+      "step": 2018
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.1201132535934448,
+      "learning_rate": 7.250160550268329e-05,
+      "loss": 0.8696,
+      "step": 2019
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.248964548110962,
+      "learning_rate": 7.238315469645815e-05,
+      "loss": 0.8216,
+      "step": 2020
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.132161259651184,
+      "learning_rate": 7.226476128454415e-05,
+      "loss": 0.6487,
+      "step": 2021
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.1532809734344482,
+      "learning_rate": 7.214642539608457e-05,
+      "loss": 0.7051,
+      "step": 2022
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.2578155994415283,
+      "learning_rate": 7.202814716015987e-05,
+      "loss": 0.8329,
+      "step": 2023
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.1598081588745117,
+      "learning_rate": 7.190992670578758e-05,
+      "loss": 0.8972,
+      "step": 2024
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.4254798889160156,
+      "learning_rate": 7.179176416192223e-05,
+      "loss": 1.0065,
+      "step": 2025
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.1195855140686035,
+      "learning_rate": 7.167365965745529e-05,
+      "loss": 0.6957,
+      "step": 2026
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.5119630098342896,
+      "learning_rate": 7.155561332121477e-05,
+      "loss": 1.0989,
+      "step": 2027
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.9728872776031494,
+      "learning_rate": 7.143762528196524e-05,
+      "loss": 0.696,
+      "step": 2028
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.5496985912322998,
+      "learning_rate": 7.131969566840788e-05,
+      "loss": 0.9291,
+      "step": 2029
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.1622270345687866,
+      "learning_rate": 7.120182460917987e-05,
+      "loss": 0.6708,
+      "step": 2030
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.5910062789916992,
+      "learning_rate": 7.108401223285463e-05,
+      "loss": 1.271,
+      "step": 2031
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.4135732650756836,
+      "learning_rate": 7.096625866794165e-05,
+      "loss": 0.7867,
+      "step": 2032
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.1207356452941895,
+      "learning_rate": 7.084856404288619e-05,
+      "loss": 0.8527,
+      "step": 2033
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.189270257949829,
+      "learning_rate": 7.07309284860692e-05,
+      "loss": 1.0127,
+      "step": 2034
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.9003916382789612,
+      "learning_rate": 7.061335212580717e-05,
+      "loss": 0.6056,
+      "step": 2035
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.007911205291748,
+      "learning_rate": 7.049583509035218e-05,
+      "loss": 0.8295,
+      "step": 2036
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.3060964345932007,
+      "learning_rate": 7.037837750789144e-05,
+      "loss": 0.7436,
+      "step": 2037
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.2931039333343506,
+      "learning_rate": 7.026097950654733e-05,
+      "loss": 0.8215,
+      "step": 2038
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.0952670574188232,
+      "learning_rate": 7.014364121437727e-05,
+      "loss": 0.5497,
+      "step": 2039
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.260249376296997,
+      "learning_rate": 7.002636275937366e-05,
+      "loss": 1.1938,
+      "step": 2040
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.329917311668396,
+      "learning_rate": 6.990914426946341e-05,
+      "loss": 0.9008,
+      "step": 2041
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.9461268782615662,
+      "learning_rate": 6.979198587250812e-05,
+      "loss": 0.6912,
+      "step": 2042
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.3266767263412476,
+      "learning_rate": 6.967488769630394e-05,
+      "loss": 0.6323,
+      "step": 2043
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.455551028251648,
+      "learning_rate": 6.955784986858117e-05,
+      "loss": 1.2073,
+      "step": 2044
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.1189351081848145,
+      "learning_rate": 6.944087251700437e-05,
+      "loss": 0.8706,
+      "step": 2045
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.1993297338485718,
+      "learning_rate": 6.932395576917209e-05,
+      "loss": 0.8263,
+      "step": 2046
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.2151789665222168,
+      "learning_rate": 6.920709975261686e-05,
+      "loss": 1.04,
+      "step": 2047
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.1951751708984375,
+      "learning_rate": 6.909030459480487e-05,
+      "loss": 0.804,
+      "step": 2048
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.9988811612129211,
+      "learning_rate": 6.89735704231359e-05,
+      "loss": 0.7224,
+      "step": 2049
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.8926522731781006,
+      "learning_rate": 6.885689736494337e-05,
+      "loss": 0.4905,
+      "step": 2050
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.233476996421814,
+      "learning_rate": 6.87402855474939e-05,
+      "loss": 1.0149,
+      "step": 2051
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.099674940109253,
+      "learning_rate": 6.862373509798725e-05,
+      "loss": 0.8594,
+      "step": 2052
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.9110369682312012,
+      "learning_rate": 6.850724614355649e-05,
+      "loss": 0.6211,
+      "step": 2053
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.037454605102539,
+      "learning_rate": 6.839081881126735e-05,
+      "loss": 0.6983,
+      "step": 2054
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.5326181650161743,
+      "learning_rate": 6.827445322811849e-05,
+      "loss": 1.3385,
+      "step": 2055
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.00337815284729,
+      "learning_rate": 6.81581495210411e-05,
+      "loss": 0.7778,
+      "step": 2056
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.9707116484642029,
+      "learning_rate": 6.804190781689904e-05,
+      "loss": 0.493,
+      "step": 2057
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.5447719097137451,
+      "learning_rate": 6.792572824248839e-05,
+      "loss": 1.3885,
+      "step": 2058
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.3587324619293213,
+      "learning_rate": 6.78096109245375e-05,
+      "loss": 1.0177,
+      "step": 2059
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.4563665390014648,
+      "learning_rate": 6.769355598970686e-05,
+      "loss": 1.0932,
+      "step": 2060
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.1776785850524902,
+      "learning_rate": 6.757756356458888e-05,
+      "loss": 1.0123,
+      "step": 2061
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.409449577331543,
+      "learning_rate": 6.746163377570777e-05,
+      "loss": 0.9802,
+      "step": 2062
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.5689934492111206,
+      "learning_rate": 6.73457667495193e-05,
+      "loss": 0.939,
+      "step": 2063
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.2428174018859863,
+      "learning_rate": 6.722996261241116e-05,
+      "loss": 1.004,
+      "step": 2064
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.0462515354156494,
+      "learning_rate": 6.711422149070201e-05,
+      "loss": 0.809,
+      "step": 2065
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.9034873843193054,
+      "learning_rate": 6.699854351064201e-05,
+      "loss": 0.6096,
+      "step": 2066
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.0596833229064941,
+      "learning_rate": 6.688292879841231e-05,
+      "loss": 0.7736,
+      "step": 2067
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.184257984161377,
+      "learning_rate": 6.676737748012523e-05,
+      "loss": 0.9299,
+      "step": 2068
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.2924439907073975,
+      "learning_rate": 6.665188968182379e-05,
+      "loss": 1.0715,
+      "step": 2069
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.367877721786499,
+      "learning_rate": 6.65364655294817e-05,
+      "loss": 1.1742,
+      "step": 2070
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.4632872343063354,
+      "learning_rate": 6.642110514900342e-05,
+      "loss": 1.1471,
+      "step": 2071
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.3757456541061401,
+      "learning_rate": 6.630580866622369e-05,
+      "loss": 1.0932,
+      "step": 2072
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.8886003494262695,
+      "learning_rate": 6.619057620690756e-05,
+      "loss": 0.6071,
+      "step": 2073
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.5674740672111511,
+      "learning_rate": 6.607540789675029e-05,
+      "loss": 0.2693,
+      "step": 2074
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.8528835773468018,
+      "learning_rate": 6.596030386137723e-05,
+      "loss": 0.4279,
+      "step": 2075
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.6943017244338989,
+      "learning_rate": 6.58452642263435e-05,
+      "loss": 0.3272,
+      "step": 2076
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.0943104028701782,
+      "learning_rate": 6.573028911713397e-05,
+      "loss": 0.7113,
+      "step": 2077
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.8859809041023254,
+      "learning_rate": 6.561537865916328e-05,
+      "loss": 0.7564,
+      "step": 2078
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.8713234066963196,
+      "learning_rate": 6.550053297777538e-05,
+      "loss": 0.4953,
+      "step": 2079
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.8634008765220642,
+      "learning_rate": 6.538575219824358e-05,
+      "loss": 0.4017,
+      "step": 2080
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.8085753321647644,
+      "learning_rate": 6.527103644577056e-05,
+      "loss": 0.395,
+      "step": 2081
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.8304145336151123,
+      "learning_rate": 6.515638584548787e-05,
+      "loss": 0.541,
+      "step": 2082
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.8201612830162048,
+      "learning_rate": 6.504180052245609e-05,
+      "loss": 0.5326,
+      "step": 2083
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.7562471032142639,
+      "learning_rate": 6.492728060166453e-05,
+      "loss": 0.4054,
+      "step": 2084
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.9142252206802368,
+      "learning_rate": 6.48128262080313e-05,
+      "loss": 0.5096,
+      "step": 2085
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.7330839037895203,
+      "learning_rate": 6.469843746640286e-05,
+      "loss": 0.4187,
+      "step": 2086
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.9274792075157166,
+      "learning_rate": 6.458411450155418e-05,
+      "loss": 0.5977,
+      "step": 2087
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.8541355729103088,
+      "learning_rate": 6.446985743818841e-05,
+      "loss": 0.4998,
+      "step": 2088
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.8299204111099243,
+      "learning_rate": 6.435566640093689e-05,
+      "loss": 0.4044,
+      "step": 2089
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.0775256156921387,
+      "learning_rate": 6.424154151435886e-05,
+      "loss": 0.4907,
+      "step": 2090
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.031470775604248,
+      "learning_rate": 6.412748290294141e-05,
+      "loss": 0.5825,
+      "step": 2091
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.0457535982131958,
+      "learning_rate": 6.40134906910994e-05,
+      "loss": 0.5561,
+      "step": 2092
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.9657010436058044,
+      "learning_rate": 6.389956500317523e-05,
+      "loss": 0.3679,
+      "step": 2093
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.9064496755599976,
+      "learning_rate": 6.378570596343871e-05,
+      "loss": 0.4382,
+      "step": 2094
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.033677577972412,
+      "learning_rate": 6.367191369608691e-05,
+      "loss": 0.4445,
+      "step": 2095
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.8130964040756226,
+      "learning_rate": 6.355818832524424e-05,
+      "loss": 0.2655,
+      "step": 2096
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.8505420088768005,
+      "learning_rate": 6.344452997496196e-05,
+      "loss": 0.3172,
+      "step": 2097
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.4362200498580933,
+      "learning_rate": 6.333093876921825e-05,
+      "loss": 0.5314,
+      "step": 2098
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.2488155364990234,
+      "learning_rate": 6.32174148319182e-05,
+      "loss": 0.4211,
+      "step": 2099
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.3689916133880615,
+      "learning_rate": 6.310395828689331e-05,
+      "loss": 0.6113,
+      "step": 2100
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.188043236732483,
+      "learning_rate": 6.299056925790173e-05,
+      "loss": 0.459,
+      "step": 2101
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.1546956300735474,
+      "learning_rate": 6.287724786862782e-05,
+      "loss": 0.4895,
+      "step": 2102
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.079819679260254,
+      "learning_rate": 6.276399424268233e-05,
+      "loss": 0.3741,
+      "step": 2103
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.0333255529403687,
+      "learning_rate": 6.265080850360199e-05,
+      "loss": 0.3441,
+      "step": 2104
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.281787395477295,
+      "learning_rate": 6.253769077484945e-05,
+      "loss": 0.4487,
+      "step": 2105
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.5447965860366821,
+      "learning_rate": 6.242464117981327e-05,
+      "loss": 0.7834,
+      "step": 2106
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.9776402711868286,
+      "learning_rate": 6.231165984180762e-05,
+      "loss": 0.4425,
+      "step": 2107
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.036454677581787,
+      "learning_rate": 6.219874688407221e-05,
+      "loss": 0.4159,
+      "step": 2108
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.1195133924484253,
+      "learning_rate": 6.208590242977223e-05,
+      "loss": 0.3327,
+      "step": 2109
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.0579090118408203,
+      "learning_rate": 6.197312660199805e-05,
+      "loss": 0.5053,
+      "step": 2110
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.0781733989715576,
+      "learning_rate": 6.186041952376532e-05,
+      "loss": 0.3166,
+      "step": 2111
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.1364599466323853,
+      "learning_rate": 6.174778131801454e-05,
+      "loss": 0.4686,
+      "step": 2112
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.6187351942062378,
+      "learning_rate": 6.163521210761114e-05,
+      "loss": 0.8032,
+      "step": 2113
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.0550283193588257,
+      "learning_rate": 6.152271201534539e-05,
+      "loss": 0.4735,
+      "step": 2114
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.0076221227645874,
+      "learning_rate": 6.141028116393204e-05,
+      "loss": 0.5,
+      "step": 2115
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.0741852521896362,
+      "learning_rate": 6.129791967601031e-05,
+      "loss": 0.5462,
+      "step": 2116
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.0995590686798096,
+      "learning_rate": 6.11856276741439e-05,
+      "loss": 0.3663,
+      "step": 2117
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.0732592344284058,
+      "learning_rate": 6.107340528082054e-05,
+      "loss": 0.4206,
+      "step": 2118
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.1388449668884277,
+      "learning_rate": 6.096125261845212e-05,
+      "loss": 0.4032,
+      "step": 2119
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.468422770500183,
+      "learning_rate": 6.084916980937451e-05,
+      "loss": 0.3697,
+      "step": 2120
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.2283095121383667,
+      "learning_rate": 6.07371569758473e-05,
+      "loss": 0.7188,
+      "step": 2121
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.4419937133789062,
+      "learning_rate": 6.062521424005378e-05,
+      "loss": 0.7527,
+      "step": 2122
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.368731141090393,
+      "learning_rate": 6.051334172410074e-05,
+      "loss": 0.4648,
+      "step": 2123
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.0447849035263062,
+      "learning_rate": 6.0401539550018494e-05,
+      "loss": 0.3418,
+      "step": 2124
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.1050643920898438,
+      "learning_rate": 6.028980783976051e-05,
+      "loss": 0.4347,
+      "step": 2125
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.1442632675170898,
+      "learning_rate": 6.017814671520339e-05,
+      "loss": 0.4525,
+      "step": 2126
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.100126028060913,
+      "learning_rate": 6.006655629814686e-05,
+      "loss": 0.3395,
+      "step": 2127
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.2252593040466309,
+      "learning_rate": 5.995503671031342e-05,
+      "loss": 0.6215,
+      "step": 2128
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.6980013251304626,
+      "learning_rate": 5.9843588073348315e-05,
+      "loss": 0.2842,
+      "step": 2129
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.2332143783569336,
+      "learning_rate": 5.9732210508819384e-05,
+      "loss": 0.4217,
+      "step": 2130
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.9792746305465698,
+      "learning_rate": 5.962090413821704e-05,
+      "loss": 0.4573,
+      "step": 2131
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.107351541519165,
+      "learning_rate": 5.950966908295394e-05,
+      "loss": 0.3899,
+      "step": 2132
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.6578118205070496,
+      "learning_rate": 5.9398505464364926e-05,
+      "loss": 0.2116,
+      "step": 2133
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.0342390537261963,
+      "learning_rate": 5.928741340370701e-05,
+      "loss": 0.3738,
+      "step": 2134
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.2607660293579102,
+      "learning_rate": 5.9176393022159174e-05,
+      "loss": 0.4923,
+      "step": 2135
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.0737662315368652,
+      "learning_rate": 5.906544444082207e-05,
+      "loss": 0.4356,
+      "step": 2136
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.0083526372909546,
+      "learning_rate": 5.895456778071809e-05,
+      "loss": 0.3503,
+      "step": 2137
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.2637425661087036,
+      "learning_rate": 5.884376316279126e-05,
+      "loss": 0.6458,
+      "step": 2138
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.065369725227356,
+      "learning_rate": 5.8733030707906936e-05,
+      "loss": 0.5274,
+      "step": 2139
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.150827169418335,
+      "learning_rate": 5.862237053685175e-05,
+      "loss": 0.3604,
+      "step": 2140
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.1921899318695068,
+      "learning_rate": 5.851178277033349e-05,
+      "loss": 0.5229,
+      "step": 2141
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.0592162609100342,
+      "learning_rate": 5.840126752898106e-05,
+      "loss": 0.4295,
+      "step": 2142
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.2355237007141113,
+      "learning_rate": 5.8290824933344175e-05,
+      "loss": 0.7435,
+      "step": 2143
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.973853349685669,
+      "learning_rate": 5.818045510389322e-05,
+      "loss": 0.2713,
+      "step": 2144
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.0709655284881592,
+      "learning_rate": 5.807015816101943e-05,
+      "loss": 0.2912,
+      "step": 2145
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.268304705619812,
+      "learning_rate": 5.795993422503436e-05,
+      "loss": 0.5308,
+      "step": 2146
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.886254608631134,
+      "learning_rate": 5.784978341616992e-05,
+      "loss": 0.2374,
+      "step": 2147
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.9648481011390686,
+      "learning_rate": 5.7739705854578424e-05,
+      "loss": 0.4132,
+      "step": 2148
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.0699951648712158,
+      "learning_rate": 5.7629701660332104e-05,
+      "loss": 0.418,
+      "step": 2149
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.2619209289550781,
+      "learning_rate": 5.7519770953423256e-05,
+      "loss": 0.5038,
+      "step": 2150
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.9739295840263367,
+      "learning_rate": 5.740991385376393e-05,
+      "loss": 0.5273,
+      "step": 2151
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.7552351951599121,
+      "learning_rate": 5.7300130481186054e-05,
+      "loss": 0.2951,
+      "step": 2152
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.2434828281402588,
+      "learning_rate": 5.7190420955440996e-05,
+      "loss": 0.581,
+      "step": 2153
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.1353987455368042,
+      "learning_rate": 5.708078539619954e-05,
+      "loss": 0.4426,
+      "step": 2154
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.0610759258270264,
+      "learning_rate": 5.697122392305194e-05,
+      "loss": 0.4811,
+      "step": 2155
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 2.2417352199554443,
+      "learning_rate": 5.6861736655507546e-05,
+      "loss": 0.3413,
+      "step": 2156
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.0961496829986572,
+      "learning_rate": 5.6752323712994684e-05,
+      "loss": 0.3918,
+      "step": 2157
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.965447187423706,
+      "learning_rate": 5.664298521486076e-05,
+      "loss": 0.3328,
+      "step": 2158
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.3192046880722046,
+      "learning_rate": 5.6533721280371945e-05,
+      "loss": 0.5889,
+      "step": 2159
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.0044875144958496,
+      "learning_rate": 5.6424532028713e-05,
+      "loss": 0.3278,
+      "step": 2160
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.924491286277771,
+      "learning_rate": 5.631541757898727e-05,
+      "loss": 0.2975,
+      "step": 2161
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.2042051553726196,
+      "learning_rate": 5.6206378050216434e-05,
+      "loss": 0.5941,
+      "step": 2162
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.0397312641143799,
+      "learning_rate": 5.609741356134061e-05,
+      "loss": 0.312,
+      "step": 2163
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.298201322555542,
+      "learning_rate": 5.598852423121793e-05,
+      "loss": 0.5786,
+      "step": 2164
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.0598686933517456,
+      "learning_rate": 5.587971017862453e-05,
+      "loss": 0.2725,
+      "step": 2165
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.0505437850952148,
+      "learning_rate": 5.577097152225456e-05,
+      "loss": 0.3511,
+      "step": 2166
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.3280149698257446,
+      "learning_rate": 5.566230838071983e-05,
+      "loss": 0.6197,
+      "step": 2167
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.2126686573028564,
+      "learning_rate": 5.555372087254976e-05,
+      "loss": 0.4556,
+      "step": 2168
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.9505837559700012,
+      "learning_rate": 5.544520911619128e-05,
+      "loss": 0.3118,
+      "step": 2169
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.4259228706359863,
+      "learning_rate": 5.533677323000882e-05,
+      "loss": 0.5703,
+      "step": 2170
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.0361673831939697,
+      "learning_rate": 5.522841333228389e-05,
+      "loss": 0.3863,
+      "step": 2171
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.2724729776382446,
+      "learning_rate": 5.512012954121513e-05,
+      "loss": 0.6094,
+      "step": 2172
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.155977725982666,
+      "learning_rate": 5.501192197491829e-05,
+      "loss": 0.3416,
+      "step": 2173
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 2.008774757385254,
+      "learning_rate": 5.490379075142586e-05,
+      "loss": 0.8598,
+      "step": 2174
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.293354868888855,
+      "learning_rate": 5.479573598868701e-05,
+      "loss": 0.5247,
+      "step": 2175
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.1473639011383057,
+      "learning_rate": 5.468775780456768e-05,
+      "loss": 0.5186,
+      "step": 2176
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.0745112895965576,
+      "learning_rate": 5.457985631685015e-05,
+      "loss": 0.5412,
+      "step": 2177
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.0768483877182007,
+      "learning_rate": 5.4472031643233054e-05,
+      "loss": 0.6722,
+      "step": 2178
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.7210749387741089,
+      "learning_rate": 5.4364283901331206e-05,
+      "loss": 0.9224,
+      "step": 2179
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.2692290544509888,
+      "learning_rate": 5.4256613208675586e-05,
+      "loss": 0.6347,
+      "step": 2180
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.937003493309021,
+      "learning_rate": 5.4149019682713155e-05,
+      "loss": 0.3242,
+      "step": 2181
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.2912002801895142,
+      "learning_rate": 5.4041503440806594e-05,
+      "loss": 0.4896,
+      "step": 2182
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.1830394268035889,
+      "learning_rate": 5.3934064600234256e-05,
+      "loss": 0.5103,
+      "step": 2183
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.0488359928131104,
+      "learning_rate": 5.382670327819024e-05,
+      "loss": 0.3721,
+      "step": 2184
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.9774886965751648,
+      "learning_rate": 5.371941959178393e-05,
+      "loss": 0.3123,
+      "step": 2185
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.9702259302139282,
+      "learning_rate": 5.361221365804006e-05,
+      "loss": 0.3097,
+      "step": 2186
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.9657133221626282,
+      "learning_rate": 5.350508559389863e-05,
+      "loss": 0.3055,
+      "step": 2187
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.1606115102767944,
+      "learning_rate": 5.339803551621461e-05,
+      "loss": 0.474,
+      "step": 2188
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.0346829891204834,
+      "learning_rate": 5.329106354175792e-05,
+      "loss": 0.4327,
+      "step": 2189
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.5432517528533936,
+      "learning_rate": 5.318416978721327e-05,
+      "loss": 0.7782,
+      "step": 2190
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.9665737152099609,
+      "learning_rate": 5.307735436918015e-05,
+      "loss": 0.3279,
+      "step": 2191
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.0604126453399658,
+      "learning_rate": 5.297061740417249e-05,
+      "loss": 0.3749,
+      "step": 2192
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.3815175294876099,
+      "learning_rate": 5.286395900861862e-05,
+      "loss": 0.6345,
+      "step": 2193
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.2199488878250122,
+      "learning_rate": 5.2757379298861345e-05,
+      "loss": 0.3725,
+      "step": 2194
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.2785658836364746,
+      "learning_rate": 5.265087839115745e-05,
+      "loss": 0.5089,
+      "step": 2195
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.1950833797454834,
+      "learning_rate": 5.254445640167781e-05,
+      "loss": 0.4159,
+      "step": 2196
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.044345736503601,
+      "learning_rate": 5.243811344650731e-05,
+      "loss": 0.3404,
+      "step": 2197
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.1644901037216187,
+      "learning_rate": 5.233184964164453e-05,
+      "loss": 0.4536,
+      "step": 2198
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.763175368309021,
+      "learning_rate": 5.222566510300172e-05,
+      "loss": 0.8698,
+      "step": 2199
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.374753713607788,
+      "learning_rate": 5.211955994640466e-05,
+      "loss": 0.5751,
+      "step": 2200
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.0620756149291992,
+      "learning_rate": 5.2013534287592635e-05,
+      "loss": 0.4053,
+      "step": 2201
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.3876912593841553,
+      "learning_rate": 5.1907588242218095e-05,
+      "loss": 0.5241,
+      "step": 2202
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.2697367668151855,
+      "learning_rate": 5.1801721925846686e-05,
+      "loss": 0.4191,
+      "step": 2203
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.1614696979522705,
+      "learning_rate": 5.1695935453957106e-05,
+      "loss": 0.4245,
+      "step": 2204
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.3249655961990356,
+      "learning_rate": 5.159022894194101e-05,
+      "loss": 0.3672,
+      "step": 2205
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.215848684310913,
+      "learning_rate": 5.148460250510273e-05,
+      "loss": 0.3222,
+      "step": 2206
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.129189133644104,
+      "learning_rate": 5.1379056258659254e-05,
+      "loss": 0.3361,
+      "step": 2207
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.1304503679275513,
+      "learning_rate": 5.12735903177402e-05,
+      "loss": 0.3433,
+      "step": 2208
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.2168861627578735,
+      "learning_rate": 5.1168204797387526e-05,
+      "loss": 0.6113,
+      "step": 2209
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.3831435441970825,
+      "learning_rate": 5.106289981255543e-05,
+      "loss": 0.6427,
+      "step": 2210
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.728682518005371,
+      "learning_rate": 5.0957675478110295e-05,
+      "loss": 0.7229,
+      "step": 2211
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.1218372583389282,
+      "learning_rate": 5.08525319088306e-05,
+      "loss": 0.4039,
+      "step": 2212
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.3881877660751343,
+      "learning_rate": 5.0747469219406636e-05,
+      "loss": 0.5552,
+      "step": 2213
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.7441228628158569,
+      "learning_rate": 5.0642487524440435e-05,
+      "loss": 0.2112,
+      "step": 2214
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.9063644409179688,
+      "learning_rate": 5.053758693844583e-05,
+      "loss": 0.2105,
+      "step": 2215
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.262115240097046,
+      "learning_rate": 5.0432767575848084e-05,
+      "loss": 0.6285,
+      "step": 2216
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.1467372179031372,
+      "learning_rate": 5.032802955098383e-05,
+      "loss": 0.5605,
+      "step": 2217
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.2581697702407837,
+      "learning_rate": 5.022337297810099e-05,
+      "loss": 0.7714,
+      "step": 2218
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.472180724143982,
+      "learning_rate": 5.011879797135878e-05,
+      "loss": 0.6244,
+      "step": 2219
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.2381073236465454,
+      "learning_rate": 5.001430464482726e-05,
+      "loss": 0.3459,
+      "step": 2220
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.65796959400177,
+      "learning_rate": 4.990989311248742e-05,
+      "loss": 0.6309,
+      "step": 2221
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.3571516275405884,
+      "learning_rate": 4.9805563488231175e-05,
+      "loss": 0.4847,
+      "step": 2222
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.4138978719711304,
+      "learning_rate": 4.9701315885860937e-05,
+      "loss": 0.5352,
+      "step": 2223
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.268650770187378,
+      "learning_rate": 4.9597150419089686e-05,
+      "loss": 0.3221,
+      "step": 2224
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.0456558465957642,
+      "learning_rate": 4.94930672015409e-05,
+      "loss": 0.5437,
+      "step": 2225
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.2303794622421265,
+      "learning_rate": 4.938906634674821e-05,
+      "loss": 0.4286,
+      "step": 2226
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.1339678764343262,
+      "learning_rate": 4.928514796815545e-05,
+      "loss": 0.519,
+      "step": 2227
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.1388310194015503,
+      "learning_rate": 4.918131217911658e-05,
+      "loss": 0.3987,
+      "step": 2228
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.1729490756988525,
+      "learning_rate": 4.9077559092895296e-05,
+      "loss": 0.3761,
+      "step": 2229
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.2407212257385254,
+      "learning_rate": 4.897388882266529e-05,
+      "loss": 0.3952,
+      "step": 2230
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.1271698474884033,
+      "learning_rate": 4.887030148150975e-05,
+      "loss": 0.3412,
+      "step": 2231
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.3033124208450317,
+      "learning_rate": 4.876679718242143e-05,
+      "loss": 0.5853,
+      "step": 2232
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.2884212732315063,
+      "learning_rate": 4.8663376038302606e-05,
+      "loss": 0.5988,
+      "step": 2233
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.3859330415725708,
+      "learning_rate": 4.856003816196476e-05,
+      "loss": 0.5825,
+      "step": 2234
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.530526876449585,
+      "learning_rate": 4.84567836661285e-05,
+      "loss": 0.6353,
+      "step": 2235
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.39138662815094,
+      "learning_rate": 4.835361266342365e-05,
+      "loss": 0.5961,
+      "step": 2236
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.0767650604248047,
+      "learning_rate": 4.825052526638883e-05,
+      "loss": 0.4035,
+      "step": 2237
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.9491590857505798,
+      "learning_rate": 4.814752158747146e-05,
+      "loss": 0.4804,
+      "step": 2238
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.1009563207626343,
+      "learning_rate": 4.804460173902764e-05,
+      "loss": 0.3648,
+      "step": 2239
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.2546765804290771,
+      "learning_rate": 4.794176583332213e-05,
+      "loss": 0.4318,
+      "step": 2240
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.1268149614334106,
+      "learning_rate": 4.783901398252805e-05,
+      "loss": 0.367,
+      "step": 2241
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.9629319906234741,
+      "learning_rate": 4.773634629872677e-05,
+      "loss": 0.3666,
+      "step": 2242
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.2154501676559448,
+      "learning_rate": 4.7633762893908e-05,
+      "loss": 0.4603,
+      "step": 2243
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.1271207332611084,
+      "learning_rate": 4.7531263879969406e-05,
+      "loss": 0.7909,
+      "step": 2244
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.8619101643562317,
+      "learning_rate": 4.742884936871664e-05,
+      "loss": 0.2442,
+      "step": 2245
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.8308319449424744,
+      "learning_rate": 4.7326519471863116e-05,
+      "loss": 0.3029,
+      "step": 2246
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.9519083499908447,
+      "learning_rate": 4.7224274301030085e-05,
+      "loss": 0.2644,
+      "step": 2247
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.3647533655166626,
+      "learning_rate": 4.712211396774628e-05,
+      "loss": 0.5271,
+      "step": 2248
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.305850625038147,
+      "learning_rate": 4.7020038583447845e-05,
+      "loss": 0.635,
+      "step": 2249
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.9908753037452698,
+      "learning_rate": 4.6918048259478445e-05,
+      "loss": 0.2768,
+      "step": 2250
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.2427970170974731,
+      "learning_rate": 4.6816143107088734e-05,
+      "loss": 0.5618,
+      "step": 2251
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.0422190427780151,
+      "learning_rate": 4.671432323743667e-05,
+      "loss": 0.3206,
+      "step": 2252
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.9672759771347046,
+      "learning_rate": 4.6612588761586996e-05,
+      "loss": 0.3396,
+      "step": 2253
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.9199987053871155,
+      "learning_rate": 4.6510939790511505e-05,
+      "loss": 0.3626,
+      "step": 2254
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.0544426441192627,
+      "learning_rate": 4.640937643508857e-05,
+      "loss": 0.3362,
+      "step": 2255
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.022659182548523,
+      "learning_rate": 4.6307898806103196e-05,
+      "loss": 0.4554,
+      "step": 2256
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.1253160238265991,
+      "learning_rate": 4.620650701424689e-05,
+      "loss": 0.4351,
+      "step": 2257
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.3265272378921509,
+      "learning_rate": 4.610520117011761e-05,
+      "loss": 0.56,
+      "step": 2258
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.306456208229065,
+      "learning_rate": 4.600398138421946e-05,
+      "loss": 0.4775,
+      "step": 2259
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.0992588996887207,
+      "learning_rate": 4.590284776696267e-05,
+      "loss": 0.4471,
+      "step": 2260
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.5208834409713745,
+      "learning_rate": 4.5801800428663594e-05,
+      "loss": 0.7269,
+      "step": 2261
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.2537895441055298,
+      "learning_rate": 4.570083947954433e-05,
+      "loss": 0.407,
+      "step": 2262
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.3734372854232788,
+      "learning_rate": 4.5599965029732796e-05,
+      "loss": 0.7895,
+      "step": 2263
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.1329195499420166,
+      "learning_rate": 4.5499177189262624e-05,
+      "loss": 0.4148,
+      "step": 2264
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.6518480777740479,
+      "learning_rate": 4.539847606807289e-05,
+      "loss": 0.5311,
+      "step": 2265
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.8638691902160645,
+      "learning_rate": 4.529786177600809e-05,
+      "loss": 0.3688,
+      "step": 2266
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.7491494417190552,
+      "learning_rate": 4.5197334422817985e-05,
+      "loss": 0.4297,
+      "step": 2267
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.2566337585449219,
+      "learning_rate": 4.509689411815761e-05,
+      "loss": 0.5839,
+      "step": 2268
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.0148571729660034,
+      "learning_rate": 4.499654097158695e-05,
+      "loss": 0.2864,
+      "step": 2269
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.9205653667449951,
+      "learning_rate": 4.489627509257089e-05,
+      "loss": 0.2769,
+      "step": 2270
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.584425687789917,
+      "learning_rate": 4.479609659047927e-05,
+      "loss": 0.4954,
+      "step": 2271
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.1181970834732056,
+      "learning_rate": 4.4696005574586474e-05,
+      "loss": 0.4403,
+      "step": 2272
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.8639723658561707,
+      "learning_rate": 4.4596002154071515e-05,
+      "loss": 0.2166,
+      "step": 2273
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.1667298078536987,
+      "learning_rate": 4.449608643801778e-05,
+      "loss": 0.3535,
+      "step": 2274
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.1740378141403198,
+      "learning_rate": 4.43962585354132e-05,
+      "loss": 0.4346,
+      "step": 2275
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.2857762575149536,
+      "learning_rate": 4.42965185551497e-05,
+      "loss": 0.6013,
+      "step": 2276
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.3287543058395386,
+      "learning_rate": 4.41968666060234e-05,
+      "loss": 0.5719,
+      "step": 2277
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.3587270975112915,
+      "learning_rate": 4.409730279673431e-05,
+      "loss": 0.5678,
+      "step": 2278
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.4869335889816284,
+      "learning_rate": 4.399782723588645e-05,
+      "loss": 0.4488,
+      "step": 2279
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.63266921043396,
+      "learning_rate": 4.389844003198745e-05,
+      "loss": 0.5575,
+      "step": 2280
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.1633329391479492,
+      "learning_rate": 4.379914129344857e-05,
+      "loss": 0.3981,
+      "step": 2281
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.0538408756256104,
+      "learning_rate": 4.3699931128584666e-05,
+      "loss": 0.3228,
+      "step": 2282
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.7124279737472534,
+      "learning_rate": 4.36008096456139e-05,
+      "loss": 0.8432,
+      "step": 2283
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.0280526876449585,
+      "learning_rate": 4.350177695265768e-05,
+      "loss": 0.3095,
+      "step": 2284
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.974923849105835,
+      "learning_rate": 4.340283315774059e-05,
+      "loss": 0.2465,
+      "step": 2285
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.1634050607681274,
+      "learning_rate": 4.3303978368790296e-05,
+      "loss": 0.3219,
+      "step": 2286
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.2221295833587646,
+      "learning_rate": 4.320521269363734e-05,
+      "loss": 0.6253,
+      "step": 2287
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.442355751991272,
+      "learning_rate": 4.310653624001497e-05,
+      "loss": 0.5688,
+      "step": 2288
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.9941724538803101,
+      "learning_rate": 4.3007949115559275e-05,
+      "loss": 0.3502,
+      "step": 2289
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.345016598701477,
+      "learning_rate": 4.29094514278088e-05,
+      "loss": 0.506,
+      "step": 2290
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.1910314559936523,
+      "learning_rate": 4.2811043284204516e-05,
+      "loss": 0.4473,
+      "step": 2291
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.248070240020752,
+      "learning_rate": 4.2712724792089804e-05,
+      "loss": 0.3997,
+      "step": 2292
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.9537377953529358,
+      "learning_rate": 4.2614496058710204e-05,
+      "loss": 0.2997,
+      "step": 2293
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.08860445022583,
+      "learning_rate": 4.2516357191213345e-05,
+      "loss": 0.3847,
+      "step": 2294
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.0493781566619873,
+      "learning_rate": 4.2418308296648775e-05,
+      "loss": 0.2967,
+      "step": 2295
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.9866517782211304,
+      "learning_rate": 4.232034948196806e-05,
+      "loss": 0.3186,
+      "step": 2296
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.015709638595581,
+      "learning_rate": 4.222248085402435e-05,
+      "loss": 0.2899,
+      "step": 2297
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.0806543827056885,
+      "learning_rate": 4.212470251957244e-05,
+      "loss": 0.4528,
+      "step": 2298
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.2434687614440918,
+      "learning_rate": 4.2027014585268725e-05,
+      "loss": 0.6677,
+      "step": 2299
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.008055567741394,
+      "learning_rate": 4.1929417157670974e-05,
+      "loss": 0.299,
+      "step": 2300
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.2269865274429321,
+      "learning_rate": 4.183191034323815e-05,
+      "loss": 0.4347,
+      "step": 2301
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.9386024475097656,
+      "learning_rate": 4.173449424833037e-05,
+      "loss": 0.3465,
+      "step": 2302
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.2053933143615723,
+      "learning_rate": 4.1637168979208936e-05,
+      "loss": 0.4694,
+      "step": 2303
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.9446846842765808,
+      "learning_rate": 4.153993464203594e-05,
+      "loss": 0.2619,
+      "step": 2304
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.3338463306427002,
+      "learning_rate": 4.144279134287435e-05,
+      "loss": 0.4563,
+      "step": 2305
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.197728157043457,
+      "learning_rate": 4.1345739187687735e-05,
+      "loss": 0.4348,
+      "step": 2306
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.8967550992965698,
+      "learning_rate": 4.1248778282340446e-05,
+      "loss": 0.1982,
+      "step": 2307
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.874433696269989,
+      "learning_rate": 4.11519087325971e-05,
+      "loss": 0.2406,
+      "step": 2308
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.0845016241073608,
+      "learning_rate": 4.1055130644122696e-05,
+      "loss": 0.3382,
+      "step": 2309
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.096617579460144,
+      "learning_rate": 4.09584441224826e-05,
+      "loss": 0.385,
+      "step": 2310
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.0413081645965576,
+      "learning_rate": 4.086184927314218e-05,
+      "loss": 0.2901,
+      "step": 2311
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.065138578414917,
+      "learning_rate": 4.0765346201466766e-05,
+      "loss": 0.3099,
+      "step": 2312
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.1384869813919067,
+      "learning_rate": 4.066893501272176e-05,
+      "loss": 0.4254,
+      "step": 2313
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.032501459121704,
+      "learning_rate": 4.0572615812072136e-05,
+      "loss": 0.3063,
+      "step": 2314
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.5010746717453003,
+      "learning_rate": 4.047638870458266e-05,
+      "loss": 0.5251,
+      "step": 2315
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.0425777435302734,
+      "learning_rate": 4.0380253795217524e-05,
+      "loss": 0.287,
+      "step": 2316
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.8620198369026184,
+      "learning_rate": 4.028421118884053e-05,
+      "loss": 0.2375,
+      "step": 2317
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.0534603595733643,
+      "learning_rate": 4.018826099021464e-05,
+      "loss": 0.3245,
+      "step": 2318
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.8742746114730835,
+      "learning_rate": 4.009240330400203e-05,
+      "loss": 0.3547,
+      "step": 2319
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.9023158550262451,
+      "learning_rate": 3.9996638234764085e-05,
+      "loss": 0.5178,
+      "step": 2320
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.1507781744003296,
+      "learning_rate": 3.990096588696099e-05,
+      "loss": 0.37,
+      "step": 2321
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.4944819211959839,
+      "learning_rate": 3.980538636495197e-05,
+      "loss": 0.4217,
+      "step": 2322
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.1764553785324097,
+      "learning_rate": 3.9709899772994815e-05,
+      "loss": 0.3115,
+      "step": 2323
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.1834471225738525,
+      "learning_rate": 3.961450621524615e-05,
+      "loss": 0.4248,
+      "step": 2324
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.237874984741211,
+      "learning_rate": 3.951920579576093e-05,
+      "loss": 0.6359,
+      "step": 2325
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.123441457748413,
+      "learning_rate": 3.9423998618492604e-05,
+      "loss": 0.3109,
+      "step": 2326
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.0116443634033203,
+      "learning_rate": 3.932888478729288e-05,
+      "loss": 0.3113,
+      "step": 2327
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.0924867391586304,
+      "learning_rate": 3.9233864405911726e-05,
+      "loss": 0.2679,
+      "step": 2328
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.3951201438903809,
+      "learning_rate": 3.913893757799708e-05,
+      "loss": 0.4859,
+      "step": 2329
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.3379281759262085,
+      "learning_rate": 3.904410440709483e-05,
+      "loss": 0.6546,
+      "step": 2330
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.9945024847984314,
+      "learning_rate": 3.894936499664881e-05,
+      "loss": 0.2794,
+      "step": 2331
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.0659005641937256,
+      "learning_rate": 3.885471945000046e-05,
+      "loss": 0.4962,
+      "step": 2332
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.343820333480835,
+      "learning_rate": 3.8760167870388924e-05,
+      "loss": 0.7368,
+      "step": 2333
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.9904674291610718,
+      "learning_rate": 3.8665710360950736e-05,
+      "loss": 0.285,
+      "step": 2334
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.2321860790252686,
+      "learning_rate": 3.857134702471997e-05,
+      "loss": 0.3921,
+      "step": 2335
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.082987904548645,
+      "learning_rate": 3.847707796462786e-05,
+      "loss": 0.422,
+      "step": 2336
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.3982007503509521,
+      "learning_rate": 3.838290328350282e-05,
+      "loss": 0.4305,
+      "step": 2337
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.945181667804718,
+      "learning_rate": 3.828882308407037e-05,
+      "loss": 0.2157,
+      "step": 2338
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.1719783544540405,
+      "learning_rate": 3.8194837468952944e-05,
+      "loss": 0.2699,
+      "step": 2339
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.8167923092842102,
+      "learning_rate": 3.8100946540669715e-05,
+      "loss": 0.3927,
+      "step": 2340
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.3482931852340698,
+      "learning_rate": 3.800715040163675e-05,
+      "loss": 0.5332,
+      "step": 2341
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.9033116102218628,
+      "learning_rate": 3.7913449154166586e-05,
+      "loss": 0.2163,
+      "step": 2342
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.194532036781311,
+      "learning_rate": 3.781984290046828e-05,
+      "loss": 0.4082,
+      "step": 2343
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.3558499813079834,
+      "learning_rate": 3.7726331742647226e-05,
+      "loss": 0.4672,
+      "step": 2344
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.1556100845336914,
+      "learning_rate": 3.76329157827052e-05,
+      "loss": 0.3475,
+      "step": 2345
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.085553526878357,
+      "learning_rate": 3.7539595122540096e-05,
+      "loss": 0.347,
+      "step": 2346
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.914618968963623,
+      "learning_rate": 3.7446369863945827e-05,
+      "loss": 0.2416,
+      "step": 2347
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.248558759689331,
+      "learning_rate": 3.735324010861217e-05,
+      "loss": 0.4279,
+      "step": 2348
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.3649533987045288,
+      "learning_rate": 3.726020595812493e-05,
+      "loss": 0.3499,
+      "step": 2349
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.2154896259307861,
+      "learning_rate": 3.716726751396543e-05,
+      "loss": 0.5063,
+      "step": 2350
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.166061520576477,
+      "learning_rate": 3.707442487751064e-05,
+      "loss": 0.3876,
+      "step": 2351
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.124474048614502,
+      "learning_rate": 3.698167815003314e-05,
+      "loss": 0.4285,
+      "step": 2352
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.103945016860962,
+      "learning_rate": 3.688902743270077e-05,
+      "loss": 0.2512,
+      "step": 2353
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.8782851099967957,
+      "learning_rate": 3.679647282657668e-05,
+      "loss": 0.2574,
+      "step": 2354
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.2587329149246216,
+      "learning_rate": 3.670401443261913e-05,
+      "loss": 0.4832,
+      "step": 2355
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.9656249284744263,
+      "learning_rate": 3.661165235168157e-05,
+      "loss": 0.3533,
+      "step": 2356
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.1450453996658325,
+      "learning_rate": 3.651938668451227e-05,
+      "loss": 0.4874,
+      "step": 2357
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.9184198379516602,
+      "learning_rate": 3.64272175317543e-05,
+      "loss": 0.374,
+      "step": 2358
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.2353371381759644,
+      "learning_rate": 3.6335144993945646e-05,
+      "loss": 0.4365,
+      "step": 2359
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.098734736442566,
+      "learning_rate": 3.6243169171518724e-05,
+      "loss": 0.3502,
+      "step": 2360
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.0059494972229004,
+      "learning_rate": 3.61512901648005e-05,
+      "loss": 0.3741,
+      "step": 2361
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.4236122369766235,
+      "learning_rate": 3.60595080740123e-05,
+      "loss": 0.6132,
+      "step": 2362
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.201125144958496,
+      "learning_rate": 3.596782299926986e-05,
+      "loss": 0.4928,
+      "step": 2363
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.2504150867462158,
+      "learning_rate": 3.587623504058299e-05,
+      "loss": 0.4114,
+      "step": 2364
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.357285976409912,
+      "learning_rate": 3.578474429785553e-05,
+      "loss": 0.4101,
+      "step": 2365
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.2139153480529785,
+      "learning_rate": 3.569335087088539e-05,
+      "loss": 0.441,
+      "step": 2366
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.486601710319519,
+      "learning_rate": 3.560205485936428e-05,
+      "loss": 0.4436,
+      "step": 2367
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.061099886894226,
+      "learning_rate": 3.551085636287755e-05,
+      "loss": 0.3589,
+      "step": 2368
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.5647423267364502,
+      "learning_rate": 3.541975548090436e-05,
+      "loss": 0.5031,
+      "step": 2369
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.248528003692627,
+      "learning_rate": 3.5328752312817306e-05,
+      "loss": 0.6558,
+      "step": 2370
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.2119498252868652,
+      "learning_rate": 3.523784695788236e-05,
+      "loss": 0.509,
+      "step": 2371
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.6099773645401,
+      "learning_rate": 3.514703951525886e-05,
+      "loss": 0.4712,
+      "step": 2372
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.9579163193702698,
+      "learning_rate": 3.5056330083999235e-05,
+      "loss": 0.4728,
+      "step": 2373
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.9517368078231812,
+      "learning_rate": 3.49657187630492e-05,
+      "loss": 0.3334,
+      "step": 2374
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.0279760360717773,
+      "learning_rate": 3.487520565124729e-05,
+      "loss": 0.3993,
+      "step": 2375
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.0859653949737549,
+      "learning_rate": 3.47847908473249e-05,
+      "loss": 0.4396,
+      "step": 2376
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.1414768695831299,
+      "learning_rate": 3.4694474449906354e-05,
+      "loss": 0.4042,
+      "step": 2377
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.0698778629302979,
+      "learning_rate": 3.460425655750848e-05,
+      "loss": 0.4052,
+      "step": 2378
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.0497967004776,
+      "learning_rate": 3.451413726854069e-05,
+      "loss": 0.4626,
+      "step": 2379
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.1601060628890991,
+      "learning_rate": 3.4424116681304905e-05,
+      "loss": 0.4332,
+      "step": 2380
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.2115287780761719,
+      "learning_rate": 3.4334194893995304e-05,
+      "loss": 0.4783,
+      "step": 2381
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.2976157665252686,
+      "learning_rate": 3.4244372004698346e-05,
+      "loss": 0.2723,
+      "step": 2382
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.005487322807312,
+      "learning_rate": 3.415464811139254e-05,
+      "loss": 0.5934,
+      "step": 2383
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.8844375014305115,
+      "learning_rate": 3.406502331194855e-05,
+      "loss": 0.3645,
+      "step": 2384
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.8819464445114136,
+      "learning_rate": 3.3975497704128804e-05,
+      "loss": 0.3613,
+      "step": 2385
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.8191189169883728,
+      "learning_rate": 3.388607138558758e-05,
+      "loss": 0.299,
+      "step": 2386
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.0504474639892578,
+      "learning_rate": 3.379674445387092e-05,
+      "loss": 0.4023,
+      "step": 2387
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.0183119773864746,
+      "learning_rate": 3.370751700641636e-05,
+      "loss": 0.3868,
+      "step": 2388
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.4171295166015625,
+      "learning_rate": 3.3618389140552974e-05,
+      "loss": 0.4165,
+      "step": 2389
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.227512001991272,
+      "learning_rate": 3.3529360953501135e-05,
+      "loss": 0.491,
+      "step": 2390
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.2160135507583618,
+      "learning_rate": 3.3440432542372625e-05,
+      "loss": 0.4508,
+      "step": 2391
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.185299038887024,
+      "learning_rate": 3.335160400417024e-05,
+      "loss": 0.4429,
+      "step": 2392
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.095727562904358,
+      "learning_rate": 3.326287543578799e-05,
+      "loss": 0.3752,
+      "step": 2393
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.0111972093582153,
+      "learning_rate": 3.3174246934010665e-05,
+      "loss": 0.2842,
+      "step": 2394
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.4653085470199585,
+      "learning_rate": 3.3085718595514086e-05,
+      "loss": 0.6001,
+      "step": 2395
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.3188362121582031,
+      "learning_rate": 3.299729051686468e-05,
+      "loss": 0.3737,
+      "step": 2396
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.0367945432662964,
+      "learning_rate": 3.2908962794519495e-05,
+      "loss": 0.4076,
+      "step": 2397
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.2478039264678955,
+      "learning_rate": 3.282073552482627e-05,
+      "loss": 0.4904,
+      "step": 2398
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.0667824745178223,
+      "learning_rate": 3.273260880402303e-05,
+      "loss": 0.3117,
+      "step": 2399
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.9630646705627441,
+      "learning_rate": 3.264458272823814e-05,
+      "loss": 0.3421,
+      "step": 2400
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.0218043327331543,
+      "learning_rate": 3.255665739349019e-05,
+      "loss": 0.4979,
+      "step": 2401
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.0328983068466187,
+      "learning_rate": 3.246883289568796e-05,
+      "loss": 0.3001,
+      "step": 2402
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.900199830532074,
+      "learning_rate": 3.238110933063014e-05,
+      "loss": 0.2591,
+      "step": 2403
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.41981840133667,
+      "learning_rate": 3.2293486794005303e-05,
+      "loss": 0.6724,
+      "step": 2404
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.0938422679901123,
+      "learning_rate": 3.220596538139196e-05,
+      "loss": 0.5295,
+      "step": 2405
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.074320673942566,
+      "learning_rate": 3.21185451882582e-05,
+      "loss": 0.2767,
+      "step": 2406
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.9862778186798096,
+      "learning_rate": 3.2031226309961675e-05,
+      "loss": 0.2652,
+      "step": 2407
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.1148425340652466,
+      "learning_rate": 3.1944008841749664e-05,
+      "loss": 0.2569,
+      "step": 2408
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.2387659549713135,
+      "learning_rate": 3.18568928787587e-05,
+      "loss": 0.4366,
+      "step": 2409
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.2367215156555176,
+      "learning_rate": 3.1769878516014635e-05,
+      "loss": 0.5176,
+      "step": 2410
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.0763795375823975,
+      "learning_rate": 3.168296584843248e-05,
+      "loss": 0.3899,
+      "step": 2411
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.3850998878479004,
+      "learning_rate": 3.1596154970816385e-05,
+      "loss": 0.5462,
+      "step": 2412
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.4004285335540771,
+      "learning_rate": 3.1509445977859405e-05,
+      "loss": 0.365,
+      "step": 2413
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.9528489708900452,
+      "learning_rate": 3.142283896414343e-05,
+      "loss": 0.2813,
+      "step": 2414
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.764185905456543,
+      "learning_rate": 3.133633402413917e-05,
+      "loss": 0.6395,
+      "step": 2415
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.0002202987670898,
+      "learning_rate": 3.124993125220603e-05,
+      "loss": 0.2165,
+      "step": 2416
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.2747881412506104,
+      "learning_rate": 3.116363074259188e-05,
+      "loss": 0.47,
+      "step": 2417
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.329888939857483,
+      "learning_rate": 3.1077432589433046e-05,
+      "loss": 0.4552,
+      "step": 2418
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.0620276927947998,
+      "learning_rate": 3.09913368867543e-05,
+      "loss": 0.3858,
+      "step": 2419
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.7584563493728638,
+      "learning_rate": 3.090534372846858e-05,
+      "loss": 0.7671,
+      "step": 2420
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.1476322412490845,
+      "learning_rate": 3.081945320837698e-05,
+      "loss": 0.4855,
+      "step": 2421
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.1755834817886353,
+      "learning_rate": 3.073366542016862e-05,
+      "loss": 0.335,
+      "step": 2422
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.312078595161438,
+      "learning_rate": 3.064798045742062e-05,
+      "loss": 0.5109,
+      "step": 2423
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.0246888399124146,
+      "learning_rate": 3.056239841359791e-05,
+      "loss": 0.2884,
+      "step": 2424
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.1685364246368408,
+      "learning_rate": 3.047691938205309e-05,
+      "loss": 0.4957,
+      "step": 2425
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.8868763446807861,
+      "learning_rate": 3.0391543456026543e-05,
+      "loss": 0.2575,
+      "step": 2426
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.6953266859054565,
+      "learning_rate": 3.0306270728646042e-05,
+      "loss": 0.5522,
+      "step": 2427
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.304674506187439,
+      "learning_rate": 3.0221101292926823e-05,
+      "loss": 0.216,
+      "step": 2428
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.3259423971176147,
+      "learning_rate": 3.0136035241771524e-05,
+      "loss": 0.4338,
+      "step": 2429
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.0263630151748657,
+      "learning_rate": 3.0051072667969956e-05,
+      "loss": 0.4382,
+      "step": 2430
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.4216090440750122,
+      "learning_rate": 2.996621366419905e-05,
+      "loss": 0.4007,
+      "step": 2431
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.4432778358459473,
+      "learning_rate": 2.9881458323022733e-05,
+      "loss": 0.5448,
+      "step": 2432
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.9634547233581543,
+      "learning_rate": 2.9796806736891975e-05,
+      "loss": 0.3259,
+      "step": 2433
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.2892754077911377,
+      "learning_rate": 2.971225899814446e-05,
+      "loss": 0.5287,
+      "step": 2434
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.2004163265228271,
+      "learning_rate": 2.9627815199004593e-05,
+      "loss": 0.4014,
+      "step": 2435
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.484840750694275,
+      "learning_rate": 2.9543475431583505e-05,
+      "loss": 0.4791,
+      "step": 2436
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.1654618978500366,
+      "learning_rate": 2.9459239787878776e-05,
+      "loss": 0.3014,
+      "step": 2437
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.3821611404418945,
+      "learning_rate": 2.937510835977439e-05,
+      "loss": 0.5759,
+      "step": 2438
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.085803508758545,
+      "learning_rate": 2.9291081239040605e-05,
+      "loss": 0.3286,
+      "step": 2439
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.9716854691505432,
+      "learning_rate": 2.920715851733413e-05,
+      "loss": 0.2903,
+      "step": 2440
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.0556275844573975,
+      "learning_rate": 2.9123340286197565e-05,
+      "loss": 0.4209,
+      "step": 2441
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.300984263420105,
+      "learning_rate": 2.9039626637059612e-05,
+      "loss": 0.3497,
+      "step": 2442
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.464999794960022,
+      "learning_rate": 2.8956017661234846e-05,
+      "loss": 0.5126,
+      "step": 2443
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.9424381852149963,
+      "learning_rate": 2.8872513449923797e-05,
+      "loss": 0.3025,
+      "step": 2444
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.313447117805481,
+      "learning_rate": 2.8789114094212575e-05,
+      "loss": 0.4539,
+      "step": 2445
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.538489580154419,
+      "learning_rate": 2.870581968507295e-05,
+      "loss": 0.6295,
+      "step": 2446
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.1631628274917603,
+      "learning_rate": 2.8622630313362303e-05,
+      "loss": 0.3899,
+      "step": 2447
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.0515528917312622,
+      "learning_rate": 2.8539546069823327e-05,
+      "loss": 0.33,
+      "step": 2448
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.3940054178237915,
+      "learning_rate": 2.8456567045084105e-05,
+      "loss": 0.4198,
+      "step": 2449
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.207826018333435,
+      "learning_rate": 2.8373693329657878e-05,
+      "loss": 0.5596,
+      "step": 2450
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.2761030197143555,
+      "learning_rate": 2.8290925013943147e-05,
+      "loss": 0.496,
+      "step": 2451
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.7271332740783691,
+      "learning_rate": 2.8208262188223337e-05,
+      "loss": 0.7644,
+      "step": 2452
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.387328028678894,
+      "learning_rate": 2.8125704942666794e-05,
+      "loss": 0.5055,
+      "step": 2453
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.200985312461853,
+      "learning_rate": 2.80432533673268e-05,
+      "loss": 0.4813,
+      "step": 2454
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.1662079095840454,
+      "learning_rate": 2.796090755214127e-05,
+      "loss": 0.4065,
+      "step": 2455
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.2971113920211792,
+      "learning_rate": 2.7878667586932776e-05,
+      "loss": 0.4729,
+      "step": 2456
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.2580766677856445,
+      "learning_rate": 2.7796533561408502e-05,
+      "loss": 0.3174,
+      "step": 2457
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.2634315490722656,
+      "learning_rate": 2.771450556515999e-05,
+      "loss": 0.3987,
+      "step": 2458
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.1510578393936157,
+      "learning_rate": 2.763258368766315e-05,
+      "loss": 0.3606,
+      "step": 2459
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.9381775856018066,
+      "learning_rate": 2.7550768018278107e-05,
+      "loss": 0.2334,
+      "step": 2460
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.036440134048462,
+      "learning_rate": 2.7469058646249255e-05,
+      "loss": 0.3621,
+      "step": 2461
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.370670199394226,
+      "learning_rate": 2.7387455660704857e-05,
+      "loss": 0.7745,
+      "step": 2462
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.025439977645874,
+      "learning_rate": 2.7305959150657326e-05,
+      "loss": 0.3705,
+      "step": 2463
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.114072322845459,
+      "learning_rate": 2.722456920500274e-05,
+      "loss": 0.4108,
+      "step": 2464
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.997879683971405,
+      "learning_rate": 2.7143285912521117e-05,
+      "loss": 0.2627,
+      "step": 2465
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.13973069190979,
+      "learning_rate": 2.7062109361876003e-05,
+      "loss": 0.4388,
+      "step": 2466
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.333604335784912,
+      "learning_rate": 2.6981039641614523e-05,
+      "loss": 0.4221,
+      "step": 2467
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.3350666761398315,
+      "learning_rate": 2.6900076840167377e-05,
+      "loss": 0.5773,
+      "step": 2468
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.058940052986145,
+      "learning_rate": 2.6819221045848538e-05,
+      "loss": 0.3743,
+      "step": 2469
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.0323387384414673,
+      "learning_rate": 2.673847234685528e-05,
+      "loss": 0.2662,
+      "step": 2470
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.5275335311889648,
+      "learning_rate": 2.6657830831268034e-05,
+      "loss": 0.6337,
+      "step": 2471
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.729491949081421,
+      "learning_rate": 2.6577296587050424e-05,
+      "loss": 0.9006,
+      "step": 2472
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.2076356410980225,
+      "learning_rate": 2.6496869702048925e-05,
+      "loss": 0.4235,
+      "step": 2473
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.8329384326934814,
+      "learning_rate": 2.6416550263992936e-05,
+      "loss": 0.2444,
+      "step": 2474
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.3011808395385742,
+      "learning_rate": 2.633633836049476e-05,
+      "loss": 0.295,
+      "step": 2475
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.2912174463272095,
+      "learning_rate": 2.6256234079049286e-05,
+      "loss": 0.4929,
+      "step": 2476
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.389458179473877,
+      "learning_rate": 2.6176237507034044e-05,
+      "loss": 0.3581,
+      "step": 2477
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.1142305135726929,
+      "learning_rate": 2.6096348731709033e-05,
+      "loss": 0.4159,
+      "step": 2478
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.8560962677001953,
+      "learning_rate": 2.601656784021679e-05,
+      "loss": 0.2975,
+      "step": 2479
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.4160083532333374,
+      "learning_rate": 2.5936894919582067e-05,
+      "loss": 0.385,
+      "step": 2480
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.0558053255081177,
+      "learning_rate": 2.58573300567118e-05,
+      "loss": 0.4253,
+      "step": 2481
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.969610333442688,
+      "learning_rate": 2.5777873338395235e-05,
+      "loss": 0.2576,
+      "step": 2482
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.3804552555084229,
+      "learning_rate": 2.5698524851303488e-05,
+      "loss": 0.5133,
+      "step": 2483
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.0985430479049683,
+      "learning_rate": 2.5619284681989624e-05,
+      "loss": 0.3691,
+      "step": 2484
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.1674649715423584,
+      "learning_rate": 2.55401529168887e-05,
+      "loss": 0.329,
+      "step": 2485
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.197788119316101,
+      "learning_rate": 2.546112964231735e-05,
+      "loss": 0.3808,
+      "step": 2486
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.2476664781570435,
+      "learning_rate": 2.5382214944474012e-05,
+      "loss": 0.4382,
+      "step": 2487
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.19683039188385,
+      "learning_rate": 2.5303408909438578e-05,
+      "loss": 0.4408,
+      "step": 2488
+    },
+    {
+      "epoch": 2.4,
+      "eval_loss": 1.3875722885131836,
+      "eval_runtime": 28.1984,
+      "eval_samples_per_second": 3.014,
+      "eval_steps_per_second": 1.525,
+      "step": 2488
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.302209734916687,
+      "learning_rate": 2.5224711623172442e-05,
+      "loss": 0.4414,
+      "step": 2489
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.1015681028366089,
+      "learning_rate": 2.514612317151846e-05,
+      "loss": 0.3905,
+      "step": 2490
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.0991705656051636,
+      "learning_rate": 2.5067643640200642e-05,
+      "loss": 0.4307,
+      "step": 2491
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.901372492313385,
+      "learning_rate": 2.4989273114824228e-05,
+      "loss": 0.2361,
+      "step": 2492
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.8259288668632507,
+      "learning_rate": 2.491101168087563e-05,
+      "loss": 0.2363,
+      "step": 2493
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.2005654573440552,
+      "learning_rate": 2.4832859423722183e-05,
+      "loss": 0.4297,
+      "step": 2494
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.0432602167129517,
+      "learning_rate": 2.475481642861213e-05,
+      "loss": 0.2909,
+      "step": 2495
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.3437654972076416,
+      "learning_rate": 2.4676882780674606e-05,
+      "loss": 0.4414,
+      "step": 2496
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.3620154857635498,
+      "learning_rate": 2.4599058564919396e-05,
+      "loss": 0.4478,
+      "step": 2497
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.377049207687378,
+      "learning_rate": 2.4521343866236945e-05,
+      "loss": 0.4468,
+      "step": 2498
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.2696644067764282,
+      "learning_rate": 2.444373876939819e-05,
+      "loss": 0.4608,
+      "step": 2499
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.1020128726959229,
+      "learning_rate": 2.4366243359054642e-05,
+      "loss": 0.4635,
+      "step": 2500
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.926050066947937,
+      "learning_rate": 2.4288857719738044e-05,
+      "loss": 0.2092,
+      "step": 2501
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.5648494958877563,
+      "learning_rate": 2.42115819358604e-05,
+      "loss": 0.7375,
+      "step": 2502
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.2817730903625488,
+      "learning_rate": 2.413441609171399e-05,
+      "loss": 0.5618,
+      "step": 2503
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.5735052824020386,
+      "learning_rate": 2.405736027147107e-05,
+      "loss": 0.5577,
+      "step": 2504
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.031183123588562,
+      "learning_rate": 2.398041455918394e-05,
+      "loss": 0.4408,
+      "step": 2505
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.3467059135437012,
+      "learning_rate": 2.390357903878472e-05,
+      "loss": 0.4451,
+      "step": 2506
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.2030715942382812,
+      "learning_rate": 2.3826853794085445e-05,
+      "loss": 0.4461,
+      "step": 2507
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.080093502998352,
+      "learning_rate": 2.3750238908777796e-05,
+      "loss": 0.3372,
+      "step": 2508
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.9159194231033325,
+      "learning_rate": 2.367373446643302e-05,
+      "loss": 0.3811,
+      "step": 2509
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.255796194076538,
+      "learning_rate": 2.3597340550502015e-05,
+      "loss": 0.5714,
+      "step": 2510
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.7954977750778198,
+      "learning_rate": 2.3521057244315066e-05,
+      "loss": 0.739,
+      "step": 2511
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.65315842628479,
+      "learning_rate": 2.344488463108177e-05,
+      "loss": 0.5431,
+      "step": 2512
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.9481430053710938,
+      "learning_rate": 2.3368822793890963e-05,
+      "loss": 0.344,
+      "step": 2513
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.8809416890144348,
+      "learning_rate": 2.3292871815710783e-05,
+      "loss": 0.1632,
+      "step": 2514
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.9627254605293274,
+      "learning_rate": 2.3217031779388276e-05,
+      "loss": 0.2672,
+      "step": 2515
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.5389256477355957,
+      "learning_rate": 2.3141302767649586e-05,
+      "loss": 0.8815,
+      "step": 2516
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.2893158197402954,
+      "learning_rate": 2.306568486309965e-05,
+      "loss": 0.3986,
+      "step": 2517
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.3341116905212402,
+      "learning_rate": 2.2990178148222365e-05,
+      "loss": 0.3183,
+      "step": 2518
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.7593715190887451,
+      "learning_rate": 2.2914782705380192e-05,
+      "loss": 0.5908,
+      "step": 2519
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.4277440309524536,
+      "learning_rate": 2.2839498616814252e-05,
+      "loss": 0.6308,
+      "step": 2520
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.268128752708435,
+      "learning_rate": 2.2764325964644285e-05,
+      "loss": 0.5316,
+      "step": 2521
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.3595266342163086,
+      "learning_rate": 2.2689264830868375e-05,
+      "loss": 0.4631,
+      "step": 2522
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.1918679475784302,
+      "learning_rate": 2.261431529736299e-05,
+      "loss": 0.3397,
+      "step": 2523
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.1698386669158936,
+      "learning_rate": 2.253947744588293e-05,
+      "loss": 0.5497,
+      "step": 2524
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.8681797385215759,
+      "learning_rate": 2.2464751358061086e-05,
+      "loss": 0.309,
+      "step": 2525
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.7204582691192627,
+      "learning_rate": 2.2390137115408465e-05,
+      "loss": 0.8977,
+      "step": 2526
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.3766608238220215,
+      "learning_rate": 2.2315634799314062e-05,
+      "loss": 0.5383,
+      "step": 2527
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.0767016410827637,
+      "learning_rate": 2.224124449104485e-05,
+      "loss": 0.4095,
+      "step": 2528
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.1819124221801758,
+      "learning_rate": 2.216696627174554e-05,
+      "loss": 0.4,
+      "step": 2529
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.377302646636963,
+      "learning_rate": 2.2092800222438607e-05,
+      "loss": 0.5399,
+      "step": 2530
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.3226739168167114,
+      "learning_rate": 2.2018746424024198e-05,
+      "loss": 0.5938,
+      "step": 2531
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.3821479082107544,
+      "learning_rate": 2.194480495727999e-05,
+      "loss": 0.2878,
+      "step": 2532
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.3803555965423584,
+      "learning_rate": 2.187097590286109e-05,
+      "loss": 0.4424,
+      "step": 2533
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.3531436920166016,
+      "learning_rate": 2.1797259341300057e-05,
+      "loss": 0.4328,
+      "step": 2534
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.4491779804229736,
+      "learning_rate": 2.172365535300677e-05,
+      "loss": 0.5579,
+      "step": 2535
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.1792582273483276,
+      "learning_rate": 2.1650164018268207e-05,
+      "loss": 0.4401,
+      "step": 2536
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.2620794773101807,
+      "learning_rate": 2.1576785417248517e-05,
+      "loss": 0.4252,
+      "step": 2537
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.075289011001587,
+      "learning_rate": 2.150351962998884e-05,
+      "loss": 0.4413,
+      "step": 2538
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.0587613582611084,
+      "learning_rate": 2.1430366736407372e-05,
+      "loss": 0.3882,
+      "step": 2539
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.1711111068725586,
+      "learning_rate": 2.1357326816299052e-05,
+      "loss": 0.3951,
+      "step": 2540
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.9997272491455078,
+      "learning_rate": 2.1284399949335608e-05,
+      "loss": 0.2251,
+      "step": 2541
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.2561250925064087,
+      "learning_rate": 2.1211586215065498e-05,
+      "loss": 0.4576,
+      "step": 2542
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.2198114395141602,
+      "learning_rate": 2.113888569291375e-05,
+      "loss": 0.5197,
+      "step": 2543
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.087884783744812,
+      "learning_rate": 2.1066298462181833e-05,
+      "loss": 0.4415,
+      "step": 2544
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.8363281488418579,
+      "learning_rate": 2.099382460204778e-05,
+      "loss": 0.2089,
+      "step": 2545
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.0022271871566772,
+      "learning_rate": 2.092146419156586e-05,
+      "loss": 0.2795,
+      "step": 2546
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.191450834274292,
+      "learning_rate": 2.0849217309666575e-05,
+      "loss": 0.3494,
+      "step": 2547
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.3831026554107666,
+      "learning_rate": 2.077708403515663e-05,
+      "loss": 0.488,
+      "step": 2548
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.0185648202896118,
+      "learning_rate": 2.070506444671885e-05,
+      "loss": 0.2596,
+      "step": 2549
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.1526098251342773,
+      "learning_rate": 2.0633158622911973e-05,
+      "loss": 0.3748,
+      "step": 2550
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.4703809022903442,
+      "learning_rate": 2.056136664217065e-05,
+      "loss": 0.4452,
+      "step": 2551
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.032928466796875,
+      "learning_rate": 2.0489688582805436e-05,
+      "loss": 0.3714,
+      "step": 2552
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.4959150552749634,
+      "learning_rate": 2.041812452300254e-05,
+      "loss": 0.526,
+      "step": 2553
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.0824545621871948,
+      "learning_rate": 2.0346674540823827e-05,
+      "loss": 0.456,
+      "step": 2554
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.9160414934158325,
+      "learning_rate": 2.0275338714206725e-05,
+      "loss": 0.3043,
+      "step": 2555
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.0382851362228394,
+      "learning_rate": 2.0204117120964177e-05,
+      "loss": 0.2935,
+      "step": 2556
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.2792209386825562,
+      "learning_rate": 2.013300983878455e-05,
+      "loss": 0.4012,
+      "step": 2557
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.9995631575584412,
+      "learning_rate": 2.0062016945231438e-05,
+      "loss": 0.2868,
+      "step": 2558
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.1381494998931885,
+      "learning_rate": 1.9991138517743644e-05,
+      "loss": 0.3517,
+      "step": 2559
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.2145975828170776,
+      "learning_rate": 1.992037463363526e-05,
+      "loss": 0.4534,
+      "step": 2560
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.2908990383148193,
+      "learning_rate": 1.9849725370095274e-05,
+      "loss": 0.4325,
+      "step": 2561
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.0607693195343018,
+      "learning_rate": 1.9779190804187705e-05,
+      "loss": 0.318,
+      "step": 2562
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.806749165058136,
+      "learning_rate": 1.9708771012851513e-05,
+      "loss": 0.1497,
+      "step": 2563
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.0389070510864258,
+      "learning_rate": 1.9638466072900374e-05,
+      "loss": 0.3352,
+      "step": 2564
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.8952654600143433,
+      "learning_rate": 1.956827606102274e-05,
+      "loss": 0.6747,
+      "step": 2565
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.2378493547439575,
+      "learning_rate": 1.9498201053781654e-05,
+      "loss": 0.2968,
+      "step": 2566
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.103607177734375,
+      "learning_rate": 1.9428241127614794e-05,
+      "loss": 0.341,
+      "step": 2567
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.4015883207321167,
+      "learning_rate": 1.935839635883424e-05,
+      "loss": 0.5368,
+      "step": 2568
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.289624571800232,
+      "learning_rate": 1.9288666823626426e-05,
+      "loss": 0.3724,
+      "step": 2569
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.9046264886856079,
+      "learning_rate": 1.9219052598052206e-05,
+      "loss": 0.3472,
+      "step": 2570
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.072669506072998,
+      "learning_rate": 1.914955375804657e-05,
+      "loss": 0.3162,
+      "step": 2571
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.3176350593566895,
+      "learning_rate": 1.9080170379418616e-05,
+      "loss": 0.2784,
+      "step": 2572
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.032930612564087,
+      "learning_rate": 1.9010902537851607e-05,
+      "loss": 0.2979,
+      "step": 2573
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.0955973863601685,
+      "learning_rate": 1.894175030890269e-05,
+      "loss": 0.3136,
+      "step": 2574
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.3473103046417236,
+      "learning_rate": 1.8872713768002925e-05,
+      "loss": 0.3991,
+      "step": 2575
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.9937989711761475,
+      "learning_rate": 1.880379299045716e-05,
+      "loss": 0.32,
+      "step": 2576
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.153117299079895,
+      "learning_rate": 1.8734988051444038e-05,
+      "loss": 0.3852,
+      "step": 2577
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.0335619449615479,
+      "learning_rate": 1.8666299026015785e-05,
+      "loss": 0.2585,
+      "step": 2578
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.1094386577606201,
+      "learning_rate": 1.8597725989098145e-05,
+      "loss": 0.3719,
+      "step": 2579
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.0681630373001099,
+      "learning_rate": 1.852926901549047e-05,
+      "loss": 0.3646,
+      "step": 2580
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.2787219285964966,
+      "learning_rate": 1.846092817986543e-05,
+      "loss": 0.418,
+      "step": 2581
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.0077821016311646,
+      "learning_rate": 1.8392703556769016e-05,
+      "loss": 0.5207,
+      "step": 2582
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.9804518222808838,
+      "learning_rate": 1.832459522062041e-05,
+      "loss": 0.2823,
+      "step": 2583
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.2154254913330078,
+      "learning_rate": 1.8256603245712087e-05,
+      "loss": 0.4744,
+      "step": 2584
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.9940858483314514,
+      "learning_rate": 1.8188727706209456e-05,
+      "loss": 0.4446,
+      "step": 2585
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.0796043872833252,
+      "learning_rate": 1.8120968676150983e-05,
+      "loss": 0.3029,
+      "step": 2586
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.2031382322311401,
+      "learning_rate": 1.8053326229447998e-05,
+      "loss": 0.4059,
+      "step": 2587
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.7458419799804688,
+      "learning_rate": 1.798580043988475e-05,
+      "loss": 0.4546,
+      "step": 2588
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.965041995048523,
+      "learning_rate": 1.7918391381118158e-05,
+      "loss": 0.2469,
+      "step": 2589
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.9897942543029785,
+      "learning_rate": 1.7851099126677816e-05,
+      "loss": 0.2851,
+      "step": 2590
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.304334282875061,
+      "learning_rate": 1.7783923749966e-05,
+      "loss": 0.4326,
+      "step": 2591
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.433759331703186,
+      "learning_rate": 1.771686532425737e-05,
+      "loss": 0.4978,
+      "step": 2592
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.1951675415039062,
+      "learning_rate": 1.7649923922699096e-05,
+      "loss": 0.4889,
+      "step": 2593
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.2474220991134644,
+      "learning_rate": 1.7583099618310645e-05,
+      "loss": 0.5135,
+      "step": 2594
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.3049352169036865,
+      "learning_rate": 1.751639248398383e-05,
+      "loss": 0.29,
+      "step": 2595
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.994936466217041,
+      "learning_rate": 1.7449802592482605e-05,
+      "loss": 0.751,
+      "step": 2596
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.0624207258224487,
+      "learning_rate": 1.738333001644299e-05,
+      "loss": 0.3416,
+      "step": 2597
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.1713080406188965,
+      "learning_rate": 1.731697482837316e-05,
+      "loss": 0.4251,
+      "step": 2598
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.1803196668624878,
+      "learning_rate": 1.7250737100653127e-05,
+      "loss": 0.3597,
+      "step": 2599
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.2651060819625854,
+      "learning_rate": 1.7184616905534828e-05,
+      "loss": 0.4793,
+      "step": 2600
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.1736713647842407,
+      "learning_rate": 1.7118614315142014e-05,
+      "loss": 0.3246,
+      "step": 2601
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.0619823932647705,
+      "learning_rate": 1.7052729401470123e-05,
+      "loss": 0.2607,
+      "step": 2602
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.2877070903778076,
+      "learning_rate": 1.6986962236386212e-05,
+      "loss": 0.3589,
+      "step": 2603
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.2267651557922363,
+      "learning_rate": 1.6921312891628974e-05,
+      "loss": 0.4943,
+      "step": 2604
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 1.2303855419158936,
+      "learning_rate": 1.685578143880846e-05,
+      "loss": 0.5195,
+      "step": 2605
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.031379222869873,
+      "learning_rate": 1.6790367949406292e-05,
+      "loss": 0.4127,
+      "step": 2606
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.1098753213882446,
+      "learning_rate": 1.672507249477527e-05,
+      "loss": 0.333,
+      "step": 2607
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.3014777898788452,
+      "learning_rate": 1.6659895146139498e-05,
+      "loss": 0.4767,
+      "step": 2608
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.0662829875946045,
+      "learning_rate": 1.6594835974594267e-05,
+      "loss": 0.442,
+      "step": 2609
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.6387438774108887,
+      "learning_rate": 1.652989505110594e-05,
+      "loss": 0.489,
+      "step": 2610
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.8665334582328796,
+      "learning_rate": 1.646507244651188e-05,
+      "loss": 0.2739,
+      "step": 2611
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.9079167246818542,
+      "learning_rate": 1.640036823152044e-05,
+      "loss": 0.177,
+      "step": 2612
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.1278053522109985,
+      "learning_rate": 1.633578247671079e-05,
+      "loss": 0.1977,
+      "step": 2613
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.1719645261764526,
+      "learning_rate": 1.6271315252532877e-05,
+      "loss": 0.4519,
+      "step": 2614
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.4588279724121094,
+      "learning_rate": 1.6206966629307373e-05,
+      "loss": 0.4433,
+      "step": 2615
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.8502780199050903,
+      "learning_rate": 1.6142736677225602e-05,
+      "loss": 0.1827,
+      "step": 2616
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.2789843082427979,
+      "learning_rate": 1.6078625466349416e-05,
+      "loss": 0.4402,
+      "step": 2617
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.2401293516159058,
+      "learning_rate": 1.601463306661112e-05,
+      "loss": 0.5667,
+      "step": 2618
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.4416359663009644,
+      "learning_rate": 1.5950759547813498e-05,
+      "loss": 0.4659,
+      "step": 2619
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.694312572479248,
+      "learning_rate": 1.5887004979629595e-05,
+      "loss": 0.6871,
+      "step": 2620
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.3438770771026611,
+      "learning_rate": 1.5823369431602687e-05,
+      "loss": 0.4474,
+      "step": 2621
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.0863066911697388,
+      "learning_rate": 1.5759852973146263e-05,
+      "loss": 0.3095,
+      "step": 2622
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.8890448212623596,
+      "learning_rate": 1.5696455673543955e-05,
+      "loss": 0.2267,
+      "step": 2623
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.167129397392273,
+      "learning_rate": 1.563317760194931e-05,
+      "loss": 0.3832,
+      "step": 2624
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.2073942422866821,
+      "learning_rate": 1.5570018827385873e-05,
+      "loss": 0.3542,
+      "step": 2625
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.1186821460723877,
+      "learning_rate": 1.550697941874711e-05,
+      "loss": 0.323,
+      "step": 2626
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.2039204835891724,
+      "learning_rate": 1.544405944479617e-05,
+      "loss": 0.4761,
+      "step": 2627
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.3113536834716797,
+      "learning_rate": 1.5381258974166063e-05,
+      "loss": 0.5072,
+      "step": 2628
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.8303695321083069,
+      "learning_rate": 1.5318578075359314e-05,
+      "loss": 0.2384,
+      "step": 2629
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.0342708826065063,
+      "learning_rate": 1.525601681674811e-05,
+      "loss": 0.4,
+      "step": 2630
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.3412367105484009,
+      "learning_rate": 1.5193575266574097e-05,
+      "loss": 0.423,
+      "step": 2631
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.3921438455581665,
+      "learning_rate": 1.5131253492948332e-05,
+      "loss": 0.4646,
+      "step": 2632
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.6994187831878662,
+      "learning_rate": 1.5069051563851216e-05,
+      "loss": 0.7081,
+      "step": 2633
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.210211157798767,
+      "learning_rate": 1.5006969547132502e-05,
+      "loss": 0.3938,
+      "step": 2634
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.8103139400482178,
+      "learning_rate": 1.4945007510511046e-05,
+      "loss": 0.2196,
+      "step": 2635
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.6641013622283936,
+      "learning_rate": 1.4883165521574857e-05,
+      "loss": 0.5646,
+      "step": 2636
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.088254451751709,
+      "learning_rate": 1.4821443647781041e-05,
+      "loss": 0.3477,
+      "step": 2637
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.6455141305923462,
+      "learning_rate": 1.4759841956455632e-05,
+      "loss": 0.5446,
+      "step": 2638
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.1219127178192139,
+      "learning_rate": 1.4698360514793563e-05,
+      "loss": 0.331,
+      "step": 2639
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.228132963180542,
+      "learning_rate": 1.4636999389858686e-05,
+      "loss": 0.4265,
+      "step": 2640
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.4756618738174438,
+      "learning_rate": 1.4575758648583487e-05,
+      "loss": 0.7019,
+      "step": 2641
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.95322585105896,
+      "learning_rate": 1.4514638357769241e-05,
+      "loss": 0.301,
+      "step": 2642
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.2768433094024658,
+      "learning_rate": 1.4453638584085744e-05,
+      "loss": 0.4116,
+      "step": 2643
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.3799253702163696,
+      "learning_rate": 1.4392759394071433e-05,
+      "loss": 0.4148,
+      "step": 2644
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.175272822380066,
+      "learning_rate": 1.4332000854133151e-05,
+      "loss": 0.3961,
+      "step": 2645
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 1.1589256525039673,
+      "learning_rate": 1.4271363030546116e-05,
+      "loss": 0.4175,
+      "step": 2646
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.0186060667037964,
+      "learning_rate": 1.4210845989453949e-05,
+      "loss": 0.3345,
+      "step": 2647
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.1230732202529907,
+      "learning_rate": 1.4150449796868442e-05,
+      "loss": 0.36,
+      "step": 2648
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.1087164878845215,
+      "learning_rate": 1.4090174518669605e-05,
+      "loss": 0.3748,
+      "step": 2649
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.1244251728057861,
+      "learning_rate": 1.4030020220605497e-05,
+      "loss": 0.2803,
+      "step": 2650
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.1560827493667603,
+      "learning_rate": 1.3969986968292366e-05,
+      "loss": 0.4778,
+      "step": 2651
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.9187933206558228,
+      "learning_rate": 1.391007482721425e-05,
+      "loss": 0.3128,
+      "step": 2652
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.490533471107483,
+      "learning_rate": 1.3850283862723176e-05,
+      "loss": 0.4867,
+      "step": 2653
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.209583044052124,
+      "learning_rate": 1.3790614140038918e-05,
+      "loss": 0.3516,
+      "step": 2654
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.234925389289856,
+      "learning_rate": 1.3731065724249107e-05,
+      "loss": 0.5393,
+      "step": 2655
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.2917462587356567,
+      "learning_rate": 1.3671638680308976e-05,
+      "loss": 0.4702,
+      "step": 2656
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.0113184452056885,
+      "learning_rate": 1.3612333073041342e-05,
+      "loss": 0.2885,
+      "step": 2657
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.164046049118042,
+      "learning_rate": 1.355314896713665e-05,
+      "loss": 0.4355,
+      "step": 2658
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.1011210680007935,
+      "learning_rate": 1.3494086427152732e-05,
+      "loss": 0.3958,
+      "step": 2659
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.180242896080017,
+      "learning_rate": 1.343514551751486e-05,
+      "loss": 0.3329,
+      "step": 2660
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.20024836063385,
+      "learning_rate": 1.337632630251559e-05,
+      "loss": 0.4231,
+      "step": 2661
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.0782047510147095,
+      "learning_rate": 1.3317628846314808e-05,
+      "loss": 0.344,
+      "step": 2662
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.1018481254577637,
+      "learning_rate": 1.3259053212939507e-05,
+      "loss": 0.3958,
+      "step": 2663
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.0811465978622437,
+      "learning_rate": 1.320059946628381e-05,
+      "loss": 0.3304,
+      "step": 2664
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.1098639965057373,
+      "learning_rate": 1.3142267670108953e-05,
+      "loss": 0.3149,
+      "step": 2665
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.4178462028503418,
+      "learning_rate": 1.308405788804308e-05,
+      "loss": 0.3847,
+      "step": 2666
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.8337834477424622,
+      "learning_rate": 1.3025970183581232e-05,
+      "loss": 0.1917,
+      "step": 2667
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.899633526802063,
+      "learning_rate": 1.2968004620085363e-05,
+      "loss": 0.2307,
+      "step": 2668
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.339258074760437,
+      "learning_rate": 1.2910161260784149e-05,
+      "loss": 0.362,
+      "step": 2669
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.5382481813430786,
+      "learning_rate": 1.2852440168772933e-05,
+      "loss": 0.4168,
+      "step": 2670
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.2240447998046875,
+      "learning_rate": 1.279484140701373e-05,
+      "loss": 0.3134,
+      "step": 2671
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.4382638931274414,
+      "learning_rate": 1.2737365038335141e-05,
+      "loss": 0.4588,
+      "step": 2672
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.4672664403915405,
+      "learning_rate": 1.2680011125432212e-05,
+      "loss": 0.804,
+      "step": 2673
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.018543004989624,
+      "learning_rate": 1.2622779730866424e-05,
+      "loss": 0.3097,
+      "step": 2674
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.3507752418518066,
+      "learning_rate": 1.2565670917065647e-05,
+      "loss": 0.3549,
+      "step": 2675
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.420772910118103,
+      "learning_rate": 1.2508684746324022e-05,
+      "loss": 0.3168,
+      "step": 2676
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.2173686027526855,
+      "learning_rate": 1.2451821280801912e-05,
+      "loss": 0.3887,
+      "step": 2677
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.013791561126709,
+      "learning_rate": 1.239508058252578e-05,
+      "loss": 0.416,
+      "step": 2678
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 1.1254581212997437,
+      "learning_rate": 1.2338462713388302e-05,
+      "loss": 0.3371,
+      "step": 2679
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 1.3764967918395996,
+      "learning_rate": 1.2281967735148045e-05,
+      "loss": 0.3323,
+      "step": 2680
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 1.2175017595291138,
+      "learning_rate": 1.2225595709429577e-05,
+      "loss": 0.4322,
+      "step": 2681
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.9489290118217468,
+      "learning_rate": 1.216934669772335e-05,
+      "loss": 0.1919,
+      "step": 2682
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 1.3265076875686646,
+      "learning_rate": 1.211322076138567e-05,
+      "loss": 0.436,
+      "step": 2683
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 1.6516838073730469,
+      "learning_rate": 1.2057217961638514e-05,
+      "loss": 0.7439,
+      "step": 2684
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 1.1286579370498657,
+      "learning_rate": 1.2001338359569583e-05,
+      "loss": 0.5829,
+      "step": 2685
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 1.074607491493225,
+      "learning_rate": 1.194558201613223e-05,
+      "loss": 0.3329,
+      "step": 2686
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.897049605846405,
+      "learning_rate": 1.1889948992145294e-05,
+      "loss": 0.221,
+      "step": 2687
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.8550513982772827,
+      "learning_rate": 1.1834439348293105e-05,
+      "loss": 0.2026,
+      "step": 2688
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.5562115907669067,
+      "learning_rate": 1.1779053145125492e-05,
+      "loss": 0.8372,
+      "step": 2689
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.3165887594223022,
+      "learning_rate": 1.1723790443057533e-05,
+      "loss": 0.5669,
+      "step": 2690
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.4699093103408813,
+      "learning_rate": 1.1668651302369632e-05,
+      "loss": 0.4754,
+      "step": 2691
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.1091874837875366,
+      "learning_rate": 1.1613635783207413e-05,
+      "loss": 0.4171,
+      "step": 2692
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.1977604627609253,
+      "learning_rate": 1.155874394558168e-05,
+      "loss": 0.388,
+      "step": 2693
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.3541066646575928,
+      "learning_rate": 1.150397584936827e-05,
+      "loss": 0.5098,
+      "step": 2694
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.4909467697143555,
+      "learning_rate": 1.1449331554308065e-05,
+      "loss": 0.6722,
+      "step": 2695
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.1262574195861816,
+      "learning_rate": 1.1394811120006967e-05,
+      "loss": 0.3256,
+      "step": 2696
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.2156447172164917,
+      "learning_rate": 1.1340414605935643e-05,
+      "loss": 0.4615,
+      "step": 2697
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.238400936126709,
+      "learning_rate": 1.1286142071429723e-05,
+      "loss": 0.285,
+      "step": 2698
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.21889328956604,
+      "learning_rate": 1.1231993575689506e-05,
+      "loss": 0.3242,
+      "step": 2699
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.147277593612671,
+      "learning_rate": 1.1177969177780051e-05,
+      "loss": 0.5024,
+      "step": 2700
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.8607897162437439,
+      "learning_rate": 1.112406893663101e-05,
+      "loss": 0.2267,
+      "step": 2701
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.7340608835220337,
+      "learning_rate": 1.1070292911036603e-05,
+      "loss": 0.6051,
+      "step": 2702
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.8612796068191528,
+      "learning_rate": 1.1016641159655574e-05,
+      "loss": 0.3697,
+      "step": 2703
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.6801596879959106,
+      "learning_rate": 1.0963113741011138e-05,
+      "loss": 0.1928,
+      "step": 2704
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.342772364616394,
+      "learning_rate": 1.0909710713490824e-05,
+      "loss": 0.4311,
+      "step": 2705
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.0663646459579468,
+      "learning_rate": 1.0856432135346511e-05,
+      "loss": 0.4227,
+      "step": 2706
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.1165860891342163,
+      "learning_rate": 1.080327806469436e-05,
+      "loss": 0.2888,
+      "step": 2707
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.1469182968139648,
+      "learning_rate": 1.0750248559514661e-05,
+      "loss": 0.4555,
+      "step": 2708
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.4056341648101807,
+      "learning_rate": 1.0697343677651863e-05,
+      "loss": 0.5328,
+      "step": 2709
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.9922555685043335,
+      "learning_rate": 1.0644563476814418e-05,
+      "loss": 0.2429,
+      "step": 2710
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.1998769044876099,
+      "learning_rate": 1.0591908014574907e-05,
+      "loss": 0.3518,
+      "step": 2711
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.6901216506958008,
+      "learning_rate": 1.0539377348369711e-05,
+      "loss": 0.7558,
+      "step": 2712
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.06063711643219,
+      "learning_rate": 1.0486971535499129e-05,
+      "loss": 0.3632,
+      "step": 2713
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.7559407949447632,
+      "learning_rate": 1.0434690633127326e-05,
+      "loss": 0.6834,
+      "step": 2714
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.5990970134735107,
+      "learning_rate": 1.038253469828214e-05,
+      "loss": 0.7073,
+      "step": 2715
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.2649844884872437,
+      "learning_rate": 1.0330503787855092e-05,
+      "loss": 0.4412,
+      "step": 2716
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.3318960666656494,
+      "learning_rate": 1.0278597958601422e-05,
+      "loss": 0.354,
+      "step": 2717
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.6753989458084106,
+      "learning_rate": 1.0226817267139824e-05,
+      "loss": 0.54,
+      "step": 2718
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.4355957508087158,
+      "learning_rate": 1.0175161769952548e-05,
+      "loss": 0.5194,
+      "step": 2719
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.2790511846542358,
+      "learning_rate": 1.0123631523385231e-05,
+      "loss": 0.6185,
+      "step": 2720
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.0246258974075317,
+      "learning_rate": 1.0072226583646981e-05,
+      "loss": 0.1929,
+      "step": 2721
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.2745766639709473,
+      "learning_rate": 1.0020947006810153e-05,
+      "loss": 0.3861,
+      "step": 2722
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.1661189794540405,
+      "learning_rate": 9.96979284881036e-06,
+      "loss": 0.3073,
+      "step": 2723
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.1576132774353027,
+      "learning_rate": 9.918764165446389e-06,
+      "loss": 0.4697,
+      "step": 2724
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.875788688659668,
+      "learning_rate": 9.867861012380255e-06,
+      "loss": 0.2326,
+      "step": 2725
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.3949896097183228,
+      "learning_rate": 9.817083445136932e-06,
+      "loss": 0.5425,
+      "step": 2726
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.2175787687301636,
+      "learning_rate": 9.766431519104447e-06,
+      "loss": 0.3765,
+      "step": 2727
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.4833611249923706,
+      "learning_rate": 9.71590528953381e-06,
+      "loss": 0.5869,
+      "step": 2728
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.0458709001541138,
+      "learning_rate": 9.665504811538884e-06,
+      "loss": 0.4591,
+      "step": 2729
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.4730372428894043,
+      "learning_rate": 9.615230140096367e-06,
+      "loss": 0.5924,
+      "step": 2730
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.1391056776046753,
+      "learning_rate": 9.565081330045716e-06,
+      "loss": 0.3789,
+      "step": 2731
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.3682113885879517,
+      "learning_rate": 9.515058436089158e-06,
+      "loss": 0.2783,
+      "step": 2732
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.0194765329360962,
+      "learning_rate": 9.465161512791504e-06,
+      "loss": 0.2986,
+      "step": 2733
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.2771401405334473,
+      "learning_rate": 9.415390614580175e-06,
+      "loss": 0.3452,
+      "step": 2734
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.1327400207519531,
+      "learning_rate": 9.365745795745145e-06,
+      "loss": 0.4536,
+      "step": 2735
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.046260952949524,
+      "learning_rate": 9.316227110438849e-06,
+      "loss": 0.4057,
+      "step": 2736
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.086682677268982,
+      "learning_rate": 9.266834612676134e-06,
+      "loss": 0.4568,
+      "step": 2737
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.1781115531921387,
+      "learning_rate": 9.217568356334166e-06,
+      "loss": 0.3688,
+      "step": 2738
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.3109606504440308,
+      "learning_rate": 9.168428395152486e-06,
+      "loss": 0.3833,
+      "step": 2739
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.3341201543807983,
+      "learning_rate": 9.119414782732829e-06,
+      "loss": 0.4116,
+      "step": 2740
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.1230562925338745,
+      "learning_rate": 9.070527572539067e-06,
+      "loss": 0.3619,
+      "step": 2741
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.5410712957382202,
+      "learning_rate": 9.021766817897292e-06,
+      "loss": 0.6043,
+      "step": 2742
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.9172453880310059,
+      "learning_rate": 8.97313257199557e-06,
+      "loss": 0.3465,
+      "step": 2743
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.0832079648971558,
+      "learning_rate": 8.92462488788398e-06,
+      "loss": 0.2514,
+      "step": 2744
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.1828948259353638,
+      "learning_rate": 8.8762438184746e-06,
+      "loss": 0.3498,
+      "step": 2745
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.2682899236679077,
+      "learning_rate": 8.827989416541394e-06,
+      "loss": 0.452,
+      "step": 2746
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.2498657703399658,
+      "learning_rate": 8.779861734720104e-06,
+      "loss": 0.5518,
+      "step": 2747
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.219550371170044,
+      "learning_rate": 8.73186082550828e-06,
+      "loss": 0.4601,
+      "step": 2748
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.1973583698272705,
+      "learning_rate": 8.683986741265151e-06,
+      "loss": 0.4653,
+      "step": 2749
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.2459843158721924,
+      "learning_rate": 8.63623953421168e-06,
+      "loss": 0.4605,
+      "step": 2750
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.2489614486694336,
+      "learning_rate": 8.588619256430383e-06,
+      "loss": 0.3642,
+      "step": 2751
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.1608028411865234,
+      "learning_rate": 8.541125959865295e-06,
+      "loss": 0.3983,
+      "step": 2752
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.3494408130645752,
+      "learning_rate": 8.493759696322031e-06,
+      "loss": 0.4337,
+      "step": 2753
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.4363130331039429,
+      "learning_rate": 8.44652051746754e-06,
+      "loss": 0.5642,
+      "step": 2754
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.4055970907211304,
+      "learning_rate": 8.399408474830187e-06,
+      "loss": 0.3069,
+      "step": 2755
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.1449294090270996,
+      "learning_rate": 8.352423619799682e-06,
+      "loss": 0.4492,
+      "step": 2756
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.1602027416229248,
+      "learning_rate": 8.305566003626974e-06,
+      "loss": 0.3871,
+      "step": 2757
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.279659390449524,
+      "learning_rate": 8.258835677424207e-06,
+      "loss": 0.4457,
+      "step": 2758
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.2665632963180542,
+      "learning_rate": 8.212232692164687e-06,
+      "loss": 0.4153,
+      "step": 2759
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.9248252511024475,
+      "learning_rate": 8.165757098682863e-06,
+      "loss": 0.1849,
+      "step": 2760
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.4571242332458496,
+      "learning_rate": 8.119408947674157e-06,
+      "loss": 0.5424,
+      "step": 2761
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.1283844709396362,
+      "learning_rate": 8.073188289694985e-06,
+      "loss": 0.4197,
+      "step": 2762
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.0835353136062622,
+      "learning_rate": 8.027095175162757e-06,
+      "loss": 0.3708,
+      "step": 2763
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.2077913284301758,
+      "learning_rate": 7.981129654355698e-06,
+      "loss": 0.3373,
+      "step": 2764
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.2850297689437866,
+      "learning_rate": 7.93529177741288e-06,
+      "loss": 0.4046,
+      "step": 2765
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.9630417823791504,
+      "learning_rate": 7.889581594334103e-06,
+      "loss": 0.302,
+      "step": 2766
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.1772398948669434,
+      "learning_rate": 7.84399915497995e-06,
+      "loss": 0.3867,
+      "step": 2767
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.3503518104553223,
+      "learning_rate": 7.798544509071596e-06,
+      "loss": 0.3149,
+      "step": 2768
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.1282005310058594,
+      "learning_rate": 7.753217706190862e-06,
+      "loss": 0.3866,
+      "step": 2769
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.9566932916641235,
+      "learning_rate": 7.708018795780079e-06,
+      "loss": 0.428,
+      "step": 2770
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.6949591636657715,
+      "learning_rate": 7.662947827142123e-06,
+      "loss": 1.0602,
+      "step": 2771
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.5378934144973755,
+      "learning_rate": 7.6180048494402656e-06,
+      "loss": 0.6878,
+      "step": 2772
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.2534940242767334,
+      "learning_rate": 7.5731899116981696e-06,
+      "loss": 0.3789,
+      "step": 2773
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.306345820426941,
+      "learning_rate": 7.528503062799882e-06,
+      "loss": 0.4367,
+      "step": 2774
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.062415361404419,
+      "learning_rate": 7.4839443514896775e-06,
+      "loss": 0.4448,
+      "step": 2775
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.4112834930419922,
+      "learning_rate": 7.439513826372091e-06,
+      "loss": 0.5658,
+      "step": 2776
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.1098510026931763,
+      "learning_rate": 7.3952115359117725e-06,
+      "loss": 0.3965,
+      "step": 2777
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.332319974899292,
+      "learning_rate": 7.351037528433594e-06,
+      "loss": 0.4484,
+      "step": 2778
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.1495813131332397,
+      "learning_rate": 7.306991852122413e-06,
+      "loss": 0.3646,
+      "step": 2779
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.1314551830291748,
+      "learning_rate": 7.263074555023133e-06,
+      "loss": 0.3927,
+      "step": 2780
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.2844690084457397,
+      "learning_rate": 7.219285685040627e-06,
+      "loss": 0.2945,
+      "step": 2781
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.8933795690536499,
+      "learning_rate": 7.175625289939688e-06,
+      "loss": 0.2147,
+      "step": 2782
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 1.2195204496383667,
+      "learning_rate": 7.132093417344929e-06,
+      "loss": 0.4342,
+      "step": 2783
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 1.2162507772445679,
+      "learning_rate": 7.0886901147408226e-06,
+      "loss": 0.3175,
+      "step": 2784
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.8239806294441223,
+      "learning_rate": 7.045415429471569e-06,
+      "loss": 0.2895,
+      "step": 2785
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 1.3585073947906494,
+      "learning_rate": 7.002269408741074e-06,
+      "loss": 0.4936,
+      "step": 2786
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 1.044557809829712,
+      "learning_rate": 6.959252099612873e-06,
+      "loss": 0.3316,
+      "step": 2787
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.9965437054634094,
+      "learning_rate": 6.9163635490101654e-06,
+      "loss": 0.4026,
+      "step": 2788
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 1.2901641130447388,
+      "learning_rate": 6.873603803715642e-06,
+      "loss": 0.4783,
+      "step": 2789
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 1.0988811254501343,
+      "learning_rate": 6.830972910371519e-06,
+      "loss": 0.4749,
+      "step": 2790
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 1.1592463254928589,
+      "learning_rate": 6.788470915479475e-06,
+      "loss": 0.3396,
+      "step": 2791
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 1.192961573600769,
+      "learning_rate": 6.7460978654005634e-06,
+      "loss": 0.2703,
+      "step": 2792
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.6996821165084839,
+      "learning_rate": 6.703853806355187e-06,
+      "loss": 0.6261,
+      "step": 2793
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.1305897235870361,
+      "learning_rate": 6.661738784423041e-06,
+      "loss": 0.4905,
+      "step": 2794
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.7390271425247192,
+      "learning_rate": 6.6197528455431e-06,
+      "loss": 0.6344,
+      "step": 2795
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.9697789549827576,
+      "learning_rate": 6.577896035513509e-06,
+      "loss": 0.3131,
+      "step": 2796
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.1802033185958862,
+      "learning_rate": 6.536168399991535e-06,
+      "loss": 0.3639,
+      "step": 2797
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.0719521045684814,
+      "learning_rate": 6.494569984493575e-06,
+      "loss": 0.2925,
+      "step": 2798
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.4077153205871582,
+      "learning_rate": 6.453100834395098e-06,
+      "loss": 0.4654,
+      "step": 2799
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.2906742095947266,
+      "learning_rate": 6.411760994930516e-06,
+      "loss": 0.4707,
+      "step": 2800
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.1949652433395386,
+      "learning_rate": 6.370550511193202e-06,
+      "loss": 0.4016,
+      "step": 2801
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 1.407058835029602,
+      "learning_rate": 6.329469428135462e-06,
+      "loss": 0.5245,
+      "step": 2802
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.9813088178634644,
+      "learning_rate": 6.288517790568424e-06,
+      "loss": 0.3017,
+      "step": 2803
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.1702017784118652,
+      "learning_rate": 6.247695643161993e-06,
+      "loss": 0.4415,
+      "step": 2804
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.032965064048767,
+      "learning_rate": 6.207003030444914e-06,
+      "loss": 0.3122,
+      "step": 2805
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.9007138013839722,
+      "learning_rate": 6.166439996804538e-06,
+      "loss": 0.3185,
+      "step": 2806
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.3804324865341187,
+      "learning_rate": 6.126006586486918e-06,
+      "loss": 0.4441,
+      "step": 2807
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.2858846187591553,
+      "learning_rate": 6.085702843596691e-06,
+      "loss": 0.4465,
+      "step": 2808
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.1197959184646606,
+      "learning_rate": 6.045528812097093e-06,
+      "loss": 0.3603,
+      "step": 2809
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.3286681175231934,
+      "learning_rate": 6.00548453580986e-06,
+      "loss": 0.5375,
+      "step": 2810
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.308885097503662,
+      "learning_rate": 5.965570058415137e-06,
+      "loss": 0.5407,
+      "step": 2811
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.5666403770446777,
+      "learning_rate": 5.925785423451569e-06,
+      "loss": 0.482,
+      "step": 2812
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.2590450048446655,
+      "learning_rate": 5.886130674316106e-06,
+      "loss": 0.5029,
+      "step": 2813
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.8867388963699341,
+      "learning_rate": 5.846605854264039e-06,
+      "loss": 0.2502,
+      "step": 2814
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.2063246965408325,
+      "learning_rate": 5.807211006408908e-06,
+      "loss": 0.3824,
+      "step": 2815
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.8002722263336182,
+      "learning_rate": 5.767946173722574e-06,
+      "loss": 0.5485,
+      "step": 2816
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.7626540660858154,
+      "learning_rate": 5.728811399034958e-06,
+      "loss": 0.6995,
+      "step": 2817
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.8790903091430664,
+      "learning_rate": 5.689806725034191e-06,
+      "loss": 0.2667,
+      "step": 2818
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.3238037824630737,
+      "learning_rate": 5.6509321942664334e-06,
+      "loss": 0.5077,
+      "step": 2819
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.1437560319900513,
+      "learning_rate": 5.612187849135947e-06,
+      "loss": 0.3441,
+      "step": 2820
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.3718996047973633,
+      "learning_rate": 5.573573731904951e-06,
+      "loss": 0.5211,
+      "step": 2821
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.0614670515060425,
+      "learning_rate": 5.5350898846935894e-06,
+      "loss": 0.2664,
+      "step": 2822
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.1064701080322266,
+      "learning_rate": 5.496736349479989e-06,
+      "loss": 0.4569,
+      "step": 2823
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 1.287601113319397,
+      "learning_rate": 5.458513168100046e-06,
+      "loss": 0.5515,
+      "step": 2824
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 1.123247504234314,
+      "learning_rate": 5.420420382247521e-06,
+      "loss": 0.3065,
+      "step": 2825
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.889412522315979,
+      "learning_rate": 5.382458033473897e-06,
+      "loss": 0.2599,
+      "step": 2826
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 1.0928308963775635,
+      "learning_rate": 5.344626163188412e-06,
+      "loss": 0.2895,
+      "step": 2827
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 1.2472971677780151,
+      "learning_rate": 5.306924812657987e-06,
+      "loss": 0.3615,
+      "step": 2828
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 1.1193463802337646,
+      "learning_rate": 5.269354023007114e-06,
+      "loss": 0.438,
+      "step": 2829
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.9217707514762878,
+      "learning_rate": 5.231913835217941e-06,
+      "loss": 0.3097,
+      "step": 2830
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 1.294617772102356,
+      "learning_rate": 5.194604290130106e-06,
+      "loss": 0.5423,
+      "step": 2831
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 1.2628179788589478,
+      "learning_rate": 5.157425428440762e-06,
+      "loss": 0.4024,
+      "step": 2832
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.9742478728294373,
+      "learning_rate": 5.120377290704512e-06,
+      "loss": 0.3548,
+      "step": 2833
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 1.4343669414520264,
+      "learning_rate": 5.083459917333361e-06,
+      "loss": 0.319,
+      "step": 2834
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 1.464012861251831,
+      "learning_rate": 5.046673348596681e-06,
+      "loss": 0.5287,
+      "step": 2835
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 1.1931774616241455,
+      "learning_rate": 5.010017624621152e-06,
+      "loss": 0.3692,
+      "step": 2836
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 1.4124280214309692,
+      "learning_rate": 4.973492785390763e-06,
+      "loss": 0.4301,
+      "step": 2837
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.9147828817367554,
+      "learning_rate": 4.937098870746671e-06,
+      "loss": 0.2231,
+      "step": 2838
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 1.17789626121521,
+      "learning_rate": 4.900835920387287e-06,
+      "loss": 0.3442,
+      "step": 2839
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 1.1786454916000366,
+      "learning_rate": 4.864703973868123e-06,
+      "loss": 0.3725,
+      "step": 2840
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.9402766227722168,
+      "learning_rate": 4.82870307060182e-06,
+      "loss": 0.2451,
+      "step": 2841
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 1.3032366037368774,
+      "learning_rate": 4.792833249858075e-06,
+      "loss": 0.4427,
+      "step": 2842
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 1.2752867937088013,
+      "learning_rate": 4.757094550763549e-06,
+      "loss": 0.5675,
+      "step": 2843
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.0933231115341187,
+      "learning_rate": 4.721487012301962e-06,
+      "loss": 0.3532,
+      "step": 2844
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.2620882987976074,
+      "learning_rate": 4.686010673313909e-06,
+      "loss": 0.4495,
+      "step": 2845
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.1065340042114258,
+      "learning_rate": 4.650665572496868e-06,
+      "loss": 0.464,
+      "step": 2846
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.9144507050514221,
+      "learning_rate": 4.615451748405164e-06,
+      "loss": 0.2445,
+      "step": 2847
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.1331830024719238,
+      "learning_rate": 4.580369239449989e-06,
+      "loss": 0.3741,
+      "step": 2848
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.3149724006652832,
+      "learning_rate": 4.545418083899216e-06,
+      "loss": 0.4282,
+      "step": 2849
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.846543550491333,
+      "learning_rate": 4.510598319877463e-06,
+      "loss": 0.5719,
+      "step": 2850
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.2957957983016968,
+      "learning_rate": 4.475909985366066e-06,
+      "loss": 0.5492,
+      "step": 2851
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.146454095840454,
+      "learning_rate": 4.441353118202942e-06,
+      "loss": 0.4014,
+      "step": 2852
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.318638563156128,
+      "learning_rate": 4.406927756082618e-06,
+      "loss": 0.5174,
+      "step": 2853
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.845371425151825,
+      "learning_rate": 4.372633936556195e-06,
+      "loss": 0.2613,
+      "step": 2854
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.4363237619400024,
+      "learning_rate": 4.338471697031304e-06,
+      "loss": 0.4672,
+      "step": 2855
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.9613076448440552,
+      "learning_rate": 4.304441074771986e-06,
+      "loss": 0.3455,
+      "step": 2856
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.239383339881897,
+      "learning_rate": 4.270542106898756e-06,
+      "loss": 0.4925,
+      "step": 2857
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.1034066677093506,
+      "learning_rate": 4.236774830388534e-06,
+      "loss": 0.3413,
+      "step": 2858
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.3597811460494995,
+      "learning_rate": 4.2031392820745665e-06,
+      "loss": 0.347,
+      "step": 2859
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.3148061037063599,
+      "learning_rate": 4.169635498646407e-06,
+      "loss": 0.4692,
+      "step": 2860
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.3651525974273682,
+      "learning_rate": 4.136263516649891e-06,
+      "loss": 0.5351,
+      "step": 2861
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.1956440210342407,
+      "learning_rate": 4.103023372487094e-06,
+      "loss": 0.4367,
+      "step": 2862
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.114611268043518,
+      "learning_rate": 4.069915102416291e-06,
+      "loss": 0.4394,
+      "step": 2863
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.038771390914917,
+      "learning_rate": 4.036938742551871e-06,
+      "loss": 0.2957,
+      "step": 2864
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 1.0876121520996094,
+      "learning_rate": 4.004094328864338e-06,
+      "loss": 0.3007,
+      "step": 2865
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 1.1242945194244385,
+      "learning_rate": 3.971381897180326e-06,
+      "loss": 0.2948,
+      "step": 2866
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 1.0400104522705078,
+      "learning_rate": 3.938801483182458e-06,
+      "loss": 0.4133,
+      "step": 2867
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 1.0308600664138794,
+      "learning_rate": 3.906353122409334e-06,
+      "loss": 0.2428,
+      "step": 2868
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 1.1233997344970703,
+      "learning_rate": 3.8740368502555856e-06,
+      "loss": 0.3727,
+      "step": 2869
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 1.2486021518707275,
+      "learning_rate": 3.841852701971668e-06,
+      "loss": 0.3912,
+      "step": 2870
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 1.3520262241363525,
+      "learning_rate": 3.8098007126639857e-06,
+      "loss": 0.6819,
+      "step": 2871
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 1.4625575542449951,
+      "learning_rate": 3.7778809172947514e-06,
+      "loss": 0.6033,
+      "step": 2872
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.9293382167816162,
+      "learning_rate": 3.7460933506820032e-06,
+      "loss": 0.3032,
+      "step": 2873
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 1.2324880361557007,
+      "learning_rate": 3.7144380474995326e-06,
+      "loss": 0.3844,
+      "step": 2874
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.906033992767334,
+      "learning_rate": 3.682915042276816e-06,
+      "loss": 0.2879,
+      "step": 2875
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.577123999595642,
+      "learning_rate": 3.6515243693991256e-06,
+      "loss": 0.4891,
+      "step": 2876
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.0596559047698975,
+      "learning_rate": 3.6202660631072802e-06,
+      "loss": 0.2943,
+      "step": 2877
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.8984016180038452,
+      "learning_rate": 3.5891401574977555e-06,
+      "loss": 0.1917,
+      "step": 2878
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.0884068012237549,
+      "learning_rate": 3.5581466865226287e-06,
+      "loss": 0.3737,
+      "step": 2879
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.3667714595794678,
+      "learning_rate": 3.5272856839894814e-06,
+      "loss": 0.5069,
+      "step": 2880
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.6226465702056885,
+      "learning_rate": 3.4965571835614e-06,
+      "loss": 0.6759,
+      "step": 2881
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.6317212581634521,
+      "learning_rate": 3.4659612187569604e-06,
+      "loss": 0.8197,
+      "step": 2882
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.2995193004608154,
+      "learning_rate": 3.435497822950148e-06,
+      "loss": 0.4017,
+      "step": 2883
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.2911146879196167,
+      "learning_rate": 3.405167029370368e-06,
+      "loss": 0.522,
+      "step": 2884
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 1.1538434028625488,
+      "learning_rate": 3.3749688711023492e-06,
+      "loss": 0.2955,
+      "step": 2885
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.2104440927505493,
+      "learning_rate": 3.344903381086159e-06,
+      "loss": 0.3577,
+      "step": 2886
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.281571388244629,
+      "learning_rate": 3.3149705921171752e-06,
+      "loss": 0.4891,
+      "step": 2887
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.0648771524429321,
+      "learning_rate": 3.2851705368459883e-06,
+      "loss": 0.3683,
+      "step": 2888
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.3280951976776123,
+      "learning_rate": 3.2555032477784017e-06,
+      "loss": 0.4345,
+      "step": 2889
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.4385279417037964,
+      "learning_rate": 3.2259687572754464e-06,
+      "loss": 0.7291,
+      "step": 2890
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.8171462416648865,
+      "learning_rate": 3.1965670975532544e-06,
+      "loss": 0.2052,
+      "step": 2891
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.12943434715271,
+      "learning_rate": 3.167298300683047e-06,
+      "loss": 0.4011,
+      "step": 2892
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.2015761137008667,
+      "learning_rate": 3.138162398591174e-06,
+      "loss": 0.4803,
+      "step": 2893
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.9145457148551941,
+      "learning_rate": 3.109159423058991e-06,
+      "loss": 0.2703,
+      "step": 2894
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.4654762744903564,
+      "learning_rate": 3.080289405722886e-06,
+      "loss": 0.5258,
+      "step": 2895
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.0232746601104736,
+      "learning_rate": 3.0515523780741403e-06,
+      "loss": 0.2846,
+      "step": 2896
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.163627028465271,
+      "learning_rate": 3.0229483714590835e-06,
+      "loss": 0.2929,
+      "step": 2897
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.0843936204910278,
+      "learning_rate": 2.994477417078867e-06,
+      "loss": 0.3374,
+      "step": 2898
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.0600237846374512,
+      "learning_rate": 2.9661395459895103e-06,
+      "loss": 0.3439,
+      "step": 2899
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.2277902364730835,
+      "learning_rate": 2.937934789101912e-06,
+      "loss": 0.4068,
+      "step": 2900
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.0997445583343506,
+      "learning_rate": 2.9098631771817403e-06,
+      "loss": 0.3053,
+      "step": 2901
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.4788862466812134,
+      "learning_rate": 2.8819247408494316e-06,
+      "loss": 0.5985,
+      "step": 2902
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.0369575023651123,
+      "learning_rate": 2.854119510580136e-06,
+      "loss": 0.2808,
+      "step": 2903
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.1584135293960571,
+      "learning_rate": 2.826447516703745e-06,
+      "loss": 0.4053,
+      "step": 2904
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 1.258483648300171,
+      "learning_rate": 2.7989087894047945e-06,
+      "loss": 0.3445,
+      "step": 2905
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.4125854969024658,
+      "learning_rate": 2.7715033587224214e-06,
+      "loss": 0.4582,
+      "step": 2906
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.1342895030975342,
+      "learning_rate": 2.744231254550436e-06,
+      "loss": 0.3992,
+      "step": 2907
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.0188727378845215,
+      "learning_rate": 2.7170925066371673e-06,
+      "loss": 0.3321,
+      "step": 2908
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.8607746362686157,
+      "learning_rate": 2.6900871445854626e-06,
+      "loss": 0.3204,
+      "step": 2909
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.1790703535079956,
+      "learning_rate": 2.663215197852717e-06,
+      "loss": 0.3829,
+      "step": 2910
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.5910663604736328,
+      "learning_rate": 2.636476695750775e-06,
+      "loss": 0.4193,
+      "step": 2911
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.5405675172805786,
+      "learning_rate": 2.6098716674459176e-06,
+      "loss": 0.5446,
+      "step": 2912
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.7244126200675964,
+      "learning_rate": 2.583400141958847e-06,
+      "loss": 0.2043,
+      "step": 2913
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.202313780784607,
+      "learning_rate": 2.5570621481646043e-06,
+      "loss": 0.3616,
+      "step": 2914
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.1371735334396362,
+      "learning_rate": 2.5308577147926394e-06,
+      "loss": 0.3795,
+      "step": 2915
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.038470983505249,
+      "learning_rate": 2.504786870426656e-06,
+      "loss": 0.2639,
+      "step": 2916
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.292428970336914,
+      "learning_rate": 2.4788496435046437e-06,
+      "loss": 0.4548,
+      "step": 2917
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.237728238105774,
+      "learning_rate": 2.453046062318887e-06,
+      "loss": 0.3921,
+      "step": 2918
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.202520489692688,
+      "learning_rate": 2.4273761550158563e-06,
+      "loss": 0.5007,
+      "step": 2919
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.2154775857925415,
+      "learning_rate": 2.4018399495961964e-06,
+      "loss": 0.4211,
+      "step": 2920
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.4609144926071167,
+      "learning_rate": 2.37643747391475e-06,
+      "loss": 0.6236,
+      "step": 2921
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.2069356441497803,
+      "learning_rate": 2.3511687556804634e-06,
+      "loss": 0.3363,
+      "step": 2922
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.7999249696731567,
+      "learning_rate": 2.326033822456386e-06,
+      "loss": 0.2567,
+      "step": 2923
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.2546871900558472,
+      "learning_rate": 2.3010327016596278e-06,
+      "loss": 0.3224,
+      "step": 2924
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.238844394683838,
+      "learning_rate": 2.2761654205613603e-06,
+      "loss": 0.4932,
+      "step": 2925
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 1.4584134817123413,
+      "learning_rate": 2.2514320062867325e-06,
+      "loss": 0.6393,
+      "step": 2926
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.3758424520492554,
+      "learning_rate": 2.2268324858148724e-06,
+      "loss": 0.5076,
+      "step": 2927
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.5602376461029053,
+      "learning_rate": 2.2023668859788985e-06,
+      "loss": 0.5522,
+      "step": 2928
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.5447279214859009,
+      "learning_rate": 2.178035233465797e-06,
+      "loss": 0.6943,
+      "step": 2929
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.119056224822998,
+      "learning_rate": 2.1538375548164903e-06,
+      "loss": 0.4048,
+      "step": 2930
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.579376220703125,
+      "learning_rate": 2.129773876425711e-06,
+      "loss": 0.5035,
+      "step": 2931
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.2969238758087158,
+      "learning_rate": 2.105844224542061e-06,
+      "loss": 0.4769,
+      "step": 2932
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.8729587197303772,
+      "learning_rate": 2.0820486252679655e-06,
+      "loss": 0.2227,
+      "step": 2933
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.4230064153671265,
+      "learning_rate": 2.058387104559578e-06,
+      "loss": 0.468,
+      "step": 2934
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.446077585220337,
+      "learning_rate": 2.034859688226823e-06,
+      "loss": 0.3544,
+      "step": 2935
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.2802209854125977,
+      "learning_rate": 2.0114664019333512e-06,
+      "loss": 0.4822,
+      "step": 2936
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.3602778911590576,
+      "learning_rate": 1.9882072711964876e-06,
+      "loss": 0.4064,
+      "step": 2937
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.9845119714736938,
+      "learning_rate": 1.9650823213872014e-06,
+      "loss": 0.6687,
+      "step": 2938
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.3791077136993408,
+      "learning_rate": 1.942091577730176e-06,
+      "loss": 0.4893,
+      "step": 2939
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.945283055305481,
+      "learning_rate": 1.919235065303629e-06,
+      "loss": 0.2887,
+      "step": 2940
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.4401484727859497,
+      "learning_rate": 1.8965128090393802e-06,
+      "loss": 0.3711,
+      "step": 2941
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.3931560516357422,
+      "learning_rate": 1.8739248337227844e-06,
+      "loss": 0.5732,
+      "step": 2942
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.1365431547164917,
+      "learning_rate": 1.851471163992785e-06,
+      "loss": 0.3554,
+      "step": 2943
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.4375975131988525,
+      "learning_rate": 1.8291518243417627e-06,
+      "loss": 0.4748,
+      "step": 2944
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.0275496244430542,
+      "learning_rate": 1.8069668391155759e-06,
+      "loss": 0.2599,
+      "step": 2945
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.8617614507675171,
+      "learning_rate": 1.7849162325135754e-06,
+      "loss": 0.1928,
+      "step": 2946
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.23322594165802,
+      "learning_rate": 1.7630000285884795e-06,
+      "loss": 0.305,
+      "step": 2947
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.6121059656143188,
+      "learning_rate": 1.7412182512464292e-06,
+      "loss": 0.4491,
+      "step": 2948
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.5377715826034546,
+      "learning_rate": 1.7195709242469465e-06,
+      "loss": 0.49,
+      "step": 2949
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.2269455194473267,
+      "learning_rate": 1.6980580712028514e-06,
+      "loss": 0.3071,
+      "step": 2950
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.3769416809082031,
+      "learning_rate": 1.6766797155803315e-06,
+      "loss": 0.4086,
+      "step": 2951
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.137343168258667,
+      "learning_rate": 1.6554358806988024e-06,
+      "loss": 0.3242,
+      "step": 2952
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.8603381514549255,
+      "learning_rate": 1.6343265897310055e-06,
+      "loss": 0.2855,
+      "step": 2953
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.6535147428512573,
+      "learning_rate": 1.6133518657028972e-06,
+      "loss": 0.7309,
+      "step": 2954
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.2747342586517334,
+      "learning_rate": 1.5925117314936484e-06,
+      "loss": 0.4667,
+      "step": 2955
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.1546356678009033,
+      "learning_rate": 1.5718062098356023e-06,
+      "loss": 0.4735,
+      "step": 2956
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 1.2965714931488037,
+      "learning_rate": 1.5512353233143178e-06,
+      "loss": 0.3944,
+      "step": 2957
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.7952131032943726,
+      "learning_rate": 1.530799094368443e-06,
+      "loss": 0.5763,
+      "step": 2958
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.420607566833496,
+      "learning_rate": 1.5104975452897573e-06,
+      "loss": 0.4716,
+      "step": 2959
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.3332853317260742,
+      "learning_rate": 1.490330698223158e-06,
+      "loss": 0.4557,
+      "step": 2960
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.3416107892990112,
+      "learning_rate": 1.4702985751665482e-06,
+      "loss": 0.4973,
+      "step": 2961
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.9640762209892273,
+      "learning_rate": 1.4504011979709491e-06,
+      "loss": 0.3745,
+      "step": 2962
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.2113425731658936,
+      "learning_rate": 1.4306385883403329e-06,
+      "loss": 0.4207,
+      "step": 2963
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.397342562675476,
+      "learning_rate": 1.411010767831733e-06,
+      "loss": 0.5736,
+      "step": 2964
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.103853464126587,
+      "learning_rate": 1.391517757855107e-06,
+      "loss": 0.3447,
+      "step": 2965
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.2431985139846802,
+      "learning_rate": 1.3721595796733626e-06,
+      "loss": 0.4443,
+      "step": 2966
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.4864449501037598,
+      "learning_rate": 1.3529362544023733e-06,
+      "loss": 0.314,
+      "step": 2967
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.0626156330108643,
+      "learning_rate": 1.3338478030108792e-06,
+      "loss": 0.4855,
+      "step": 2968
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.3877254724502563,
+      "learning_rate": 1.3148942463205166e-06,
+      "loss": 0.4789,
+      "step": 2969
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.9821919202804565,
+      "learning_rate": 1.2960756050057476e-06,
+      "loss": 0.2581,
+      "step": 2970
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.2562929391860962,
+      "learning_rate": 1.27739189959393e-06,
+      "loss": 0.5034,
+      "step": 2971
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.0050729513168335,
+      "learning_rate": 1.2588431504651915e-06,
+      "loss": 0.3125,
+      "step": 2972
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.1035274267196655,
+      "learning_rate": 1.240429377852431e-06,
+      "loss": 0.4549,
+      "step": 2973
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.6296160221099854,
+      "learning_rate": 1.2221506018413597e-06,
+      "loss": 0.7023,
+      "step": 2974
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.1510668992996216,
+      "learning_rate": 1.2040068423704171e-06,
+      "loss": 0.3249,
+      "step": 2975
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.4502555131912231,
+      "learning_rate": 1.1859981192307718e-06,
+      "loss": 0.41,
+      "step": 2976
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.4568229913711548,
+      "learning_rate": 1.1681244520662798e-06,
+      "loss": 0.6871,
+      "step": 2977
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 1.3338775634765625,
+      "learning_rate": 1.1503858603734846e-06,
+      "loss": 0.5211,
+      "step": 2978
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.2410080432891846,
+      "learning_rate": 1.1327823635016026e-06,
+      "loss": 0.3128,
+      "step": 2979
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.0830005407333374,
+      "learning_rate": 1.1153139806524964e-06,
+      "loss": 0.272,
+      "step": 2980
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.1216962337493896,
+      "learning_rate": 1.0979807308805906e-06,
+      "loss": 0.4014,
+      "step": 2981
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.6853399276733398,
+      "learning_rate": 1.0807826330929694e-06,
+      "loss": 0.6638,
+      "step": 2982
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.0854101181030273,
+      "learning_rate": 1.0637197060492793e-06,
+      "loss": 0.2654,
+      "step": 2983
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.1459499597549438,
+      "learning_rate": 1.0467919683617017e-06,
+      "loss": 0.3811,
+      "step": 2984
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.1812524795532227,
+      "learning_rate": 1.0299994384949802e-06,
+      "loss": 0.3847,
+      "step": 2985
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.2767400741577148,
+      "learning_rate": 1.0133421347663374e-06,
+      "loss": 0.5387,
+      "step": 2986
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.6375187635421753,
+      "learning_rate": 9.968200753455308e-07,
+      "loss": 0.6675,
+      "step": 2987
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 1.0926377773284912,
+      "learning_rate": 9.804332782547693e-07,
+      "loss": 0.3265,
+      "step": 2988
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 1.2261865139007568,
+      "learning_rate": 9.641817613687404e-07,
+      "loss": 0.4266,
+      "step": 2989
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 1.200217604637146,
+      "learning_rate": 9.480655424145418e-07,
+      "loss": 0.3329,
+      "step": 2990
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 1.0978448390960693,
+      "learning_rate": 9.320846389717086e-07,
+      "loss": 0.3821,
+      "step": 2991
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 1.2996379137039185,
+      "learning_rate": 9.162390684721577e-07,
+      "loss": 0.428,
+      "step": 2992
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 1.2850451469421387,
+      "learning_rate": 9.005288482002022e-07,
+      "loss": 0.4459,
+      "step": 2993
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.824524462223053,
+      "learning_rate": 8.849539952925089e-07,
+      "loss": 0.2747,
+      "step": 2994
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 1.188064694404602,
+      "learning_rate": 8.695145267380717e-07,
+      "loss": 0.4026,
+      "step": 2995
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 1.0427355766296387,
+      "learning_rate": 8.54210459378238e-07,
+      "loss": 0.4607,
+      "step": 2996
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.7268507480621338,
+      "learning_rate": 8.390418099066266e-07,
+      "loss": 0.2667,
+      "step": 2997
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 1.2438222169876099,
+      "learning_rate": 8.240085948691689e-07,
+      "loss": 0.3948,
+      "step": 2998
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 1.2525782585144043,
+      "learning_rate": 8.091108306640532e-07,
+      "loss": 0.5778,
+      "step": 2999
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.011322259902954,
+      "learning_rate": 7.943485335417249e-07,
+      "loss": 0.4481,
+      "step": 3000
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.1529054641723633,
+      "learning_rate": 7.797217196048451e-07,
+      "loss": 0.3791,
+      "step": 3001
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.1260803937911987,
+      "learning_rate": 7.652304048083175e-07,
+      "loss": 0.4783,
+      "step": 3002
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.1295585632324219,
+      "learning_rate": 7.508746049592063e-07,
+      "loss": 0.3297,
+      "step": 3003
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.0275527238845825,
+      "learning_rate": 7.366543357168048e-07,
+      "loss": 0.2958,
+      "step": 3004
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.4418631792068481,
+      "learning_rate": 7.225696125925246e-07,
+      "loss": 0.5509,
+      "step": 3005
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.5385364294052124,
+      "learning_rate": 7.086204509499511e-07,
+      "loss": 0.3432,
+      "step": 3006
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.9590983986854553,
+      "learning_rate": 6.948068660048019e-07,
+      "loss": 0.1927,
+      "step": 3007
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.226137399673462,
+      "learning_rate": 6.81128872824871e-07,
+      "loss": 0.5011,
+      "step": 3008
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.3301578760147095,
+      "learning_rate": 6.67586486330099e-07,
+      "loss": 0.5187,
+      "step": 3009
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.4750556945800781,
+      "learning_rate": 6.541797212924888e-07,
+      "loss": 0.4413,
+      "step": 3010
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.5725219249725342,
+      "learning_rate": 6.409085923360785e-07,
+      "loss": 0.4741,
+      "step": 3011
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.067269206047058,
+      "learning_rate": 6.27773113936983e-07,
+      "loss": 0.2581,
+      "step": 3012
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.217340350151062,
+      "learning_rate": 6.147733004233796e-07,
+      "loss": 0.3935,
+      "step": 3013
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.3149571418762207,
+      "learning_rate": 6.019091659753978e-07,
+      "loss": 0.4206,
+      "step": 3014
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.2064661979675293,
+      "learning_rate": 5.891807246251879e-07,
+      "loss": 0.4474,
+      "step": 3015
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.103737473487854,
+      "learning_rate": 5.765879902569493e-07,
+      "loss": 0.3483,
+      "step": 3016
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.2660019397735596,
+      "learning_rate": 5.641309766067499e-07,
+      "loss": 0.3404,
+      "step": 3017
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 1.3228251934051514,
+      "learning_rate": 5.518096972627063e-07,
+      "loss": 0.414,
+      "step": 3018
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.9580221772193909,
+      "learning_rate": 5.396241656648037e-07,
+      "loss": 0.2349,
+      "step": 3019
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 1.2516742944717407,
+      "learning_rate": 5.275743951050071e-07,
+      "loss": 0.5049,
+      "step": 3020
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.7590013742446899,
+      "learning_rate": 5.156603987271635e-07,
+      "loss": 0.1885,
+      "step": 3021
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 1.0176297426223755,
+      "learning_rate": 5.038821895270302e-07,
+      "loss": 0.462,
+      "step": 3022
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 1.2553941011428833,
+      "learning_rate": 4.922397803522466e-07,
+      "loss": 0.5014,
+      "step": 3023
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 1.3685301542282104,
+      "learning_rate": 4.807331839023349e-07,
+      "loss": 0.6243,
+      "step": 3024
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 1.0733470916748047,
+      "learning_rate": 4.6936241272863e-07,
+      "loss": 0.425,
+      "step": 3025
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 1.3270857334136963,
+      "learning_rate": 4.581274792343493e-07,
+      "loss": 0.551,
+      "step": 3026
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 1.0223822593688965,
+      "learning_rate": 4.470283956745508e-07,
+      "loss": 0.3436,
+      "step": 3027
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 1.0676158666610718,
+      "learning_rate": 4.360651741560501e-07,
+      "loss": 0.3039,
+      "step": 3028
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.9071052670478821,
+      "learning_rate": 4.2523782663753117e-07,
+      "loss": 0.2872,
+      "step": 3029
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 1.0709900856018066,
+      "learning_rate": 4.145463649294079e-07,
+      "loss": 0.3877,
+      "step": 3030
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.3956645727157593,
+      "learning_rate": 4.039908006939069e-07,
+      "loss": 0.4614,
+      "step": 3031
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.042420506477356,
+      "learning_rate": 3.935711454450125e-07,
+      "loss": 0.3412,
+      "step": 3032
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.9218428134918213,
+      "learning_rate": 3.832874105484524e-07,
+      "loss": 0.1879,
+      "step": 3033
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.0542484521865845,
+      "learning_rate": 3.7313960722169803e-07,
+      "loss": 0.3305,
+      "step": 3034
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.9976268410682678,
+      "learning_rate": 3.6312774653395054e-07,
+      "loss": 0.2927,
+      "step": 3035
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.625490665435791,
+      "learning_rate": 3.5325183940611293e-07,
+      "loss": 0.452,
+      "step": 3036
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.3954522609710693,
+      "learning_rate": 3.435118966107903e-07,
+      "loss": 0.4176,
+      "step": 3037
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.1922634840011597,
+      "learning_rate": 3.3390792877230347e-07,
+      "loss": 0.3824,
+      "step": 3038
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.2640684843063354,
+      "learning_rate": 3.2443994636663363e-07,
+      "loss": 0.3669,
+      "step": 3039
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 1.0344746112823486,
+      "learning_rate": 3.151079597214085e-07,
+      "loss": 0.3111,
+      "step": 3040
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.2467656135559082,
+      "learning_rate": 3.0591197901595757e-07,
+      "loss": 0.6657,
+      "step": 3041
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.0825926065444946,
+      "learning_rate": 2.968520142812292e-07,
+      "loss": 0.3323,
+      "step": 3042
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.1362985372543335,
+      "learning_rate": 2.8792807539979036e-07,
+      "loss": 0.4313,
+      "step": 3043
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.332108974456787,
+      "learning_rate": 2.7914017210585454e-07,
+      "loss": 0.6444,
+      "step": 3044
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.0163555145263672,
+      "learning_rate": 2.7048831398525375e-07,
+      "loss": 0.2917,
+      "step": 3045
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.1551223993301392,
+      "learning_rate": 2.6197251047541117e-07,
+      "loss": 0.3461,
+      "step": 3046
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.4475940465927124,
+      "learning_rate": 2.535927708653268e-07,
+      "loss": 0.5594,
+      "step": 3047
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.1431162357330322,
+      "learning_rate": 2.4534910429560575e-07,
+      "loss": 0.3737,
+      "step": 3048
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.167028546333313,
+      "learning_rate": 2.372415197584299e-07,
+      "loss": 0.327,
+      "step": 3049
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.110597014427185,
+      "learning_rate": 2.292700260975028e-07,
+      "loss": 0.315,
+      "step": 3050
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 1.1866278648376465,
+      "learning_rate": 2.2143463200813285e-07,
+      "loss": 0.3174,
+      "step": 3051
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.0540951490402222,
+      "learning_rate": 2.1373534603713608e-07,
+      "loss": 0.4002,
+      "step": 3052
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.6657365560531616,
+      "learning_rate": 2.0617217658287779e-07,
+      "loss": 0.6559,
+      "step": 3053
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.9559653401374817,
+      "learning_rate": 1.9874513189523102e-07,
+      "loss": 0.2344,
+      "step": 3054
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.9411253333091736,
+      "learning_rate": 1.914542200756181e-07,
+      "loss": 0.2153,
+      "step": 3055
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.9949620366096497,
+      "learning_rate": 1.842994490769273e-07,
+      "loss": 0.2434,
+      "step": 3056
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.1056835651397705,
+      "learning_rate": 1.772808267035547e-07,
+      "loss": 0.2845,
+      "step": 3057
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.0158149003982544,
+      "learning_rate": 1.703983606114179e-07,
+      "loss": 0.2857,
+      "step": 3058
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.2848867177963257,
+      "learning_rate": 1.6365205830787267e-07,
+      "loss": 0.5692,
+      "step": 3059
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.018404483795166,
+      "learning_rate": 1.5704192715178257e-07,
+      "loss": 0.3901,
+      "step": 3060
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 1.9516348838806152,
+      "learning_rate": 1.5056797435344937e-07,
+      "loss": 0.6798,
+      "step": 3061
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.1646894216537476,
+      "learning_rate": 1.4423020697465484e-07,
+      "loss": 0.3379,
+      "step": 3062
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.3057594299316406,
+      "learning_rate": 1.3802863192860504e-07,
+      "loss": 0.4302,
+      "step": 3063
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.0812674760818481,
+      "learning_rate": 1.3196325597995828e-07,
+      "loss": 0.3526,
+      "step": 3064
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.0315954685211182,
+      "learning_rate": 1.2603408574483887e-07,
+      "loss": 0.325,
+      "step": 3065
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.7634711265563965,
+      "learning_rate": 1.2024112769074002e-07,
+      "loss": 0.1287,
+      "step": 3066
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.4026670455932617,
+      "learning_rate": 1.145843881366071e-07,
+      "loss": 0.5154,
+      "step": 3067
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.2732499837875366,
+      "learning_rate": 1.0906387325280987e-07,
+      "loss": 0.4641,
+      "step": 3068
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.0098817348480225,
+      "learning_rate": 1.0367958906110087e-07,
+      "loss": 0.3296,
+      "step": 3069
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.400363564491272,
+      "learning_rate": 9.843154143465704e-08,
+      "loss": 0.4768,
+      "step": 3070
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.1068822145462036,
+      "learning_rate": 9.331973609801037e-08,
+      "loss": 0.3435,
+      "step": 3071
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 1.1789405345916748,
+      "learning_rate": 8.834417862711718e-08,
+      "loss": 0.2829,
+      "step": 3072
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 1.0974359512329102,
+      "learning_rate": 8.350487444930278e-08,
+      "loss": 0.3189,
+      "step": 3073
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 1.0662319660186768,
+      "learning_rate": 7.880182884327513e-08,
+      "loss": 0.1984,
+      "step": 3074
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.9128117561340332,
+      "learning_rate": 7.423504693908346e-08,
+      "loss": 0.3309,
+      "step": 3075
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 1.2563143968582153,
+      "learning_rate": 6.980453371817353e-08,
+      "loss": 0.3568,
+      "step": 3076
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.6046451330184937,
+      "learning_rate": 6.55102940133323e-08,
+      "loss": 0.1443,
+      "step": 3077
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 1.0894300937652588,
+      "learning_rate": 6.135233250871563e-08,
+      "loss": 0.2837,
+      "step": 3078
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 1.0161566734313965,
+      "learning_rate": 5.7330653739806614e-08,
+      "loss": 0.3274,
+      "step": 3079
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 1.2504090070724487,
+      "learning_rate": 5.344526209344336e-08,
+      "loss": 0.5186,
+      "step": 3080
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.9541034698486328,
+      "learning_rate": 4.9696161807805117e-08,
+      "loss": 0.2524,
+      "step": 3081
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 1.2277942895889282,
+      "learning_rate": 4.608335697238453e-08,
+      "loss": 0.3188,
+      "step": 3082
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.3345773220062256,
+      "learning_rate": 4.260685152804311e-08,
+      "loss": 0.5299,
+      "step": 3083
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.1143313646316528,
+      "learning_rate": 3.926664926692802e-08,
+      "loss": 0.2181,
+      "step": 3084
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.5047612190246582,
+      "learning_rate": 3.606275383251367e-08,
+      "loss": 0.5564,
+      "step": 3085
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.266991376876831,
+      "learning_rate": 3.299516871962949e-08,
+      "loss": 0.4578,
+      "step": 3086
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.2709894180297852,
+      "learning_rate": 3.006389727436276e-08,
+      "loss": 0.5225,
+      "step": 3087
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.071608543395996,
+      "learning_rate": 2.726894269414193e-08,
+      "loss": 0.4342,
+      "step": 3088
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.1185423135757446,
+      "learning_rate": 2.4610308027708806e-08,
+      "loss": 0.4581,
+      "step": 3089
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.1018141508102417,
+      "learning_rate": 2.2087996175076953e-08,
+      "loss": 0.4682,
+      "step": 3090
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.2371251583099365,
+      "learning_rate": 1.970200988758719e-08,
+      "loss": 0.2975,
+      "step": 3091
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.2653716802597046,
+      "learning_rate": 1.745235176786597e-08,
+      "loss": 0.4085,
+      "step": 3092
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 1.0708848237991333,
+      "learning_rate": 1.533902426983924e-08,
+      "loss": 0.4154,
+      "step": 3093
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 1.3194761276245117,
+      "learning_rate": 1.336202969871858e-08,
+      "loss": 0.3781,
+      "step": 3094
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 1.083683967590332,
+      "learning_rate": 1.1521370210987314e-08,
+      "loss": 0.2848,
+      "step": 3095
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 1.0737828016281128,
+      "learning_rate": 9.817047814442148e-09,
+      "loss": 0.2543,
+      "step": 3096
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.9343252182006836,
+      "learning_rate": 8.249064368151537e-09,
+      "loss": 0.247,
+      "step": 3097
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 1.1907275915145874,
+      "learning_rate": 6.817421582483441e-09,
+      "loss": 0.3894,
+      "step": 3098
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 1.0009660720825195,
+      "learning_rate": 5.5221210190498085e-09,
+      "loss": 0.2432,
+      "step": 3099
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 1.1556274890899658,
+      "learning_rate": 4.363164090775973e-09,
+      "loss": 0.3347,
+      "step": 3100
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 1.2074859142303467,
+      "learning_rate": 3.340552061831259e-09,
+      "loss": 0.3179,
+      "step": 3101
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 1.529038429260254,
+      "learning_rate": 2.4542860476844954e-09,
+      "loss": 0.8287,
+      "step": 3102
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.0030752420425415,
+      "learning_rate": 1.7043670150901358e-09,
+      "loss": 0.2722,
+      "step": 3103
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.2378499507904053,
+      "learning_rate": 1.0907957820327496e-09,
+      "loss": 0.3671,
+      "step": 3104
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.0954062938690186,
+      "learning_rate": 6.135730178102872e-10,
+      "loss": 0.3245,
+      "step": 3105
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.0416110754013062,
+      "learning_rate": 2.726992429646913e-10,
+      "loss": 0.3151,
+      "step": 3106
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.0947151184082031,
+      "learning_rate": 6.817482933740848e-11,
+      "loss": 0.3989,
+      "step": 3107
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.2456382513046265,
+      "learning_rate": 0.0,
+      "loss": 0.4042,
+      "step": 3108
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 3108,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 4.416706180131324e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}