diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17185 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 50,
+  "global_step": 2397,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 25.505249363840598,
+      "learning_rate": 2.7777777777777776e-07,
+      "loss": 2.2899,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 21.678483027579784,
+      "learning_rate": 5.555555555555555e-07,
+      "loss": 2.2769,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 26.59727728839949,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 2.5262,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 25.40385261848215,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 2.3225,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 20.854241840653593,
+      "learning_rate": 1.3888888888888892e-06,
+      "loss": 2.1934,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 21.361719116645002,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 2.1343,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 18.193012437207088,
+      "learning_rate": 1.944444444444445e-06,
+      "loss": 1.9818,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 15.874239105714151,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.7864,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 14.643013572082948,
+      "learning_rate": 2.5e-06,
+      "loss": 1.8432,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.8718047927552295,
+      "learning_rate": 2.7777777777777783e-06,
+      "loss": 1.5016,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.191810477054005,
+      "learning_rate": 3.055555555555556e-06,
+      "loss": 1.5519,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.767792498196569,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.5226,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.14802596353624,
+      "learning_rate": 3.6111111111111115e-06,
+      "loss": 1.6055,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.234464423613831,
+      "learning_rate": 3.88888888888889e-06,
+      "loss": 1.4491,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.560025745854116,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.6011,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.154014864210117,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 1.5542,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 15.480256268478772,
+      "learning_rate": 4.722222222222222e-06,
+      "loss": 1.5749,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4602251022849813,
+      "learning_rate": 5e-06,
+      "loss": 1.5325,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.236389799056888,
+      "learning_rate": 5.2777777777777785e-06,
+      "loss": 1.4976,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.840902720329139,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 1.4049,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.07106867400434,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 1.5106,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.453853153442332,
+      "learning_rate": 6.111111111111112e-06,
+      "loss": 1.4462,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.3165457589803577,
+      "learning_rate": 6.3888888888888885e-06,
+      "loss": 1.4898,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7503039561271176,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.463,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.5439501059492327,
+      "learning_rate": 6.944444444444445e-06,
+      "loss": 1.4495,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.926396911879542,
+      "learning_rate": 7.222222222222223e-06,
+      "loss": 1.2916,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.4345233805238116,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.4761,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.2438613906867335,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 1.4366,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7927065621259943,
+      "learning_rate": 8.055555555555557e-06,
+      "loss": 1.3505,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.806496995827704,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.4239,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.601248417159665,
+      "learning_rate": 8.611111111111112e-06,
+      "loss": 1.3235,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.033207870946212,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 1.4108,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.322468702457431,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 1.3367,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.0584313673923638,
+      "learning_rate": 9.444444444444445e-06,
+      "loss": 1.457,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.1914159123310846,
+      "learning_rate": 9.722222222222223e-06,
+      "loss": 1.2576,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.3748499855961422,
+      "learning_rate": 1e-05,
+      "loss": 1.3249,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.4771558050738527,
+      "learning_rate": 1.0277777777777777e-05,
+      "loss": 1.4292,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2813066054858466,
+      "learning_rate": 1.0555555555555557e-05,
+      "loss": 1.2969,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.8701334331838466,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 1.2833,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 10.061811285504659,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 1.4369,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.609352006325047,
+      "learning_rate": 1.138888888888889e-05,
+      "loss": 1.2606,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.0468768814183123,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 1.3861,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7283887218681504,
+      "learning_rate": 1.1944444444444444e-05,
+      "loss": 1.3036,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.398611539279862,
+      "learning_rate": 1.2222222222222224e-05,
+      "loss": 1.3445,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.600874698460049,
+      "learning_rate": 1.25e-05,
+      "loss": 1.2134,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7527398319037384,
+      "learning_rate": 1.2777777777777777e-05,
+      "loss": 1.291,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 9.650939558308956,
+      "learning_rate": 1.3055555555555557e-05,
+      "loss": 1.2675,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.9616454282610616,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.386,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.139606751272629,
+      "learning_rate": 1.3611111111111113e-05,
+      "loss": 1.2865,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.8735936514267626,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 1.3433,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 1.72141695022583,
+      "eval_runtime": 10.125,
+      "eval_samples_per_second": 88.889,
+      "eval_steps_per_second": 0.198,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6294350089585565,
+      "learning_rate": 1.416666666666667e-05,
+      "loss": 1.2407,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.156889488291122,
+      "learning_rate": 1.4444444444444446e-05,
+      "loss": 1.3154,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.2245931223694417,
+      "learning_rate": 1.4722222222222224e-05,
+      "loss": 1.2171,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 11.569847009736273,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.3136,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.338596744513355,
+      "learning_rate": 1.5277777777777777e-05,
+      "loss": 1.2707,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.071092183900207,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 1.2669,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.859387535782194,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 1.2781,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.493581477800294,
+      "learning_rate": 1.6111111111111115e-05,
+      "loss": 1.336,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7412964072476145,
+      "learning_rate": 1.638888888888889e-05,
+      "loss": 1.2252,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 51.046626501690476,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.3205,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7200789207727305,
+      "learning_rate": 1.6944444444444446e-05,
+      "loss": 1.3259,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.2690456988197156,
+      "learning_rate": 1.7222222222222224e-05,
+      "loss": 1.1771,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6318039727031586,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 1.2417,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4491280049836996,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 1.3084,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.3310764321125,
+      "learning_rate": 1.8055555555555558e-05,
+      "loss": 1.2252,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4239156975562444,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 1.3491,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 11.73567310626887,
+      "learning_rate": 1.8611111111111114e-05,
+      "loss": 1.2189,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5402945256193268,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 1.2529,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5951599456493244,
+      "learning_rate": 1.916666666666667e-05,
+      "loss": 1.3444,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.6447468558660383,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 1.1966,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5637502807174433,
+      "learning_rate": 1.9722222222222224e-05,
+      "loss": 1.2515,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.954810456141417,
+      "learning_rate": 2e-05,
+      "loss": 1.2623,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5563547933547337,
+      "learning_rate": 1.9999990870988185e-05,
+      "loss": 1.294,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.1713705320312813,
+      "learning_rate": 1.9999963483969405e-05,
+      "loss": 1.1987,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.323094673788578,
+      "learning_rate": 1.9999917838993664e-05,
+      "loss": 1.3151,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.326550336426196,
+      "learning_rate": 1.9999853936144303e-05,
+      "loss": 1.2645,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.102056442198325,
+      "learning_rate": 1.999977177553799e-05,
+      "loss": 1.3182,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.0063388604988557,
+      "learning_rate": 1.9999671357324743e-05,
+      "loss": 1.2417,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.939715055397657,
+      "learning_rate": 1.99995526816879e-05,
+      "loss": 1.3227,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5503418997446585,
+      "learning_rate": 1.9999415748844137e-05,
+      "loss": 1.2147,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 5.187003705203586,
+      "learning_rate": 1.999926055904347e-05,
+      "loss": 1.2533,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.091833875725846,
+      "learning_rate": 1.9999087112569246e-05,
+      "loss": 1.1901,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.559033125715415,
+      "learning_rate": 1.999889540973814e-05,
+      "loss": 1.2991,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6725643823687217,
+      "learning_rate": 1.999868545090017e-05,
+      "loss": 1.203,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.87593747435648,
+      "learning_rate": 1.9998457236438674e-05,
+      "loss": 1.3135,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.656309816931765,
+      "learning_rate": 1.9998210766770323e-05,
+      "loss": 1.2016,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0266369883226214,
+      "learning_rate": 1.9997946042345128e-05,
+      "loss": 1.2631,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.849488309387943,
+      "learning_rate": 1.9997663063646424e-05,
+      "loss": 1.3313,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9226696098760114,
+      "learning_rate": 1.9997361831190872e-05,
+      "loss": 1.2775,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.111181145042153,
+      "learning_rate": 1.9997042345528466e-05,
+      "loss": 1.1473,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.8980504379400087,
+      "learning_rate": 1.9996704607242516e-05,
+      "loss": 1.2775,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.4173946465347047,
+      "learning_rate": 1.9996348616949673e-05,
+      "loss": 1.1644,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1761946632829883,
+      "learning_rate": 1.99959743752999e-05,
+      "loss": 1.2316,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.5785679747035974,
+      "learning_rate": 1.9995581882976492e-05,
+      "loss": 1.2873,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1329488895977415,
+      "learning_rate": 1.999517114069606e-05,
+      "loss": 1.2313,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.487475818049426,
+      "learning_rate": 1.999474214920854e-05,
+      "loss": 1.2256,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.685971757070303,
+      "learning_rate": 1.999429490929718e-05,
+      "loss": 1.3053,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.512352189623193,
+      "learning_rate": 1.999382942177856e-05,
+      "loss": 1.2158,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.3908505476047957,
+      "learning_rate": 1.9993345687502566e-05,
+      "loss": 1.2314,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.061114056324173,
+      "learning_rate": 1.99928437073524e-05,
+      "loss": 1.2793,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 1.859473466873169,
+      "eval_runtime": 9.1823,
+      "eval_samples_per_second": 98.015,
+      "eval_steps_per_second": 0.218,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.9192371335139855,
+      "learning_rate": 1.9992323482244576e-05,
+      "loss": 1.2609,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.9505080251869393,
+      "learning_rate": 1.9991785013128922e-05,
+      "loss": 1.1779,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.5397547924992443,
+      "learning_rate": 1.9991228300988586e-05,
+      "loss": 1.3407,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.8984264717838548,
+      "learning_rate": 1.999065334684e-05,
+      "loss": 1.2566,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.521828568278579,
+      "learning_rate": 1.999006015173293e-05,
+      "loss": 1.244,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6348337597516203,
+      "learning_rate": 1.9989448716750423e-05,
+      "loss": 1.289,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0174698859014164,
+      "learning_rate": 1.998881904300884e-05,
+      "loss": 1.2262,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5765539229171517,
+      "learning_rate": 1.9988171131657848e-05,
+      "loss": 1.1793,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.6864653272743033,
+      "learning_rate": 1.9987504983880395e-05,
+      "loss": 1.2962,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.958485330501808,
+      "learning_rate": 1.9986820600892743e-05,
+      "loss": 1.2675,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.7043912282342135,
+      "learning_rate": 1.998611798394444e-05,
+      "loss": 1.1734,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.039963780720724,
+      "learning_rate": 1.998539713431832e-05,
+      "loss": 1.1895,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5136446560341326,
+      "learning_rate": 1.9984658053330516e-05,
+      "loss": 1.279,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.799209324330764,
+      "learning_rate": 1.998390074233044e-05,
+      "loss": 1.2491,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 11.50065353253629,
+      "learning_rate": 1.99831252027008e-05,
+      "loss": 1.2365,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.749202626354803,
+      "learning_rate": 1.9982331435857566e-05,
+      "loss": 1.2707,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.6638610380451615,
+      "learning_rate": 1.998151944325001e-05,
+      "loss": 1.0686,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.5514142143174414,
+      "learning_rate": 1.9980689226360668e-05,
+      "loss": 1.2969,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.334857150061336,
+      "learning_rate": 1.997984078670535e-05,
+      "loss": 1.247,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.365443569140406,
+      "learning_rate": 1.9978974125833136e-05,
+      "loss": 1.2392,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.3274757510131665,
+      "learning_rate": 1.9978089245326387e-05,
+      "loss": 1.1335,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 60.62144095007304,
+      "learning_rate": 1.9977186146800707e-05,
+      "loss": 1.3059,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7562932298333025,
+      "learning_rate": 1.9976264831904988e-05,
+      "loss": 1.1836,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 14.029642980655849,
+      "learning_rate": 1.9975325302321363e-05,
+      "loss": 1.2519,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.750620188593457,
+      "learning_rate": 1.9974367559765223e-05,
+      "loss": 1.1992,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.409791360101249,
+      "learning_rate": 1.9973391605985226e-05,
+      "loss": 1.3445,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5851268142647785,
+      "learning_rate": 1.997239744276326e-05,
+      "loss": 1.1284,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.4660973027845183,
+      "learning_rate": 1.9971385071914485e-05,
+      "loss": 1.3297,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.330685707510604,
+      "learning_rate": 1.9970354495287275e-05,
+      "loss": 1.1264,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 37.4834790464467,
+      "learning_rate": 1.9969305714763266e-05,
+      "loss": 1.279,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.6058362999231914,
+      "learning_rate": 1.9968238732257327e-05,
+      "loss": 1.1942,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 8.42263645321561,
+      "learning_rate": 1.996715354971755e-05,
+      "loss": 1.3222,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1871062359215867,
+      "learning_rate": 1.9966050169125275e-05,
+      "loss": 1.2373,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1731882495285246,
+      "learning_rate": 1.9964928592495046e-05,
+      "loss": 1.2284,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.362609188632245,
+      "learning_rate": 1.996378882187464e-05,
+      "loss": 1.1386,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.2082659767201855,
+      "learning_rate": 1.9962630859345062e-05,
+      "loss": 1.2727,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.0385194899267955,
+      "learning_rate": 1.9961454707020515e-05,
+      "loss": 1.2284,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3135064076427243,
+      "learning_rate": 1.9960260367048426e-05,
+      "loss": 1.172,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.824777991005825,
+      "learning_rate": 1.9959047841609417e-05,
+      "loss": 1.3214,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.700263181947178,
+      "learning_rate": 1.9957817132917325e-05,
+      "loss": 1.2177,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.6433133031674985,
+      "learning_rate": 1.9956568243219182e-05,
+      "loss": 1.0589,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.673764669899299,
+      "learning_rate": 1.995530117479521e-05,
+      "loss": 1.2803,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6672205115833654,
+      "learning_rate": 1.9954015929958828e-05,
+      "loss": 1.2043,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.39603969943886,
+      "learning_rate": 1.9952712511056637e-05,
+      "loss": 1.2304,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7552776074824443,
+      "learning_rate": 1.9951390920468423e-05,
+      "loss": 1.2717,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.7946635769954031,
+      "learning_rate": 1.995005116060715e-05,
+      "loss": 1.2232,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.3458906468476397,
+      "learning_rate": 1.994869323391895e-05,
+      "loss": 1.1427,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.3873940189172727,
+      "learning_rate": 1.994731714288314e-05,
+      "loss": 1.2699,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.3713962735020107,
+      "learning_rate": 1.994592289001218e-05,
+      "loss": 1.1887,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 7.485396078844685,
+      "learning_rate": 1.9944510477851705e-05,
+      "loss": 1.2083,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 1.932008147239685,
+      "eval_runtime": 8.6545,
+      "eval_samples_per_second": 103.992,
+      "eval_steps_per_second": 0.231,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.6820002497733344,
+      "learning_rate": 1.9943079908980497e-05,
+      "loss": 1.3462,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.2485792195269534,
+      "learning_rate": 1.9941631186010496e-05,
+      "loss": 1.1787,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.448137106053737,
+      "learning_rate": 1.994016431158678e-05,
+      "loss": 1.2558,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.5607991603539175,
+      "learning_rate": 1.9938679288387575e-05,
+      "loss": 1.1574,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.6034938638503675,
+      "learning_rate": 1.9937176119124237e-05,
+      "loss": 1.2556,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3885165116625493,
+      "learning_rate": 1.993565480654126e-05,
+      "loss": 1.1939,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.6175659906524338,
+      "learning_rate": 1.993411535341625e-05,
+      "loss": 1.3012,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.129179408120843,
+      "learning_rate": 1.993255776255996e-05,
+      "loss": 1.117,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4818011610654758,
+      "learning_rate": 1.993098203681623e-05,
+      "loss": 1.2988,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9426979192833407,
+      "learning_rate": 1.9929388179062033e-05,
+      "loss": 1.1978,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.165303993531214,
+      "learning_rate": 1.9927776192207432e-05,
+      "loss": 1.2702,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4981419864099108,
+      "learning_rate": 1.9926146079195597e-05,
+      "loss": 1.1473,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8952415811011036,
+      "learning_rate": 1.9924497843002792e-05,
+      "loss": 1.3092,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.818904576783481,
+      "learning_rate": 1.9922831486638375e-05,
+      "loss": 1.0656,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.680162397866089,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 1.2231,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.956199300794763,
+      "learning_rate": 1.9919444425597523e-05,
+      "loss": 1.2534,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6807922477873793,
+      "learning_rate": 1.991772372710519e-05,
+      "loss": 1.2345,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.363763614038963,
+      "learning_rate": 1.9915984920809442e-05,
+      "loss": 1.1618,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.4940367520170548,
+      "learning_rate": 1.991422800988499e-05,
+      "loss": 1.2399,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8863953145533663,
+      "learning_rate": 1.991245299753961e-05,
+      "loss": 1.2274,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.021954979921746,
+      "learning_rate": 1.9910659887014122e-05,
+      "loss": 1.1786,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.177503355612902,
+      "learning_rate": 1.990884868158239e-05,
+      "loss": 1.1719,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 5.050661422687726,
+      "learning_rate": 1.990701938455132e-05,
+      "loss": 1.2453,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.696378046145972,
+      "learning_rate": 1.9905171999260846e-05,
+      "loss": 1.1709,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.860317583588694,
+      "learning_rate": 1.9903306529083927e-05,
+      "loss": 1.2814,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.7589162171620822,
+      "learning_rate": 1.9901422977426547e-05,
+      "loss": 1.2012,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9510248154657992,
+      "learning_rate": 1.989952134772769e-05,
+      "loss": 1.2582,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.90160597815886,
+      "learning_rate": 1.989760164345937e-05,
+      "loss": 1.1304,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.0672697957412387,
+      "learning_rate": 1.9895663868126577e-05,
+      "loss": 1.2824,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.6561112218361946,
+      "learning_rate": 1.9893708025267305e-05,
+      "loss": 1.2757,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2653116679581045,
+      "learning_rate": 1.989173411845254e-05,
+      "loss": 1.1942,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.9149135777786603,
+      "learning_rate": 1.988974215128625e-05,
+      "loss": 1.2704,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.5588456509066178,
+      "learning_rate": 1.9887732127405367e-05,
+      "loss": 1.1236,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.5092854765686252,
+      "learning_rate": 1.98857040504798e-05,
+      "loss": 1.2279,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.5638463438864856,
+      "learning_rate": 1.988365792421241e-05,
+      "loss": 1.2905,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.857075640116396,
+      "learning_rate": 1.988159375233903e-05,
+      "loss": 1.1922,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.1567441876295885,
+      "learning_rate": 1.987951153862843e-05,
+      "loss": 1.2175,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4509353160997158,
+      "learning_rate": 1.9877411286882308e-05,
+      "loss": 1.3053,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.2559229759805417,
+      "learning_rate": 1.987529300093532e-05,
+      "loss": 1.1513,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.6239700219549746,
+      "learning_rate": 1.987315668465503e-05,
+      "loss": 1.2034,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.078057390406743,
+      "learning_rate": 1.9871002341941938e-05,
+      "loss": 1.1942,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.63929905259416,
+      "learning_rate": 1.9868829976729444e-05,
+      "loss": 1.225,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.3646407571065147,
+      "learning_rate": 1.9866639592983853e-05,
+      "loss": 1.0957,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4479063072999476,
+      "learning_rate": 1.9864431194704378e-05,
+      "loss": 1.2661,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4600470954732294,
+      "learning_rate": 1.9862204785923117e-05,
+      "loss": 1.1266,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4562983354094812,
+      "learning_rate": 1.985996037070505e-05,
+      "loss": 1.2426,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.05089606332769,
+      "learning_rate": 1.985769795314804e-05,
+      "loss": 1.191,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 6.964525947927704,
+      "learning_rate": 1.9855417537382807e-05,
+      "loss": 1.1981,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2734372991202425,
+      "learning_rate": 1.985311912757295e-05,
+      "loss": 1.1938,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.5981443744699897,
+      "learning_rate": 1.9850802727914902e-05,
+      "loss": 1.3014,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 2.131434679031372,
+      "eval_runtime": 8.8386,
+      "eval_samples_per_second": 101.826,
+      "eval_steps_per_second": 0.226,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1334793452059497,
+      "learning_rate": 1.9848468342637957e-05,
+      "loss": 1.1603,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.5584553013760785,
+      "learning_rate": 1.9846115976004234e-05,
+      "loss": 1.2888,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.9973887620187984,
+      "learning_rate": 1.9843745632308697e-05,
+      "loss": 1.1548,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.766492883863429,
+      "learning_rate": 1.984135731587912e-05,
+      "loss": 1.3009,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.310182791521504,
+      "learning_rate": 1.98389510310761e-05,
+      "loss": 1.2104,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.602902713470204,
+      "learning_rate": 1.9836526782293037e-05,
+      "loss": 1.1044,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.001899446126104,
+      "learning_rate": 1.983408457395613e-05,
+      "loss": 1.2189,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.369763229739522,
+      "learning_rate": 1.9831624410524364e-05,
+      "loss": 1.2268,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1790537937802807,
+      "learning_rate": 1.982914629648952e-05,
+      "loss": 1.1513,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0008039648366776,
+      "learning_rate": 1.9826650236376136e-05,
+      "loss": 1.2918,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8943475230448845,
+      "learning_rate": 1.982413623474153e-05,
+      "loss": 1.2414,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4717781955199234,
+      "learning_rate": 1.9821604296175774e-05,
+      "loss": 1.1844,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 8.398896127674554,
+      "learning_rate": 1.981905442530168e-05,
+      "loss": 1.1867,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8881666755645945,
+      "learning_rate": 1.9816486626774814e-05,
+      "loss": 1.2165,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.5589200405498094,
+      "learning_rate": 1.9813900905283464e-05,
+      "loss": 1.095,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4542439873504946,
+      "learning_rate": 1.9811297265548654e-05,
+      "loss": 1.2721,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.184104096315812,
+      "learning_rate": 1.9808675712324108e-05,
+      "loss": 1.2176,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.4529272671030034,
+      "learning_rate": 1.9806036250396266e-05,
+      "loss": 1.1736,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.940138953648698,
+      "learning_rate": 1.9803378884584266e-05,
+      "loss": 1.2941,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1777281823059202,
+      "learning_rate": 1.980070361973993e-05,
+      "loss": 1.1273,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 15.966091894655618,
+      "learning_rate": 1.9798010460747766e-05,
+      "loss": 1.2734,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.952590110577043,
+      "learning_rate": 1.9795299412524948e-05,
+      "loss": 1.1374,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.9588919771900546,
+      "learning_rate": 1.9792570480021315e-05,
+      "loss": 1.3206,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.7923539212520883,
+      "learning_rate": 1.9789823668219353e-05,
+      "loss": 1.12,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.4265775037469408,
+      "learning_rate": 1.978705898213421e-05,
+      "loss": 1.2529,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.859751699150808,
+      "learning_rate": 1.9784276426813645e-05,
+      "loss": 1.2016,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.628875443578729,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 1.2256,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.495088397107423,
+      "learning_rate": 1.977865772882046e-05,
+      "loss": 1.2426,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.081288653595053,
+      "learning_rate": 1.9775821596406475e-05,
+      "loss": 1.2255,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4601913185970388,
+      "learning_rate": 1.977296761527432e-05,
+      "loss": 1.1579,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.9289792159291452,
+      "learning_rate": 1.9770095790634792e-05,
+      "loss": 1.2396,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5329522457654283,
+      "learning_rate": 1.9767206127731283e-05,
+      "loss": 1.1133,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4799836400667294,
+      "learning_rate": 1.9764298631839746e-05,
+      "loss": 1.2747,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9950326659308697,
+      "learning_rate": 1.976137330826869e-05,
+      "loss": 1.2072,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2537155072307415,
+      "learning_rate": 1.975843016235918e-05,
+      "loss": 1.2178,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.247934100648022,
+      "learning_rate": 1.975546919948482e-05,
+      "loss": 1.1889,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.734714053241722,
+      "learning_rate": 1.975249042505174e-05,
+      "loss": 1.2283,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 11.413424179847599,
+      "learning_rate": 1.9749493844498596e-05,
+      "loss": 1.1435,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.7660748108977906,
+      "learning_rate": 1.9746479463296553e-05,
+      "loss": 1.2228,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0584030554493347,
+      "learning_rate": 1.9743447286949274e-05,
+      "loss": 1.1755,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.198561740487497,
+      "learning_rate": 1.974039732099291e-05,
+      "loss": 1.1789,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8514003598900113,
+      "learning_rate": 1.9737329570996098e-05,
+      "loss": 1.2069,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.6719625675346954,
+      "learning_rate": 1.9734244042559948e-05,
+      "loss": 1.226,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.5077647821982194,
+      "learning_rate": 1.973114074131802e-05,
+      "loss": 1.0696,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.2148023342135414,
+      "learning_rate": 1.972801967293633e-05,
+      "loss": 1.3204,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.6756311001971307,
+      "learning_rate": 1.972488084311333e-05,
+      "loss": 1.1728,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.120131183107689,
+      "learning_rate": 1.9721724257579907e-05,
+      "loss": 1.2163,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.074971214940691,
+      "learning_rate": 1.971854992209936e-05,
+      "loss": 1.2884,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 5.626031552765455,
+      "learning_rate": 1.97153578424674e-05,
+      "loss": 1.0733,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.3106888219202208,
+      "learning_rate": 1.971214802451213e-05,
+      "loss": 1.1066,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 2.200995683670044,
+      "eval_runtime": 9.2288,
+      "eval_samples_per_second": 97.521,
+      "eval_steps_per_second": 0.217,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1222357520998771,
+      "learning_rate": 1.9708920474094046e-05,
+      "loss": 1.2713,
+      "step": 251
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.094290805834737,
+      "learning_rate": 1.970567519710602e-05,
+      "loss": 1.2255,
+      "step": 252
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.852165039345409,
+      "learning_rate": 1.9702412199473277e-05,
+      "loss": 1.1585,
+      "step": 253
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 9.884006226472785,
+      "learning_rate": 1.9699131487153416e-05,
+      "loss": 1.2318,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7190151519348766,
+      "learning_rate": 1.9695833066136365e-05,
+      "loss": 1.1975,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.130320765554474,
+      "learning_rate": 1.9692516942444387e-05,
+      "loss": 1.2311,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.210332919209866,
+      "learning_rate": 1.968918312213207e-05,
+      "loss": 1.2288,
+      "step": 257
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.548131763463991,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 1.2535,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 7.636588730588618,
+      "learning_rate": 1.968246241602631e-05,
+      "loss": 1.088,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.6466118562670908,
+      "learning_rate": 1.9679075542503546e-05,
+      "loss": 1.2737,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.8836544659549443,
+      "learning_rate": 1.9675670996901783e-05,
+      "loss": 1.1747,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.394617807128666,
+      "learning_rate": 1.9672248785437053e-05,
+      "loss": 1.1933,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 6.723840821354797,
+      "learning_rate": 1.966880891435763e-05,
+      "loss": 1.1163,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5419290414603544,
+      "learning_rate": 1.966535138994404e-05,
+      "loss": 1.2668,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7275899984867513,
+      "learning_rate": 1.9661876218509045e-05,
+      "loss": 1.1051,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.2327379546094166,
+      "learning_rate": 1.965838340639761e-05,
+      "loss": 1.2571,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.29075940702595,
+      "learning_rate": 1.9654872959986936e-05,
+      "loss": 1.2123,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5482061600203874,
+      "learning_rate": 1.9651344885686394e-05,
+      "loss": 1.1596,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.5406136871326366,
+      "learning_rate": 1.9647799189937552e-05,
+      "loss": 1.1989,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 24.819582563935725,
+      "learning_rate": 1.9644235879214144e-05,
+      "loss": 1.2166,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.1149076166235483,
+      "learning_rate": 1.964065496002208e-05,
+      "loss": 1.1544,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.2802998499634266,
+      "learning_rate": 1.963705643889941e-05,
+      "loss": 1.2671,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.3415856174395566,
+      "learning_rate": 1.963344032241632e-05,
+      "loss": 1.1397,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.9004815162548123,
+      "learning_rate": 1.962980661717512e-05,
+      "loss": 1.2102,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.046154314253929,
+      "learning_rate": 1.9626155329810244e-05,
+      "loss": 1.2014,
+      "step": 275
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6750797185829793,
+      "learning_rate": 1.9622486466988216e-05,
+      "loss": 1.2545,
+      "step": 276
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 12.683203231551689,
+      "learning_rate": 1.961880003540766e-05,
+      "loss": 1.2536,
+      "step": 277
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.093665244882962,
+      "learning_rate": 1.9615096041799267e-05,
+      "loss": 1.164,
+      "step": 278
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.113229470973958,
+      "learning_rate": 1.96113744929258e-05,
+      "loss": 1.196,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.173320420645036,
+      "learning_rate": 1.9607635395582067e-05,
+      "loss": 1.2346,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.717831756579617,
+      "learning_rate": 1.9603878756594928e-05,
+      "loss": 1.1075,
+      "step": 281
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.576975336229833,
+      "learning_rate": 1.960010458282326e-05,
+      "loss": 1.2585,
+      "step": 282
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8649888008699103,
+      "learning_rate": 1.9596312881157953e-05,
+      "loss": 1.2787,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.2744740651163462,
+      "learning_rate": 1.9592503658521917e-05,
+      "loss": 1.0691,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.712041535917794,
+      "learning_rate": 1.9588676921870028e-05,
+      "loss": 1.2507,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.0295001799054835,
+      "learning_rate": 1.9584832678189154e-05,
+      "loss": 1.1671,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.841970209796399,
+      "learning_rate": 1.958097093449813e-05,
+      "loss": 1.2194,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.9572041796651463,
+      "learning_rate": 1.957709169784773e-05,
+      "loss": 1.1534,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 3.4752872835634876,
+      "learning_rate": 1.9573194975320672e-05,
+      "loss": 1.2155,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.730886528028278,
+      "learning_rate": 1.956928077403161e-05,
+      "loss": 1.1828,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.547370618665364,
+      "learning_rate": 1.9565349101127095e-05,
+      "loss": 1.2362,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1114831977267672,
+      "learning_rate": 1.9561399963785586e-05,
+      "loss": 1.0388,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0956422045704286,
+      "learning_rate": 1.955743336921743e-05,
+      "loss": 1.2709,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.5162962076734336,
+      "learning_rate": 1.9553449324664844e-05,
+      "loss": 1.1497,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8859322215056735,
+      "learning_rate": 1.95494478374019e-05,
+      "loss": 1.211,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.431121159403125,
+      "learning_rate": 1.954542891473453e-05,
+      "loss": 1.2136,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2550902393470214,
+      "learning_rate": 1.954139256400049e-05,
+      "loss": 1.2684,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.711424640536355,
+      "learning_rate": 1.953733879256936e-05,
+      "loss": 1.0888,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.5051079824666336,
+      "learning_rate": 1.953326760784252e-05,
+      "loss": 1.3047,
+      "step": 299
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2186148350577,
+      "learning_rate": 1.952917901725316e-05,
+      "loss": 1.1212,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 1.9820705652236938,
+      "eval_runtime": 8.9734,
+      "eval_samples_per_second": 100.297,
+      "eval_steps_per_second": 0.223,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5469112937770568,
+      "learning_rate": 1.9525073028266225e-05,
+      "loss": 1.1987,
+      "step": 301
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.399754630383876,
+      "learning_rate": 1.9520949648378444e-05,
+      "loss": 1.0764,
+      "step": 302
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8745155235732036,
+      "learning_rate": 1.9516808885118297e-05,
+      "loss": 1.2593,
+      "step": 303
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1398517937181643,
+      "learning_rate": 1.9512650746046003e-05,
+      "loss": 1.1552,
+      "step": 304
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.043129904117724,
+      "learning_rate": 1.950847523875349e-05,
+      "loss": 1.2,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8384980878663744,
+      "learning_rate": 1.950428237086442e-05,
+      "loss": 1.2008,
+      "step": 306
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5195191148653209,
+      "learning_rate": 1.9500072150034136e-05,
+      "loss": 1.2464,
+      "step": 307
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9993941066381796,
+      "learning_rate": 1.9495844583949668e-05,
+      "loss": 1.1422,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0771186000019095,
+      "learning_rate": 1.949159968032972e-05,
+      "loss": 1.2061,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.4641587149623163,
+      "learning_rate": 1.948733744692465e-05,
+      "loss": 1.1707,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1402262834126857,
+      "learning_rate": 1.948305789151644e-05,
+      "loss": 1.2449,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1988989357020685,
+      "learning_rate": 1.947876102191873e-05,
+      "loss": 1.1285,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.837267154691931,
+      "learning_rate": 1.947444684597674e-05,
+      "loss": 1.2483,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1039845992054638,
+      "learning_rate": 1.9470115371567312e-05,
+      "loss": 1.2146,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.132506968478357,
+      "learning_rate": 1.946576660659886e-05,
+      "loss": 1.1201,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8964780570572068,
+      "learning_rate": 1.9461400559011367e-05,
+      "loss": 1.1245,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.088591278075442,
+      "learning_rate": 1.945701723677637e-05,
+      "loss": 1.2976,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.407362728810993,
+      "learning_rate": 1.9452616647896963e-05,
+      "loss": 1.03,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7940927643022375,
+      "learning_rate": 1.9448198800407737e-05,
+      "loss": 1.1957,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.2199993008468775,
+      "learning_rate": 1.944376370237481e-05,
+      "loss": 1.226,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0451419949161613,
+      "learning_rate": 1.94393113618958e-05,
+      "loss": 1.2159,
+      "step": 321
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.373784393173936,
+      "learning_rate": 1.9434841787099804e-05,
+      "loss": 1.2204,
+      "step": 322
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0806838464858444,
+      "learning_rate": 1.9430354986147372e-05,
+      "loss": 1.1987,
+      "step": 323
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0726785713751303,
+      "learning_rate": 1.942585096723052e-05,
+      "loss": 1.231,
+      "step": 324
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5861704363704372,
+      "learning_rate": 1.9421329738572702e-05,
+      "loss": 1.1387,
+      "step": 325
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7234428648614297,
+      "learning_rate": 1.941679130842878e-05,
+      "loss": 1.2056,
+      "step": 326
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3678484193498104,
+      "learning_rate": 1.9412235685085034e-05,
+      "loss": 1.1549,
+      "step": 327
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1159121397662957,
+      "learning_rate": 1.940766287685913e-05,
+      "loss": 1.2367,
+      "step": 328
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1733246901326513,
+      "learning_rate": 1.9403072892100115e-05,
+      "loss": 1.157,
+      "step": 329
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.842593361570564,
+      "learning_rate": 1.9398465739188392e-05,
+      "loss": 1.1964,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.3235050053412762,
+      "learning_rate": 1.9393841426535712e-05,
+      "loss": 1.1806,
+      "step": 331
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1335307593055122,
+      "learning_rate": 1.9389199962585156e-05,
+      "loss": 1.1929,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0811636548917436,
+      "learning_rate": 1.938454135581112e-05,
+      "loss": 1.1746,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0791298567218608,
+      "learning_rate": 1.93798656147193e-05,
+      "loss": 1.2418,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5411330341529867,
+      "learning_rate": 1.9375172747846677e-05,
+      "loss": 1.1677,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1753906220353323,
+      "learning_rate": 1.9370462763761494e-05,
+      "loss": 1.2458,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1784007630441216,
+      "learning_rate": 1.9365735671063247e-05,
+      "loss": 1.1767,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.760878768163942,
+      "learning_rate": 1.9360991478382686e-05,
+      "loss": 1.1991,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6864257240417833,
+      "learning_rate": 1.935623019438176e-05,
+      "loss": 1.0032,
+      "step": 339
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3773484943317202,
+      "learning_rate": 1.9351451827753632e-05,
+      "loss": 1.2953,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2070623732910468,
+      "learning_rate": 1.9346656387222662e-05,
+      "loss": 1.1312,
+      "step": 341
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.10314014364944,
+      "learning_rate": 1.9341843881544372e-05,
+      "loss": 1.2368,
+      "step": 342
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7495195763510192,
+      "learning_rate": 1.9337014319505445e-05,
+      "loss": 1.2788,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3351514670841258,
+      "learning_rate": 1.9332167709923708e-05,
+      "loss": 1.1393,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.694500016828258,
+      "learning_rate": 1.9327304061648116e-05,
+      "loss": 1.1952,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1929400147303522,
+      "learning_rate": 1.9322423383558724e-05,
+      "loss": 1.2091,
+      "step": 346
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7894637405161866,
+      "learning_rate": 1.9317525684566686e-05,
+      "loss": 1.174,
+      "step": 347
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1379616956954417,
+      "learning_rate": 1.9312610973614238e-05,
+      "loss": 1.2652,
+      "step": 348
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.456590470991627,
+      "learning_rate": 1.9307679259674664e-05,
+      "loss": 1.273,
+      "step": 349
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4425426232450702,
+      "learning_rate": 1.93027305517523e-05,
+      "loss": 1.0908,
+      "step": 350
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 2.07572078704834,
+      "eval_runtime": 15.8018,
+      "eval_samples_per_second": 56.955,
+      "eval_steps_per_second": 0.127,
+      "step": 350
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.2689186254119418,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 1.1089,
+      "step": 351
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0122460721066195,
+      "learning_rate": 1.9292782190131677e-05,
+      "loss": 1.2951,
+      "step": 352
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0935600040560374,
+      "learning_rate": 1.9287782554597156e-05,
+      "loss": 1.061,
+      "step": 353
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.3737918703470573,
+      "learning_rate": 1.9282765961407297e-05,
+      "loss": 1.2383,
+      "step": 354
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.412709197943325,
+      "learning_rate": 1.927773241972141e-05,
+      "loss": 1.202,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.2113667890720095,
+      "learning_rate": 1.927268193872975e-05,
+      "loss": 1.1832,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.26046151380203,
+      "learning_rate": 1.926761452765349e-05,
+      "loss": 1.2191,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6413922275268436,
+      "learning_rate": 1.926253019574472e-05,
+      "loss": 1.1768,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.457660264234057,
+      "learning_rate": 1.9257428952286433e-05,
+      "loss": 1.1521,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.278659799279929,
+      "learning_rate": 1.925231080659249e-05,
+      "loss": 1.2855,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1050176721362794,
+      "learning_rate": 1.924717576800761e-05,
+      "loss": 1.2217,
+      "step": 361
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4181234843889508,
+      "learning_rate": 1.9242023845907362e-05,
+      "loss": 1.1306,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.211151305340775,
+      "learning_rate": 1.9236855049698132e-05,
+      "loss": 1.2003,
+      "step": 363
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.2536913564456404,
+      "learning_rate": 1.9231669388817128e-05,
+      "loss": 1.1801,
+      "step": 364
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.159618817568335,
+      "learning_rate": 1.9226466872732333e-05,
+      "loss": 1.212,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.929590460743508,
+      "learning_rate": 1.922124751094252e-05,
+      "loss": 1.1142,
+      "step": 366
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6536710220358137,
+      "learning_rate": 1.921601131297721e-05,
+      "loss": 1.2463,
+      "step": 367
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4358262354732345,
+      "learning_rate": 1.921075828839667e-05,
+      "loss": 1.1115,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.2110187167464121,
+      "learning_rate": 1.9205488446791873e-05,
+      "loss": 1.2352,
+      "step": 369
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4812948600569564,
+      "learning_rate": 1.9200201797784523e-05,
+      "loss": 1.0488,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1379979696432205,
+      "learning_rate": 1.9194898351026983e-05,
+      "loss": 1.2016,
+      "step": 371
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4886349718952565,
+      "learning_rate": 1.918957811620231e-05,
+      "loss": 1.1991,
+      "step": 372
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.483480019721692,
+      "learning_rate": 1.918424110302419e-05,
+      "loss": 1.2313,
+      "step": 373
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6625308139817123,
+      "learning_rate": 1.9178887321236965e-05,
+      "loss": 1.146,
+      "step": 374
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6686158459847893,
+      "learning_rate": 1.9173516780615575e-05,
+      "loss": 1.26,
+      "step": 375
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1700363334956327,
+      "learning_rate": 1.9168129490965568e-05,
+      "loss": 1.0946,
+      "step": 376
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1006254255384906,
+      "learning_rate": 1.9162725462123074e-05,
+      "loss": 1.1806,
+      "step": 377
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.5168545842774475,
+      "learning_rate": 1.9157304703954772e-05,
+      "loss": 1.1807,
+      "step": 378
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1902486555051361,
+      "learning_rate": 1.915186722635791e-05,
+      "loss": 1.2525,
+      "step": 379
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.144161379721218,
+      "learning_rate": 1.9146413039260234e-05,
+      "loss": 1.0775,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0673221100118442,
+      "learning_rate": 1.9140942152620012e-05,
+      "loss": 1.2335,
+      "step": 381
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.184148696665428,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 1.1141,
+      "step": 382
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9605291614082017,
+      "learning_rate": 1.9129950320697456e-05,
+      "loss": 1.1922,
+      "step": 383
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.343002879910368,
+      "learning_rate": 1.912442939548403e-05,
+      "loss": 1.1648,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7743027058254943,
+      "learning_rate": 1.9118891810865846e-05,
+      "loss": 1.25,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.662519107426177,
+      "learning_rate": 1.911333757695345e-05,
+      "loss": 1.1672,
+      "step": 386
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7652896464496066,
+      "learning_rate": 1.9107766703887764e-05,
+      "loss": 1.2137,
+      "step": 387
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.5019080057848355,
+      "learning_rate": 1.9102179201840112e-05,
+      "loss": 1.0533,
+      "step": 388
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0607987326962365,
+      "learning_rate": 1.909657508101216e-05,
+      "loss": 1.1102,
+      "step": 389
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.106328035206022,
+      "learning_rate": 1.909095435163593e-05,
+      "loss": 1.1824,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0049058818010066,
+      "learning_rate": 1.9085317023973765e-05,
+      "loss": 1.2107,
+      "step": 391
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.480554599800787,
+      "learning_rate": 1.9079663108318304e-05,
+      "loss": 1.1428,
+      "step": 392
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.054833995415477,
+      "learning_rate": 1.907399261499248e-05,
+      "loss": 1.1919,
+      "step": 393
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1185475105210057,
+      "learning_rate": 1.90683055543495e-05,
+      "loss": 1.2282,
+      "step": 394
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.072179077354714,
+      "learning_rate": 1.9062601936772807e-05,
+      "loss": 1.1355,
+      "step": 395
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1782858650515655,
+      "learning_rate": 1.9056881772676084e-05,
+      "loss": 1.1468,
+      "step": 396
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5885040734682745,
+      "learning_rate": 1.9051145072503216e-05,
+      "loss": 1.2119,
+      "step": 397
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1387003702456904,
+      "learning_rate": 1.9045391846728284e-05,
+      "loss": 1.2151,
+      "step": 398
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8557244857895485,
+      "learning_rate": 1.903962210585554e-05,
+      "loss": 1.0909,
+      "step": 399
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0970212184387709,
+      "learning_rate": 1.9033835860419398e-05,
+      "loss": 1.2862,
+      "step": 400
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 2.23453426361084,
+      "eval_runtime": 8.7668,
+      "eval_samples_per_second": 102.661,
+      "eval_steps_per_second": 0.228,
+      "step": 400
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5777745686971714,
+      "learning_rate": 1.9028033120984394e-05,
+      "loss": 1.1275,
+      "step": 401
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3125755151343144,
+      "learning_rate": 1.9022213898145176e-05,
+      "loss": 1.1676,
+      "step": 402
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.518169080212938,
+      "learning_rate": 1.901637820252651e-05,
+      "loss": 1.3018,
+      "step": 403
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2417981316079993,
+      "learning_rate": 1.901052604478321e-05,
+      "loss": 1.2017,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.0661637874618126,
+      "learning_rate": 1.9004657435600165e-05,
+      "loss": 1.1539,
+      "step": 405
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2295334582526776,
+      "learning_rate": 1.8998772385692294e-05,
+      "loss": 1.2376,
+      "step": 406
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.511554269646418,
+      "learning_rate": 1.8992870905804535e-05,
+      "loss": 1.128,
+      "step": 407
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3136263395578844,
+      "learning_rate": 1.8986953006711822e-05,
+      "loss": 1.1912,
+      "step": 408
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3969231570741996,
+      "learning_rate": 1.8981018699219077e-05,
+      "loss": 1.2067,
+      "step": 409
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5960116720407655,
+      "learning_rate": 1.8975067994161164e-05,
+      "loss": 1.239,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1653043226338953,
+      "learning_rate": 1.8969100902402903e-05,
+      "loss": 1.0378,
+      "step": 411
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2908839819212194,
+      "learning_rate": 1.896311743483901e-05,
+      "loss": 1.3097,
+      "step": 412
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.274969438605365,
+      "learning_rate": 1.895711760239413e-05,
+      "loss": 1.1304,
+      "step": 413
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.094481680920178,
+      "learning_rate": 1.895110141602276e-05,
+      "loss": 1.2434,
+      "step": 414
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6851129525593473,
+      "learning_rate": 1.8945068886709273e-05,
+      "loss": 1.2434,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 16.399447309006582,
+      "learning_rate": 1.893902002546788e-05,
+      "loss": 1.0331,
+      "step": 416
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2666488519346097,
+      "learning_rate": 1.893295484334259e-05,
+      "loss": 1.1204,
+      "step": 417
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2616900306437178,
+      "learning_rate": 1.8926873351407244e-05,
+      "loss": 1.2619,
+      "step": 418
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.291158436765171,
+      "learning_rate": 1.8920775560765435e-05,
+      "loss": 1.1074,
+      "step": 419
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1498340562281941,
+      "learning_rate": 1.8914661482550523e-05,
+      "loss": 1.2306,
+      "step": 420
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.82965514718623,
+      "learning_rate": 1.8908531127925616e-05,
+      "loss": 1.1418,
+      "step": 421
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1498056325562676,
+      "learning_rate": 1.8902384508083518e-05,
+      "loss": 1.2083,
+      "step": 422
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1797080806052807,
+      "learning_rate": 1.889622163424675e-05,
+      "loss": 1.187,
+      "step": 423
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.038365759267788,
+      "learning_rate": 1.8890042517667495e-05,
+      "loss": 1.1763,
+      "step": 424
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0458622606055223,
+      "learning_rate": 1.8883847169627606e-05,
+      "loss": 1.1468,
+      "step": 425
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.218423521844201,
+      "learning_rate": 1.887763560143856e-05,
+      "loss": 1.2353,
+      "step": 426
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.443179340197502,
+      "learning_rate": 1.8871407824441453e-05,
+      "loss": 1.1831,
+      "step": 427
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1881207571016068,
+      "learning_rate": 1.8865163850006977e-05,
+      "loss": 1.1933,
+      "step": 428
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.199724698118787,
+      "learning_rate": 1.885890368953539e-05,
+      "loss": 1.1313,
+      "step": 429
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7523996710688787,
+      "learning_rate": 1.8852627354456512e-05,
+      "loss": 1.1694,
+      "step": 430
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.2045624961818526,
+      "learning_rate": 1.884633485622969e-05,
+      "loss": 1.1492,
+      "step": 431
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9504390874684212,
+      "learning_rate": 1.8840026206343786e-05,
+      "loss": 1.2748,
+      "step": 432
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0246290714595854,
+      "learning_rate": 1.8833701416317143e-05,
+      "loss": 1.2511,
+      "step": 433
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1870947427861467,
+      "learning_rate": 1.8827360497697575e-05,
+      "loss": 1.1969,
+      "step": 434
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.3153588110001821,
+      "learning_rate": 1.8821003462062354e-05,
+      "loss": 1.2247,
+      "step": 435
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.846984372168359,
+      "learning_rate": 1.8814630321018165e-05,
+      "loss": 1.1639,
+      "step": 436
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9973200340891678,
+      "learning_rate": 1.8808241086201106e-05,
+      "loss": 1.1655,
+      "step": 437
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.130529179844024,
+      "learning_rate": 1.880183576927665e-05,
+      "loss": 1.1154,
+      "step": 438
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.8443595598307114,
+      "learning_rate": 1.8795414381939653e-05,
+      "loss": 1.2484,
+      "step": 439
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.58822190113735,
+      "learning_rate": 1.8788976935914293e-05,
+      "loss": 1.1878,
+      "step": 440
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8551463659301584,
+      "learning_rate": 1.878252344295407e-05,
+      "loss": 1.096,
+      "step": 441
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.2451802764712105,
+      "learning_rate": 1.877605391484179e-05,
+      "loss": 1.0145,
+      "step": 442
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.359883877153515,
+      "learning_rate": 1.8769568363389535e-05,
+      "loss": 1.2789,
+      "step": 443
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3394169383014622,
+      "learning_rate": 1.8763066800438638e-05,
+      "loss": 1.1248,
+      "step": 444
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1879310118481765,
+      "learning_rate": 1.8756549237859665e-05,
+      "loss": 1.2439,
+      "step": 445
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3832384698438789,
+      "learning_rate": 1.8750015687552404e-05,
+      "loss": 1.0984,
+      "step": 446
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.080374727621366,
+      "learning_rate": 1.8743466161445823e-05,
+      "loss": 1.2736,
+      "step": 447
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8906852827771856,
+      "learning_rate": 1.873690067149806e-05,
+      "loss": 1.0986,
+      "step": 448
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.4471005281188858,
+      "learning_rate": 1.8730319229696407e-05,
+      "loss": 1.1514,
+      "step": 449
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.2207633036860528,
+      "learning_rate": 1.872372184805727e-05,
+      "loss": 1.1962,
+      "step": 450
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 2.318338394165039,
+      "eval_runtime": 9.6464,
+      "eval_samples_per_second": 93.299,
+      "eval_steps_per_second": 0.207,
+      "step": 450
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.107715212040529,
+      "learning_rate": 1.871710853862617e-05,
+      "loss": 1.228,
+      "step": 451
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.050236885907851,
+      "learning_rate": 1.8710479313477697e-05,
+      "loss": 1.1436,
+      "step": 452
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5525016265274556,
+      "learning_rate": 1.870383418471551e-05,
+      "loss": 1.2076,
+      "step": 453
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.4540043231238755,
+      "learning_rate": 1.86971731644723e-05,
+      "loss": 1.1697,
+      "step": 454
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1533380255818182,
+      "learning_rate": 1.8690496264909774e-05,
+      "loss": 1.2185,
+      "step": 455
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.4069401448418322,
+      "learning_rate": 1.868380349821863e-05,
+      "loss": 1.1243,
+      "step": 456
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0384801965795751,
+      "learning_rate": 1.867709487661854e-05,
+      "loss": 1.241,
+      "step": 457
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.777439294587615,
+      "learning_rate": 1.8670370412358117e-05,
+      "loss": 1.2103,
+      "step": 458
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3739186474990128,
+      "learning_rate": 1.8663630117714904e-05,
+      "loss": 1.1171,
+      "step": 459
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1452694951069902,
+      "learning_rate": 1.865687400499535e-05,
+      "loss": 1.1242,
+      "step": 460
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.2012963731650756,
+      "learning_rate": 1.865010208653478e-05,
+      "loss": 1.2559,
+      "step": 461
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6229571091824606,
+      "learning_rate": 1.8643314374697377e-05,
+      "loss": 1.1246,
+      "step": 462
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.229485731377732,
+      "learning_rate": 1.8636510881876163e-05,
+      "loss": 1.2562,
+      "step": 463
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0757394184423095,
+      "learning_rate": 1.8629691620492973e-05,
+      "loss": 1.0952,
+      "step": 464
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0095002364126235,
+      "learning_rate": 1.862285660299843e-05,
+      "loss": 1.1171,
+      "step": 465
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1793390560508847,
+      "learning_rate": 1.861600584187192e-05,
+      "loss": 1.1747,
+      "step": 466
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8460104239695654,
+      "learning_rate": 1.8609139349621588e-05,
+      "loss": 1.2249,
+      "step": 467
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2804627494329568,
+      "learning_rate": 1.8602257138784284e-05,
+      "loss": 1.1123,
+      "step": 468
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6626087840462676,
+      "learning_rate": 1.8595359221925566e-05,
+      "loss": 1.2798,
+      "step": 469
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.338428107533118,
+      "learning_rate": 1.858844561163967e-05,
+      "loss": 1.2509,
+      "step": 470
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.546883512641666,
+      "learning_rate": 1.858151632054948e-05,
+      "loss": 1.1401,
+      "step": 471
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.4228633014732317,
+      "learning_rate": 1.857457136130651e-05,
+      "loss": 1.2417,
+      "step": 472
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.507742945499947,
+      "learning_rate": 1.8567610746590888e-05,
+      "loss": 1.0691,
+      "step": 473
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2775824776892464,
+      "learning_rate": 1.8560634489111318e-05,
+      "loss": 1.1186,
+      "step": 474
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1044012867193909,
+      "learning_rate": 1.855364260160507e-05,
+      "loss": 1.2368,
+      "step": 475
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2128588395909372,
+      "learning_rate": 1.854663509683794e-05,
+      "loss": 1.2367,
+      "step": 476
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.3087764631033088,
+      "learning_rate": 1.853961198760426e-05,
+      "loss": 1.0988,
+      "step": 477
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.069003105434836,
+      "learning_rate": 1.8532573286726828e-05,
+      "loss": 1.2081,
+      "step": 478
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.430188192380364,
+      "learning_rate": 1.8525519007056926e-05,
+      "loss": 1.2042,
+      "step": 479
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1540546720677767,
+      "learning_rate": 1.851844916147427e-05,
+      "loss": 1.1673,
+      "step": 480
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1536923213359338,
+      "learning_rate": 1.8511363762887013e-05,
+      "loss": 1.1481,
+      "step": 481
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7452711944578976,
+      "learning_rate": 1.8504262824231675e-05,
+      "loss": 1.1823,
+      "step": 482
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2067276037905468,
+      "learning_rate": 1.849714635847318e-05,
+      "loss": 1.1516,
+      "step": 483
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.5220653838680887,
+      "learning_rate": 1.8490014378604783e-05,
+      "loss": 1.1923,
+      "step": 484
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7087144988336622,
+      "learning_rate": 1.848286689764807e-05,
+      "loss": 1.1683,
+      "step": 485
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 3.4524068789271585,
+      "learning_rate": 1.8475703928652932e-05,
+      "loss": 1.1844,
+      "step": 486
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.117360434516391,
+      "learning_rate": 1.8468525484697527e-05,
+      "loss": 1.1952,
+      "step": 487
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.2568567989707327,
+      "learning_rate": 1.846133157888828e-05,
+      "loss": 1.1703,
+      "step": 488
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2999375095932106,
+      "learning_rate": 1.845412222435984e-05,
+      "loss": 1.098,
+      "step": 489
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.612386597939116,
+      "learning_rate": 1.8446897434275066e-05,
+      "loss": 1.1672,
+      "step": 490
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.906947547996957,
+      "learning_rate": 1.8439657221824994e-05,
+      "loss": 1.2178,
+      "step": 491
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.61211727324661,
+      "learning_rate": 1.8432401600228823e-05,
+      "loss": 1.1409,
+      "step": 492
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.2251048819459183,
+      "learning_rate": 1.8425130582733883e-05,
+      "loss": 1.0305,
+      "step": 493
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.420618988270917,
+      "learning_rate": 1.8417844182615616e-05,
+      "loss": 1.2206,
+      "step": 494
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.5097625115674207,
+      "learning_rate": 1.8410542413177546e-05,
+      "loss": 1.1982,
+      "step": 495
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.71948444537149,
+      "learning_rate": 1.840322528775126e-05,
+      "loss": 1.1476,
+      "step": 496
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.1794361931630004,
+      "learning_rate": 1.839589281969639e-05,
+      "loss": 1.1505,
+      "step": 497
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.6072586370432385,
+      "learning_rate": 1.8388545022400567e-05,
+      "loss": 1.178,
+      "step": 498
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.2120836952937795,
+      "learning_rate": 1.8381181909279417e-05,
+      "loss": 1.1799,
+      "step": 499
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6595066067589548,
+      "learning_rate": 1.8373803493776534e-05,
+      "loss": 1.1881,
+      "step": 500
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 2.3302342891693115,
+      "eval_runtime": 11.9091,
+      "eval_samples_per_second": 75.573,
+      "eval_steps_per_second": 0.168,
+      "step": 500
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.3971883598198371,
+      "learning_rate": 1.8366409789363443e-05,
+      "loss": 1.1369,
+      "step": 501
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.0610184192225791,
+      "learning_rate": 1.8359000809539584e-05,
+      "loss": 1.1739,
+      "step": 502
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.5052468690085083,
+      "learning_rate": 1.8351576567832297e-05,
+      "loss": 1.254,
+      "step": 503
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.379897029103935,
+      "learning_rate": 1.834413707779677e-05,
+      "loss": 1.099,
+      "step": 504
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.5621409555186805,
+      "learning_rate": 1.8336682353016055e-05,
+      "loss": 1.2643,
+      "step": 505
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.9952117558473773,
+      "learning_rate": 1.8329212407100996e-05,
+      "loss": 1.102,
+      "step": 506
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.1877374354731909,
+      "learning_rate": 1.832172725369024e-05,
+      "loss": 1.2296,
+      "step": 507
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.645724785997549,
+      "learning_rate": 1.8314226906450197e-05,
+      "loss": 1.1163,
+      "step": 508
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.3258139969363645,
+      "learning_rate": 1.8306711379075025e-05,
+      "loss": 1.2699,
+      "step": 509
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 5.440844995865374,
+      "learning_rate": 1.8299180685286586e-05,
+      "loss": 1.1198,
+      "step": 510
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8628680797378927,
+      "learning_rate": 1.8291634838834437e-05,
+      "loss": 1.1978,
+      "step": 511
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.2285561060602594,
+      "learning_rate": 1.8284073853495808e-05,
+      "loss": 1.1874,
+      "step": 512
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0432530366947863,
+      "learning_rate": 1.8276497743075556e-05,
+      "loss": 1.2376,
+      "step": 513
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.205275408087816,
+      "learning_rate": 1.8268906521406167e-05,
+      "loss": 1.0937,
+      "step": 514
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.0991944247648386,
+      "learning_rate": 1.8261300202347712e-05,
+      "loss": 1.0449,
+      "step": 515
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.4780592587966195,
+      "learning_rate": 1.8253678799787825e-05,
+      "loss": 1.111,
+      "step": 516
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9161457788583751,
+      "learning_rate": 1.8246042327641678e-05,
+      "loss": 1.2923,
+      "step": 517
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.0693133928406877,
+      "learning_rate": 1.8238390799851963e-05,
+      "loss": 1.1037,
+      "step": 518
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 7.150180870988112,
+      "learning_rate": 1.823072423038886e-05,
+      "loss": 1.0942,
+      "step": 519
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.4152616724647569,
+      "learning_rate": 1.8223042633250003e-05,
+      "loss": 1.1645,
+      "step": 520
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.034208261367486,
+      "learning_rate": 1.8215346022460475e-05,
+      "loss": 1.2496,
+      "step": 521
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.0223936768193995,
+      "learning_rate": 1.8207634412072765e-05,
+      "loss": 1.0887,
+      "step": 522
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.0887257005091793,
+      "learning_rate": 1.819990781616675e-05,
+      "loss": 1.2198,
+      "step": 523
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.870259521257212,
+      "learning_rate": 1.8192166248849666e-05,
+      "loss": 1.1245,
+      "step": 524
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.3095458734196104,
+      "learning_rate": 1.8184409724256084e-05,
+      "loss": 1.1663,
+      "step": 525
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.2465149188628106,
+      "learning_rate": 1.8176638256547888e-05,
+      "loss": 1.1007,
+      "step": 526
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 5.36661569986971,
+      "learning_rate": 1.816885185991424e-05,
+      "loss": 1.2254,
+      "step": 527
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.5200547757059286,
+      "learning_rate": 1.8161050548571562e-05,
+      "loss": 1.1694,
+      "step": 528
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 12.792249778977917,
+      "learning_rate": 1.8153234336763502e-05,
+      "loss": 1.1995,
+      "step": 529
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.2676601271959995,
+      "learning_rate": 1.8145403238760928e-05,
+      "loss": 1.175,
+      "step": 530
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.3634648128912141,
+      "learning_rate": 1.813755726886187e-05,
+      "loss": 1.1671,
+      "step": 531
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.302931597427782,
+      "learning_rate": 1.812969644139152e-05,
+      "loss": 1.1612,
+      "step": 532
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.1588216051286209,
+      "learning_rate": 1.8121820770702195e-05,
+      "loss": 1.2089,
+      "step": 533
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.4957237132218806,
+      "learning_rate": 1.8113930271173314e-05,
+      "loss": 1.1213,
+      "step": 534
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 8.249481467528316,
+      "learning_rate": 1.810602495721137e-05,
+      "loss": 1.2511,
+      "step": 535
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.534775437545512,
+      "learning_rate": 1.80981048432499e-05,
+      "loss": 1.1626,
+      "step": 536
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.873070888222635,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 1.0517,
+      "step": 537
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6214573929008316,
+      "learning_rate": 1.8082220273197645e-05,
+      "loss": 1.2057,
+      "step": 538
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.638183186061765,
+      "learning_rate": 1.8074255846108944e-05,
+      "loss": 1.2385,
+      "step": 539
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.614134266569183,
+      "learning_rate": 1.8066276677024833e-05,
+      "loss": 1.0931,
+      "step": 540
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.3820448947965522,
+      "learning_rate": 1.805828278051371e-05,
+      "loss": 1.1209,
+      "step": 541
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6565000198893707,
+      "learning_rate": 1.8050274171170835e-05,
+      "loss": 1.1933,
+      "step": 542
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.1195411433725266,
+      "learning_rate": 1.804225086361836e-05,
+      "loss": 1.0919,
+      "step": 543
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7520749641364282,
+      "learning_rate": 1.803421287250525e-05,
+      "loss": 1.1944,
+      "step": 544
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.4775718417560257,
+      "learning_rate": 1.8026160212507293e-05,
+      "loss": 1.1732,
+      "step": 545
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0660896131776303,
+      "learning_rate": 1.801809289832705e-05,
+      "loss": 1.2423,
+      "step": 546
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.1704676425589078,
+      "learning_rate": 1.8010010944693846e-05,
+      "loss": 1.1008,
+      "step": 547
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.578476492910222,
+      "learning_rate": 1.800191436636373e-05,
+      "loss": 1.116,
+      "step": 548
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8710145529061912,
+      "learning_rate": 1.7993803178119457e-05,
+      "loss": 1.0613,
+      "step": 549
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.971049801804774,
+      "learning_rate": 1.798567739477045e-05,
+      "loss": 1.1872,
+      "step": 550
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 2.2507705688476562,
+      "eval_runtime": 9.6106,
+      "eval_samples_per_second": 93.647,
+      "eval_steps_per_second": 0.208,
+      "step": 550
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8387639469616224,
+      "learning_rate": 1.7977537031152784e-05,
+      "loss": 1.1792,
+      "step": 551
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.6286308292110891,
+      "learning_rate": 1.7969382102129153e-05,
+      "loss": 1.1688,
+      "step": 552
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.278490310723867,
+      "learning_rate": 1.7961212622588847e-05,
+      "loss": 1.1291,
+      "step": 553
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.2420044178220009,
+      "learning_rate": 1.795302860744772e-05,
+      "loss": 1.1675,
+      "step": 554
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.5742171408381633,
+      "learning_rate": 1.794483007164817e-05,
+      "loss": 1.1715,
+      "step": 555
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.250899990292976,
+      "learning_rate": 1.7936617030159096e-05,
+      "loss": 1.199,
+      "step": 556
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.213581463962385,
+      "learning_rate": 1.7928389497975897e-05,
+      "loss": 1.1887,
+      "step": 557
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.563398827626112,
+      "learning_rate": 1.7920147490120412e-05,
+      "loss": 1.2505,
+      "step": 558
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.3025732594310135,
+      "learning_rate": 1.7911891021640925e-05,
+      "loss": 1.0823,
+      "step": 559
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.1102899375804909,
+      "learning_rate": 1.7903620107612114e-05,
+      "loss": 1.2222,
+      "step": 560
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.5764114937274702,
+      "learning_rate": 1.7895334763135036e-05,
+      "loss": 1.1545,
+      "step": 561
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.3318072088385624,
+      "learning_rate": 1.7887035003337082e-05,
+      "loss": 1.2116,
+      "step": 562
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9295243728922132,
+      "learning_rate": 1.7878720843371986e-05,
+      "loss": 1.0458,
+      "step": 563
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.384925256087727,
+      "learning_rate": 1.7870392298419755e-05,
+      "loss": 1.1993,
+      "step": 564
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.1043450479838783,
+      "learning_rate": 1.7862049383686667e-05,
+      "loss": 1.1441,
+      "step": 565
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.951200882869599,
+      "learning_rate": 1.7853692114405233e-05,
+      "loss": 1.2724,
+      "step": 566
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.2416378310633358,
+      "learning_rate": 1.7845320505834176e-05,
+      "loss": 1.2487,
+      "step": 567
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.1476943533305437,
+      "learning_rate": 1.78369345732584e-05,
+      "loss": 1.0998,
+      "step": 568
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.46040400834289,
+      "learning_rate": 1.782853433198896e-05,
+      "loss": 1.1683,
+      "step": 569
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.527261923342412,
+      "learning_rate": 1.7820119797363038e-05,
+      "loss": 1.1866,
+      "step": 570
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.9207290766858335,
+      "learning_rate": 1.7811690984743906e-05,
+      "loss": 1.1375,
+      "step": 571
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9944239110575535,
+      "learning_rate": 1.780324790952092e-05,
+      "loss": 1.1433,
+      "step": 572
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.9994039897859311,
+      "learning_rate": 1.7794790587109456e-05,
+      "loss": 1.2842,
+      "step": 573
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.553422950392834,
+      "learning_rate": 1.778631903295092e-05,
+      "loss": 1.0938,
+      "step": 574
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.4017553539357843,
+      "learning_rate": 1.7777833262512693e-05,
+      "loss": 1.1912,
+      "step": 575
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.213978133332068,
+      "learning_rate": 1.776933329128811e-05,
+      "loss": 1.1309,
+      "step": 576
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.9700218541607801,
+      "learning_rate": 1.776081913479645e-05,
+      "loss": 1.226,
+      "step": 577
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.7290246416014727,
+      "learning_rate": 1.775229080858287e-05,
+      "loss": 1.1901,
+      "step": 578
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6591153501096514,
+      "learning_rate": 1.7743748328218412e-05,
+      "loss": 1.2132,
+      "step": 579
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.31871419230632,
+      "learning_rate": 1.773519170929996e-05,
+      "loss": 1.0232,
+      "step": 580
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.7073121167356407,
+      "learning_rate": 1.7726620967450202e-05,
+      "loss": 1.2425,
+      "step": 581
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.1962891014364416,
+      "learning_rate": 1.771803611831762e-05,
+      "loss": 1.1555,
+      "step": 582
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7060514455391431,
+      "learning_rate": 1.7709437177576457e-05,
+      "loss": 1.191,
+      "step": 583
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.7036917648686165,
+      "learning_rate": 1.770082416092668e-05,
+      "loss": 1.1606,
+      "step": 584
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.516961076332457,
+      "learning_rate": 1.7692197084093944e-05,
+      "loss": 1.1962,
+      "step": 585
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.1689710099391983,
+      "learning_rate": 1.7683555962829603e-05,
+      "loss": 1.1499,
+      "step": 586
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.6873600622400902,
+      "learning_rate": 1.767490081291062e-05,
+      "loss": 1.161,
+      "step": 587
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.2600793155580938,
+      "learning_rate": 1.7666231650139602e-05,
+      "loss": 1.1559,
+      "step": 588
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.3311360701033437,
+      "learning_rate": 1.7657548490344713e-05,
+      "loss": 1.2122,
+      "step": 589
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.777072117798904,
+      "learning_rate": 1.7648851349379697e-05,
+      "loss": 1.1659,
+      "step": 590
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.1442995350206253,
+      "learning_rate": 1.7640140243123813e-05,
+      "loss": 1.2191,
+      "step": 591
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5326965367977092,
+      "learning_rate": 1.7631415187481818e-05,
+      "loss": 1.0422,
+      "step": 592
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.3809788889394463,
+      "learning_rate": 1.762267619838394e-05,
+      "loss": 1.2238,
+      "step": 593
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.2193651591885903,
+      "learning_rate": 1.7613923291785842e-05,
+      "loss": 1.0818,
+      "step": 594
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.021715844772141,
+      "learning_rate": 1.760515648366861e-05,
+      "loss": 1.2243,
+      "step": 595
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.0521350094464825,
+      "learning_rate": 1.7596375790038697e-05,
+      "loss": 1.2105,
+      "step": 596
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5692840506643515,
+      "learning_rate": 1.758758122692791e-05,
+      "loss": 1.1915,
+      "step": 597
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6454805688051455,
+      "learning_rate": 1.757877281039339e-05,
+      "loss": 1.1203,
+      "step": 598
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.3855630447812133,
+      "learning_rate": 1.7569950556517566e-05,
+      "loss": 1.2336,
+      "step": 599
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.554718129493987,
+      "learning_rate": 1.7561114481408125e-05,
+      "loss": 1.1508,
+      "step": 600
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 2.309680223464966,
+      "eval_runtime": 11.4401,
+      "eval_samples_per_second": 78.67,
+      "eval_steps_per_second": 0.175,
+      "step": 600
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.3409123070040216,
+      "learning_rate": 1.7552264601197995e-05,
+      "loss": 1.144,
+      "step": 601
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.3256989955185785,
+      "learning_rate": 1.754340093204531e-05,
+      "loss": 1.265,
+      "step": 602
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.439284547156723,
+      "learning_rate": 1.7534523490133378e-05,
+      "loss": 1.1423,
+      "step": 603
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8419761394649303,
+      "learning_rate": 1.752563229167065e-05,
+      "loss": 1.1372,
+      "step": 604
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.4051183360937836,
+      "learning_rate": 1.75167273528907e-05,
+      "loss": 1.2518,
+      "step": 605
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.1554963734529125,
+      "learning_rate": 1.7507808690052183e-05,
+      "loss": 1.0524,
+      "step": 606
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6226174454487892,
+      "learning_rate": 1.749887631943882e-05,
+      "loss": 1.1747,
+      "step": 607
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.5461130500300047,
+      "learning_rate": 1.7489930257359352e-05,
+      "loss": 1.1715,
+      "step": 608
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.65707410264589,
+      "learning_rate": 1.7480970520147522e-05,
+      "loss": 1.2094,
+      "step": 609
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6396829505346866,
+      "learning_rate": 1.7471997124162036e-05,
+      "loss": 1.1535,
+      "step": 610
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8264473781381552,
+      "learning_rate": 1.7463010085786544e-05,
+      "loss": 1.1889,
+      "step": 611
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5356498213228638,
+      "learning_rate": 1.74540094214296e-05,
+      "loss": 1.1619,
+      "step": 612
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.3029825538337312,
+      "learning_rate": 1.7444995147524638e-05,
+      "loss": 1.1242,
+      "step": 613
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7900642485059937,
+      "learning_rate": 1.7435967280529943e-05,
+      "loss": 1.2291,
+      "step": 614
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.671715568012744,
+      "learning_rate": 1.7426925836928612e-05,
+      "loss": 1.1812,
+      "step": 615
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.119080204413585,
+      "learning_rate": 1.7417870833228542e-05,
+      "loss": 1.183,
+      "step": 616
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6342284857835687,
+      "learning_rate": 1.740880228596237e-05,
+      "loss": 1.1966,
+      "step": 617
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.3465991756006626,
+      "learning_rate": 1.739972021168748e-05,
+      "loss": 1.0688,
+      "step": 618
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8366473541498594,
+      "learning_rate": 1.739062462698594e-05,
+      "loss": 1.1061,
+      "step": 619
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.3029503650242305,
+      "learning_rate": 1.7381515548464486e-05,
+      "loss": 1.1889,
+      "step": 620
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.7089486812378727,
+      "learning_rate": 1.7372392992754505e-05,
+      "loss": 1.2114,
+      "step": 621
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 12.361032925691427,
+      "learning_rate": 1.7363256976511972e-05,
+      "loss": 1.2171,
+      "step": 622
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.5499713874555567,
+      "learning_rate": 1.7354107516417454e-05,
+      "loss": 1.1145,
+      "step": 623
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0364009679487256,
+      "learning_rate": 1.7344944629176054e-05,
+      "loss": 1.2385,
+      "step": 624
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.070516574123712,
+      "learning_rate": 1.733576833151739e-05,
+      "loss": 1.0285,
+      "step": 625
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.2985029159625041,
+      "learning_rate": 1.7326578640195573e-05,
+      "loss": 1.2133,
+      "step": 626
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.3509404516659664,
+      "learning_rate": 1.7317375571989158e-05,
+      "loss": 1.172,
+      "step": 627
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9289765933958682,
+      "learning_rate": 1.7308159143701134e-05,
+      "loss": 1.199,
+      "step": 628
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.97450143243467,
+      "learning_rate": 1.729892937215887e-05,
+      "loss": 1.1418,
+      "step": 629
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.7958584287540105,
+      "learning_rate": 1.7289686274214116e-05,
+      "loss": 1.2172,
+      "step": 630
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.2312826511669777,
+      "learning_rate": 1.7280429866742934e-05,
+      "loss": 1.0427,
+      "step": 631
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.8649680216706086,
+      "learning_rate": 1.7271160166645695e-05,
+      "loss": 1.1706,
+      "step": 632
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.1723709429222797,
+      "learning_rate": 1.7261877190847042e-05,
+      "loss": 1.2,
+      "step": 633
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.558083269766,
+      "learning_rate": 1.725258095629585e-05,
+      "loss": 1.2228,
+      "step": 634
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.171149098960579,
+      "learning_rate": 1.7243271479965215e-05,
+      "loss": 1.0977,
+      "step": 635
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.460996594695065,
+      "learning_rate": 1.7233948778852388e-05,
+      "loss": 1.2481,
+      "step": 636
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 8.35635730398717,
+      "learning_rate": 1.722461286997879e-05,
+      "loss": 1.1176,
+      "step": 637
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7190073739578815,
+      "learning_rate": 1.721526377038994e-05,
+      "loss": 1.0469,
+      "step": 638
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.5493088351003599,
+      "learning_rate": 1.7205901497155445e-05,
+      "loss": 1.1976,
+      "step": 639
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.2461580316164138,
+      "learning_rate": 1.719652606736897e-05,
+      "loss": 1.185,
+      "step": 640
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.2722831908739207,
+      "learning_rate": 1.7187137498148196e-05,
+      "loss": 1.2232,
+      "step": 641
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.876256952704975,
+      "learning_rate": 1.717773580663479e-05,
+      "loss": 1.1836,
+      "step": 642
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.1377570473196197,
+      "learning_rate": 1.7168321009994383e-05,
+      "loss": 1.1721,
+      "step": 643
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.305396411291584,
+      "learning_rate": 1.7158893125416543e-05,
+      "loss": 1.0696,
+      "step": 644
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0955741059327733,
+      "learning_rate": 1.7149452170114713e-05,
+      "loss": 1.223,
+      "step": 645
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.0330303255825237,
+      "learning_rate": 1.7139998161326217e-05,
+      "loss": 1.1125,
+      "step": 646
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.269452114532427,
+      "learning_rate": 1.7130531116312202e-05,
+      "loss": 1.1934,
+      "step": 647
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.3490616611385327,
+      "learning_rate": 1.7121051052357628e-05,
+      "loss": 1.161,
+      "step": 648
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.3901613022480928,
+      "learning_rate": 1.7111557986771215e-05,
+      "loss": 1.2008,
+      "step": 649
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9180330949745716,
+      "learning_rate": 1.710205193688542e-05,
+      "loss": 1.1153,
+      "step": 650
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 2.3337860107421875,
+      "eval_runtime": 9.828,
+      "eval_samples_per_second": 91.575,
+      "eval_steps_per_second": 0.203,
+      "step": 650
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7828504741377469,
+      "learning_rate": 1.709253292005642e-05,
+      "loss": 1.1733,
+      "step": 651
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.814418099196492,
+      "learning_rate": 1.708300095366405e-05,
+      "loss": 1.1482,
+      "step": 652
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.4986798916082222,
+      "learning_rate": 1.70734560551118e-05,
+      "loss": 1.1166,
+      "step": 653
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7757124542841898,
+      "learning_rate": 1.706389824182677e-05,
+      "loss": 1.2158,
+      "step": 654
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.3774795433840779,
+      "learning_rate": 1.7054327531259638e-05,
+      "loss": 1.1738,
+      "step": 655
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3444670763397815,
+      "learning_rate": 1.704474394088463e-05,
+      "loss": 1.171,
+      "step": 656
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.3215639397464745,
+      "learning_rate": 1.703514748819948e-05,
+      "loss": 1.2061,
+      "step": 657
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 3.36582303153467,
+      "learning_rate": 1.702553819072543e-05,
+      "loss": 1.0477,
+      "step": 658
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.2029733886809542,
+      "learning_rate": 1.7015916066007138e-05,
+      "loss": 1.1922,
+      "step": 659
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.673630479839942,
+      "learning_rate": 1.700628113161272e-05,
+      "loss": 1.1543,
+      "step": 660
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.649716900510303,
+      "learning_rate": 1.6996633405133656e-05,
+      "loss": 1.2465,
+      "step": 661
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6245778909525712,
+      "learning_rate": 1.6986972904184783e-05,
+      "loss": 1.0823,
+      "step": 662
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.074738530975692,
+      "learning_rate": 1.6977299646404274e-05,
+      "loss": 1.1676,
+      "step": 663
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7798844584449025,
+      "learning_rate": 1.6967613649453577e-05,
+      "loss": 1.2147,
+      "step": 664
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7564959667985038,
+      "learning_rate": 1.6957914931017417e-05,
+      "loss": 1.1968,
+      "step": 665
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.925663080568778,
+      "learning_rate": 1.6948203508803733e-05,
+      "loss": 1.0461,
+      "step": 666
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5749227079038064,
+      "learning_rate": 1.693847940054366e-05,
+      "loss": 1.1768,
+      "step": 667
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.625072813507722,
+      "learning_rate": 1.6928742623991504e-05,
+      "loss": 1.159,
+      "step": 668
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.2330577594709238,
+      "learning_rate": 1.691899319692469e-05,
+      "loss": 1.0991,
+      "step": 669
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.381994654204721,
+      "learning_rate": 1.6909231137143744e-05,
+      "loss": 1.1285,
+      "step": 670
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.583064337029652,
+      "learning_rate": 1.6899456462472264e-05,
+      "loss": 1.1724,
+      "step": 671
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6876377722896896,
+      "learning_rate": 1.688966919075687e-05,
+      "loss": 1.2285,
+      "step": 672
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.3108070936321552,
+      "learning_rate": 1.6879869339867182e-05,
+      "loss": 1.1406,
+      "step": 673
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.394509470324005,
+      "learning_rate": 1.6870056927695796e-05,
+      "loss": 1.2153,
+      "step": 674
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9213063114262785,
+      "learning_rate": 1.686023197215824e-05,
+      "loss": 1.1342,
+      "step": 675
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.3308166837745965,
+      "learning_rate": 1.6850394491192933e-05,
+      "loss": 1.0995,
+      "step": 676
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8200970083343357,
+      "learning_rate": 1.684054450276118e-05,
+      "loss": 1.1693,
+      "step": 677
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.732009426030266,
+      "learning_rate": 1.6830682024847104e-05,
+      "loss": 1.2462,
+      "step": 678
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.7196941199600797,
+      "learning_rate": 1.6820807075457648e-05,
+      "loss": 1.1566,
+      "step": 679
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.2214435436121678,
+      "learning_rate": 1.681091967262251e-05,
+      "loss": 1.0477,
+      "step": 680
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.1167691123435528,
+      "learning_rate": 1.680101983439414e-05,
+      "loss": 1.1477,
+      "step": 681
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.4725807991428512,
+      "learning_rate": 1.6791107578847688e-05,
+      "loss": 1.2806,
+      "step": 682
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7097260425716874,
+      "learning_rate": 1.678118292408097e-05,
+      "loss": 1.0412,
+      "step": 683
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.5283080597817955,
+      "learning_rate": 1.6771245888214442e-05,
+      "loss": 1.1913,
+      "step": 684
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.3376297577182692,
+      "learning_rate": 1.676129648939117e-05,
+      "loss": 1.208,
+      "step": 685
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7315039475379488,
+      "learning_rate": 1.675133474577679e-05,
+      "loss": 1.1662,
+      "step": 686
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.4425467440751756,
+      "learning_rate": 1.6741360675559475e-05,
+      "loss": 1.1267,
+      "step": 687
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6823577212969671,
+      "learning_rate": 1.673137429694991e-05,
+      "loss": 1.1515,
+      "step": 688
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.3829076936435574,
+      "learning_rate": 1.6721375628181244e-05,
+      "loss": 1.0634,
+      "step": 689
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.654473637803497,
+      "learning_rate": 1.6711364687509073e-05,
+      "loss": 1.238,
+      "step": 690
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.821851044074957,
+      "learning_rate": 1.6701341493211396e-05,
+      "loss": 1.1162,
+      "step": 691
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.1867415140676787,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 1.16,
+      "step": 692
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8109419682524877,
+      "learning_rate": 1.668125841696335e-05,
+      "loss": 1.1739,
+      "step": 693
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.078639535546807,
+      "learning_rate": 1.6671198571680705e-05,
+      "loss": 1.1882,
+      "step": 694
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4653424290758905,
+      "learning_rate": 1.6661126546107945e-05,
+      "loss": 1.1199,
+      "step": 695
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.315524161543452,
+      "learning_rate": 1.6651042358634595e-05,
+      "loss": 1.2212,
+      "step": 696
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5598059590908786,
+      "learning_rate": 1.6640946027672395e-05,
+      "loss": 1.0551,
+      "step": 697
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7970293700114697,
+      "learning_rate": 1.6630837571655236e-05,
+      "loss": 1.1907,
+      "step": 698
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.467371840842433,
+      "learning_rate": 1.6620717009039175e-05,
+      "loss": 1.1831,
+      "step": 699
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.3601874552477335,
+      "learning_rate": 1.661058435830235e-05,
+      "loss": 1.1742,
+      "step": 700
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 2.2965242862701416,
+      "eval_runtime": 9.1471,
+      "eval_samples_per_second": 98.392,
+      "eval_steps_per_second": 0.219,
+      "step": 700
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.3297243101944418,
+      "learning_rate": 1.660043963794498e-05,
+      "loss": 1.1158,
+      "step": 701
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 3.8363547466779977,
+      "learning_rate": 1.659028286648932e-05,
+      "loss": 1.2041,
+      "step": 702
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.168993413099993,
+      "learning_rate": 1.658011406247963e-05,
+      "loss": 1.219,
+      "step": 703
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.423753496998417,
+      "learning_rate": 1.6569933244482137e-05,
+      "loss": 1.1791,
+      "step": 704
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.2906596091783735,
+      "learning_rate": 1.6559740431084997e-05,
+      "loss": 1.169,
+      "step": 705
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.105364504704536,
+      "learning_rate": 1.654953564089828e-05,
+      "loss": 1.2266,
+      "step": 706
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.1104261330758471,
+      "learning_rate": 1.653931889255391e-05,
+      "loss": 1.0318,
+      "step": 707
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7186194009570122,
+      "learning_rate": 1.6529090204705654e-05,
+      "loss": 1.1943,
+      "step": 708
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 6.020457097870213,
+      "learning_rate": 1.651884959602907e-05,
+      "loss": 1.2465,
+      "step": 709
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9982773782143237,
+      "learning_rate": 1.650859708522149e-05,
+      "loss": 1.0943,
+      "step": 710
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6223032807183255,
+      "learning_rate": 1.649833269100197e-05,
+      "loss": 1.097,
+      "step": 711
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.3865996466265957,
+      "learning_rate": 1.648805643211127e-05,
+      "loss": 1.2583,
+      "step": 712
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.564613062049625,
+      "learning_rate": 1.6477768327311802e-05,
+      "loss": 1.1074,
+      "step": 713
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.1204470778861833,
+      "learning_rate": 1.646746839538761e-05,
+      "loss": 1.1624,
+      "step": 714
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.343006914903349,
+      "learning_rate": 1.6457156655144343e-05,
+      "loss": 1.2725,
+      "step": 715
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.1797594067890802,
+      "learning_rate": 1.644683312540919e-05,
+      "loss": 0.9961,
+      "step": 716
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9707952461725868,
+      "learning_rate": 1.6436497825030886e-05,
+      "loss": 1.1532,
+      "step": 717
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.187816990275362,
+      "learning_rate": 1.6426150772879643e-05,
+      "loss": 1.2051,
+      "step": 718
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.0767176782841448,
+      "learning_rate": 1.641579198784713e-05,
+      "loss": 1.1343,
+      "step": 719
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 4.1614908835726485,
+      "learning_rate": 1.6405421488846446e-05,
+      "loss": 1.0993,
+      "step": 720
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.3884196013739962,
+      "learning_rate": 1.6395039294812072e-05,
+      "loss": 1.1772,
+      "step": 721
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.376000172360809,
+      "learning_rate": 1.6384645424699835e-05,
+      "loss": 1.1603,
+      "step": 722
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.1458377508985467,
+      "learning_rate": 1.63742398974869e-05,
+      "loss": 1.1309,
+      "step": 723
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.2250498579189673,
+      "learning_rate": 1.6363822732171693e-05,
+      "loss": 1.1882,
+      "step": 724
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.1152196879499772,
+      "learning_rate": 1.6353393947773902e-05,
+      "loss": 1.1127,
+      "step": 725
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7862023622573282,
+      "learning_rate": 1.634295356333443e-05,
+      "loss": 1.254,
+      "step": 726
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.4870630828724822,
+      "learning_rate": 1.6332501597915353e-05,
+      "loss": 1.2345,
+      "step": 727
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.078179487792858,
+      "learning_rate": 1.632203807059989e-05,
+      "loss": 1.0269,
+      "step": 728
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1523087098080285,
+      "learning_rate": 1.631156300049238e-05,
+      "loss": 1.195,
+      "step": 729
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.3190759231240792,
+      "learning_rate": 1.6301076406718228e-05,
+      "loss": 1.1548,
+      "step": 730
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.2485209702125848,
+      "learning_rate": 1.6290578308423882e-05,
+      "loss": 1.1503,
+      "step": 731
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.5215783597936656,
+      "learning_rate": 1.6280068724776795e-05,
+      "loss": 1.1627,
+      "step": 732
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.3651763451065655,
+      "learning_rate": 1.626954767496539e-05,
+      "loss": 1.226,
+      "step": 733
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 5.091006270325283,
+      "learning_rate": 1.6259015178199024e-05,
+      "loss": 1.1706,
+      "step": 734
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.3534800880895559,
+      "learning_rate": 1.6248471253707953e-05,
+      "loss": 1.1714,
+      "step": 735
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7907030564976056,
+      "learning_rate": 1.6237915920743303e-05,
+      "loss": 1.0396,
+      "step": 736
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.3237482083055063,
+      "learning_rate": 1.622734919857702e-05,
+      "loss": 1.208,
+      "step": 737
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.331028159356576,
+      "learning_rate": 1.6216771106501855e-05,
+      "loss": 1.154,
+      "step": 738
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8534545040838393,
+      "learning_rate": 1.6206181663831317e-05,
+      "loss": 1.189,
+      "step": 739
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.359523300247032,
+      "learning_rate": 1.6195580889899628e-05,
+      "loss": 1.0784,
+      "step": 740
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.26122589848967,
+      "learning_rate": 1.6184968804061706e-05,
+      "loss": 1.189,
+      "step": 741
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6039514866539648,
+      "learning_rate": 1.617434542569313e-05,
+      "loss": 1.0655,
+      "step": 742
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.2399797588829622,
+      "learning_rate": 1.6163710774190082e-05,
+      "loss": 1.1503,
+      "step": 743
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.3844370913044473,
+      "learning_rate": 1.6153064868969336e-05,
+      "loss": 1.2497,
+      "step": 744
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.329619775531287,
+      "learning_rate": 1.6142407729468215e-05,
+      "loss": 1.0905,
+      "step": 745
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.77223599213699,
+      "learning_rate": 1.6131739375144543e-05,
+      "loss": 1.0873,
+      "step": 746
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.5047169421100666,
+      "learning_rate": 1.612105982547663e-05,
+      "loss": 1.2364,
+      "step": 747
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.079190737387077,
+      "learning_rate": 1.6110369099963223e-05,
+      "loss": 1.097,
+      "step": 748
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.2587345079090027,
+      "learning_rate": 1.6099667218123473e-05,
+      "loss": 1.1599,
+      "step": 749
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.3444916242290486,
+      "learning_rate": 1.6088954199496902e-05,
+      "loss": 1.1618,
+      "step": 750
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 2.312548875808716,
+      "eval_runtime": 9.2917,
+      "eval_samples_per_second": 96.861,
+      "eval_steps_per_second": 0.215,
+      "step": 750
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.346957465537316,
+      "learning_rate": 1.6078230063643366e-05,
+      "loss": 1.1827,
+      "step": 751
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0615469251923098,
+      "learning_rate": 1.6067494830143014e-05,
+      "loss": 1.1521,
+      "step": 752
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.1436299980592588,
+      "learning_rate": 1.605674851859626e-05,
+      "loss": 1.1196,
+      "step": 753
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.9320815873312163,
+      "learning_rate": 1.6045991148623752e-05,
+      "loss": 1.1537,
+      "step": 754
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.4861077908468492,
+      "learning_rate": 1.6035222739866314e-05,
+      "loss": 1.1836,
+      "step": 755
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.4412962615085174,
+      "learning_rate": 1.6024443311984938e-05,
+      "loss": 1.162,
+      "step": 756
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.114170263013048,
+      "learning_rate": 1.6013652884660723e-05,
+      "loss": 1.1977,
+      "step": 757
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 6.2244693103634745,
+      "learning_rate": 1.600285147759486e-05,
+      "loss": 1.2423,
+      "step": 758
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7873979391569979,
+      "learning_rate": 1.5992039110508586e-05,
+      "loss": 1.0825,
+      "step": 759
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.3766516809462503,
+      "learning_rate": 1.5981215803143145e-05,
+      "loss": 1.2126,
+      "step": 760
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.733223506911629,
+      "learning_rate": 1.5970381575259753e-05,
+      "loss": 1.0491,
+      "step": 761
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 10.800489881441875,
+      "learning_rate": 1.5959536446639572e-05,
+      "loss": 1.1915,
+      "step": 762
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.5053981347506606,
+      "learning_rate": 1.5948680437083662e-05,
+      "loss": 1.1436,
+      "step": 763
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.340156623890832,
+      "learning_rate": 1.5937813566412955e-05,
+      "loss": 1.0569,
+      "step": 764
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.456228257393803,
+      "learning_rate": 1.5926935854468207e-05,
+      "loss": 1.1639,
+      "step": 765
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7317808511995811,
+      "learning_rate": 1.5916047321109967e-05,
+      "loss": 1.2064,
+      "step": 766
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.1698962866805322,
+      "learning_rate": 1.5905147986218546e-05,
+      "loss": 1.1275,
+      "step": 767
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.894745889194892,
+      "learning_rate": 1.589423786969398e-05,
+      "loss": 1.2017,
+      "step": 768
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.298770874668521,
+      "learning_rate": 1.5883316991455983e-05,
+      "loss": 1.1412,
+      "step": 769
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1091187094975936,
+      "learning_rate": 1.587238537144392e-05,
+      "loss": 1.1808,
+      "step": 770
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.4629701272682496,
+      "learning_rate": 1.5861443029616766e-05,
+      "loss": 1.1028,
+      "step": 771
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.4135360445770444,
+      "learning_rate": 1.5850489985953076e-05,
+      "loss": 1.2008,
+      "step": 772
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8247008160684817,
+      "learning_rate": 1.5839526260450946e-05,
+      "loss": 1.0473,
+      "step": 773
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.04787207843481,
+      "learning_rate": 1.582855187312797e-05,
+      "loss": 1.2291,
+      "step": 774
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8144757001277088,
+      "learning_rate": 1.581756684402121e-05,
+      "loss": 1.1899,
+      "step": 775
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7746735269268876,
+      "learning_rate": 1.5806571193187157e-05,
+      "loss": 1.2102,
+      "step": 776
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6830843245424503,
+      "learning_rate": 1.57955649407017e-05,
+      "loss": 1.0497,
+      "step": 777
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.5141223979611416,
+      "learning_rate": 1.578454810666008e-05,
+      "loss": 1.2354,
+      "step": 778
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.338161150942344,
+      "learning_rate": 1.577352071117685e-05,
+      "loss": 1.2005,
+      "step": 779
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.604597808282837,
+      "learning_rate": 1.5762482774385867e-05,
+      "loss": 1.1186,
+      "step": 780
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8826271226042537,
+      "learning_rate": 1.5751434316440216e-05,
+      "loss": 1.1829,
+      "step": 781
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.4074480079917513,
+      "learning_rate": 1.5740375357512198e-05,
+      "loss": 1.1331,
+      "step": 782
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7107732331564631,
+      "learning_rate": 1.5729305917793286e-05,
+      "loss": 1.136,
+      "step": 783
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.2526219122503977,
+      "learning_rate": 1.5718226017494092e-05,
+      "loss": 1.2126,
+      "step": 784
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.33511340205993,
+      "learning_rate": 1.570713567684432e-05,
+      "loss": 1.1078,
+      "step": 785
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9924248220412712,
+      "learning_rate": 1.5696034916092743e-05,
+      "loss": 1.2216,
+      "step": 786
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.950537898045203,
+      "learning_rate": 1.568492375550715e-05,
+      "loss": 1.2236,
+      "step": 787
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6508825575832302,
+      "learning_rate": 1.567380221537434e-05,
+      "loss": 1.1755,
+      "step": 788
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.6900568990453284,
+      "learning_rate": 1.566267031600003e-05,
+      "loss": 1.198,
+      "step": 789
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.3575592019147917,
+      "learning_rate": 1.5651528077708877e-05,
+      "loss": 1.1092,
+      "step": 790
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.299869944596932,
+      "learning_rate": 1.5640375520844403e-05,
+      "loss": 1.1094,
+      "step": 791
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.4803535822614522,
+      "learning_rate": 1.562921266576898e-05,
+      "loss": 1.1357,
+      "step": 792
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.454870701334724,
+      "learning_rate": 1.5618039532863764e-05,
+      "loss": 1.1236,
+      "step": 793
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.4577939604929835,
+      "learning_rate": 1.560685614252869e-05,
+      "loss": 1.1621,
+      "step": 794
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.8508122925455557,
+      "learning_rate": 1.5595662515182426e-05,
+      "loss": 1.1992,
+      "step": 795
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1444451034438647,
+      "learning_rate": 1.5584458671262316e-05,
+      "loss": 1.0495,
+      "step": 796
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.4384016643574746,
+      "learning_rate": 1.5573244631224364e-05,
+      "loss": 1.2742,
+      "step": 797
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1209815159884706,
+      "learning_rate": 1.55620204155432e-05,
+      "loss": 1.1096,
+      "step": 798
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.28470867971083,
+      "learning_rate": 1.555078604471201e-05,
+      "loss": 1.1952,
+      "step": 799
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.564948691859885,
+      "learning_rate": 1.5539541539242548e-05,
+      "loss": 1.1598,
+      "step": 800
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 2.3567399978637695,
+      "eval_runtime": 12.6573,
+      "eval_samples_per_second": 71.105,
+      "eval_steps_per_second": 0.158,
+      "step": 800
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.2388316514430047,
+      "learning_rate": 1.5528286919665054e-05,
+      "loss": 1.1902,
+      "step": 801
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.3719910721732558,
+      "learning_rate": 1.5517022206528233e-05,
+      "loss": 1.0798,
+      "step": 802
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.1024768808953036,
+      "learning_rate": 1.550574742039923e-05,
+      "loss": 1.1652,
+      "step": 803
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8414689768234727,
+      "learning_rate": 1.549446258186358e-05,
+      "loss": 1.1685,
+      "step": 804
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8434630169776192,
+      "learning_rate": 1.548316771152516e-05,
+      "loss": 1.1409,
+      "step": 805
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5605633198661784,
+      "learning_rate": 1.547186283000618e-05,
+      "loss": 1.1347,
+      "step": 806
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.973849382396006,
+      "learning_rate": 1.5460547957947105e-05,
+      "loss": 1.2064,
+      "step": 807
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.085771373595158,
+      "learning_rate": 1.5449223116006665e-05,
+      "loss": 1.123,
+      "step": 808
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9626480074342836,
+      "learning_rate": 1.5437888324861788e-05,
+      "loss": 1.1445,
+      "step": 809
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.800795523107878,
+      "learning_rate": 1.5426543605207555e-05,
+      "loss": 1.1468,
+      "step": 810
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.3050224449589503,
+      "learning_rate": 1.5415188977757183e-05,
+      "loss": 1.1531,
+      "step": 811
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0085477292137983,
+      "learning_rate": 1.540382446324198e-05,
+      "loss": 1.1858,
+      "step": 812
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9615423564363583,
+      "learning_rate": 1.53924500824113e-05,
+      "loss": 1.1574,
+      "step": 813
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 7.066923876864276,
+      "learning_rate": 1.5381065856032518e-05,
+      "loss": 0.9699,
+      "step": 814
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.766108764644666,
+      "learning_rate": 1.5369671804890978e-05,
+      "loss": 1.2349,
+      "step": 815
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.364519878614181,
+      "learning_rate": 1.5358267949789968e-05,
+      "loss": 1.0817,
+      "step": 816
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.7440546867734743,
+      "learning_rate": 1.5346854311550673e-05,
+      "loss": 1.2208,
+      "step": 817
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8567575651718746,
+      "learning_rate": 1.533543091101214e-05,
+      "loss": 1.1327,
+      "step": 818
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.4274156712589268,
+      "learning_rate": 1.532399776903124e-05,
+      "loss": 1.1327,
+      "step": 819
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7210491665005656,
+      "learning_rate": 1.5312554906482633e-05,
+      "loss": 1.1134,
+      "step": 820
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.825621703161565,
+      "learning_rate": 1.5301102344258724e-05,
+      "loss": 1.2225,
+      "step": 821
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 7.087429865201878,
+      "learning_rate": 1.5289640103269626e-05,
+      "loss": 1.132,
+      "step": 822
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8829506477209879,
+      "learning_rate": 1.5278168204443126e-05,
+      "loss": 1.0767,
+      "step": 823
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.429891088657107,
+      "learning_rate": 1.5266686668724647e-05,
+      "loss": 1.2257,
+      "step": 824
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7274596892325984,
+      "learning_rate": 1.5255195517077204e-05,
+      "loss": 1.1441,
+      "step": 825
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.2953779103125318,
+      "learning_rate": 1.5243694770481365e-05,
+      "loss": 1.079,
+      "step": 826
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9453070934974936,
+      "learning_rate": 1.523218444993522e-05,
+      "loss": 1.1749,
+      "step": 827
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.508559395602714,
+      "learning_rate": 1.5220664576454347e-05,
+      "loss": 1.1124,
+      "step": 828
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.409977473502115,
+      "learning_rate": 1.520913517107175e-05,
+      "loss": 1.2016,
+      "step": 829
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.4000036819459871,
+      "learning_rate": 1.5197596254837845e-05,
+      "loss": 1.1566,
+      "step": 830
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.0800540864720856,
+      "learning_rate": 1.5186047848820417e-05,
+      "loss": 1.1272,
+      "step": 831
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.20095783798415,
+      "learning_rate": 1.5174489974104574e-05,
+      "loss": 1.1842,
+      "step": 832
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.620015728847235,
+      "learning_rate": 1.5162922651792707e-05,
+      "loss": 1.1457,
+      "step": 833
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1976201239727056,
+      "learning_rate": 1.5151345903004457e-05,
+      "loss": 1.1501,
+      "step": 834
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.7682015942950335,
+      "learning_rate": 1.5139759748876687e-05,
+      "loss": 1.2129,
+      "step": 835
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.304753729089618,
+      "learning_rate": 1.5128164210563422e-05,
+      "loss": 1.2265,
+      "step": 836
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.9695440052867158,
+      "learning_rate": 1.5116559309235825e-05,
+      "loss": 1.0588,
+      "step": 837
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.017848948074111,
+      "learning_rate": 1.5104945066082146e-05,
+      "loss": 1.1726,
+      "step": 838
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.541574323580997,
+      "learning_rate": 1.5093321502307703e-05,
+      "loss": 1.1879,
+      "step": 839
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.106845250439184,
+      "learning_rate": 1.5081688639134821e-05,
+      "loss": 1.1191,
+      "step": 840
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.4077749098713799,
+      "learning_rate": 1.5070046497802819e-05,
+      "loss": 1.0638,
+      "step": 841
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.059927716983428,
+      "learning_rate": 1.5058395099567935e-05,
+      "loss": 1.2043,
+      "step": 842
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.3296102325273,
+      "learning_rate": 1.5046734465703328e-05,
+      "loss": 1.03,
+      "step": 843
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.243144653201744,
+      "learning_rate": 1.5035064617499005e-05,
+      "loss": 1.1892,
+      "step": 844
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 3.657643618322132,
+      "learning_rate": 1.5023385576261805e-05,
+      "loss": 1.1585,
+      "step": 845
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5829690554743803,
+      "learning_rate": 1.5011697363315347e-05,
+      "loss": 1.2336,
+      "step": 846
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.2258744412142177,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.1455,
+      "step": 847
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 5.936059841988962,
+      "learning_rate": 1.498829350767284e-05,
+      "loss": 1.1083,
+      "step": 848
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.5362877720000996,
+      "learning_rate": 1.4976577907707603e-05,
+      "loss": 1.0729,
+      "step": 849
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.472322669265795,
+      "learning_rate": 1.4964853221494667e-05,
+      "loss": 1.216,
+      "step": 850
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 2.4036521911621094,
+      "eval_runtime": 9.6209,
+      "eval_samples_per_second": 93.546,
+      "eval_steps_per_second": 0.208,
+      "step": 850
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.7652151363571775,
+      "learning_rate": 1.4953119470440977e-05,
+      "loss": 1.1561,
+      "step": 851
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.066882868258353,
+      "learning_rate": 1.4941376675970058e-05,
+      "loss": 1.1692,
+      "step": 852
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6628937860620643,
+      "learning_rate": 1.4929624859521925e-05,
+      "loss": 1.1715,
+      "step": 853
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.998597025093413,
+      "learning_rate": 1.4917864042553077e-05,
+      "loss": 1.1087,
+      "step": 854
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9616989524070805,
+      "learning_rate": 1.4906094246536435e-05,
+      "loss": 1.1407,
+      "step": 855
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.703085272253672,
+      "learning_rate": 1.4894315492961327e-05,
+      "loss": 1.1962,
+      "step": 856
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.1284604460679728,
+      "learning_rate": 1.4882527803333422e-05,
+      "loss": 1.0962,
+      "step": 857
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.8678195305823055,
+      "learning_rate": 1.487073119917472e-05,
+      "loss": 1.1571,
+      "step": 858
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.4752268829343833,
+      "learning_rate": 1.4858925702023481e-05,
+      "loss": 1.244,
+      "step": 859
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5267984063326554,
+      "learning_rate": 1.4847111333434214e-05,
+      "loss": 1.1508,
+      "step": 860
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7051600216291511,
+      "learning_rate": 1.483528811497762e-05,
+      "loss": 1.1563,
+      "step": 861
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7522546248147017,
+      "learning_rate": 1.4823456068240558e-05,
+      "loss": 1.0516,
+      "step": 862
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.3573036405997272,
+      "learning_rate": 1.4811615214826009e-05,
+      "loss": 1.1331,
+      "step": 863
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.8064048141123856,
+      "learning_rate": 1.479976557635303e-05,
+      "loss": 1.1414,
+      "step": 864
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.9705487704326785,
+      "learning_rate": 1.4787907174456719e-05,
+      "loss": 1.203,
+      "step": 865
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.6819915516822905,
+      "learning_rate": 1.4776040030788172e-05,
+      "loss": 1.1668,
+      "step": 866
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.3976950950892428,
+      "learning_rate": 1.4764164167014451e-05,
+      "loss": 1.1896,
+      "step": 867
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.4673610480719432,
+      "learning_rate": 1.4752279604818537e-05,
+      "loss": 1.1369,
+      "step": 868
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.7995435702442877,
+      "learning_rate": 1.474038636589929e-05,
+      "loss": 1.1704,
+      "step": 869
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.5188683272861574,
+      "learning_rate": 1.4728484471971411e-05,
+      "loss": 1.1024,
+      "step": 870
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.2135240947376742,
+      "learning_rate": 1.4716573944765413e-05,
+      "loss": 1.1643,
+      "step": 871
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 3.776601273590904,
+      "learning_rate": 1.4704654806027558e-05,
+      "loss": 1.1569,
+      "step": 872
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.1774321172509725,
+      "learning_rate": 1.469272707751984e-05,
+      "loss": 1.1426,
+      "step": 873
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8580262672068018,
+      "learning_rate": 1.4680790781019935e-05,
+      "loss": 1.0716,
+      "step": 874
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.343708751735701,
+      "learning_rate": 1.466884593832116e-05,
+      "loss": 1.1911,
+      "step": 875
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.156060245818154,
+      "learning_rate": 1.465689257123244e-05,
+      "loss": 1.1592,
+      "step": 876
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 3.815374727398419,
+      "learning_rate": 1.4644930701578252e-05,
+      "loss": 1.1793,
+      "step": 877
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2605759966289076,
+      "learning_rate": 1.463296035119862e-05,
+      "loss": 1.1829,
+      "step": 878
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6545977188571883,
+      "learning_rate": 1.4620981541949024e-05,
+      "loss": 1.1545,
+      "step": 879
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 5.830731028230518,
+      "learning_rate": 1.4608994295700414e-05,
+      "loss": 1.0596,
+      "step": 880
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.551723698967026,
+      "learning_rate": 1.4596998634339124e-05,
+      "loss": 1.1934,
+      "step": 881
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.2119040923278486,
+      "learning_rate": 1.4584994579766865e-05,
+      "loss": 1.1658,
+      "step": 882
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.5376317084452162,
+      "learning_rate": 1.4572982153900665e-05,
+      "loss": 1.1578,
+      "step": 883
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8708030863472591,
+      "learning_rate": 1.4560961378672845e-05,
+      "loss": 1.1609,
+      "step": 884
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 39.84893418699618,
+      "learning_rate": 1.4548932276030958e-05,
+      "loss": 1.2501,
+      "step": 885
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.1225531910401763,
+      "learning_rate": 1.4536894867937775e-05,
+      "loss": 1.1023,
+      "step": 886
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6937217642153866,
+      "learning_rate": 1.4524849176371219e-05,
+      "loss": 1.1611,
+      "step": 887
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.016398883796313,
+      "learning_rate": 1.4512795223324344e-05,
+      "loss": 1.1804,
+      "step": 888
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8839994166719624,
+      "learning_rate": 1.4500733030805286e-05,
+      "loss": 1.0826,
+      "step": 889
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.134776204038116,
+      "learning_rate": 1.4488662620837222e-05,
+      "loss": 1.1637,
+      "step": 890
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.0343821605274057,
+      "learning_rate": 1.4476584015458339e-05,
+      "loss": 1.0621,
+      "step": 891
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.9924732331723756,
+      "learning_rate": 1.4464497236721779e-05,
+      "loss": 1.2249,
+      "step": 892
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.9591402693403555,
+      "learning_rate": 1.4452402306695619e-05,
+      "loss": 1.1477,
+      "step": 893
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.295649035605896,
+      "learning_rate": 1.4440299247462802e-05,
+      "loss": 1.1441,
+      "step": 894
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.0279771420065449,
+      "learning_rate": 1.4428188081121125e-05,
+      "loss": 1.1234,
+      "step": 895
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.0546982955630777,
+      "learning_rate": 1.4416068829783187e-05,
+      "loss": 1.2424,
+      "step": 896
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.183578802300124,
+      "learning_rate": 1.4403941515576344e-05,
+      "loss": 1.0751,
+      "step": 897
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.9739979229519727,
+      "learning_rate": 1.4391806160642676e-05,
+      "loss": 1.2415,
+      "step": 898
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.4813286849602751,
+      "learning_rate": 1.4379662787138937e-05,
+      "loss": 1.0804,
+      "step": 899
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.0407041317235521,
+      "learning_rate": 1.4367511417236535e-05,
+      "loss": 1.147,
+      "step": 900
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 2.292776107788086,
+      "eval_runtime": 10.3505,
+      "eval_samples_per_second": 86.952,
+      "eval_steps_per_second": 0.193,
+      "step": 900
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8546601351572369,
+      "learning_rate": 1.4355352073121464e-05,
+      "loss": 1.178,
+      "step": 901
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8681308323362926,
+      "learning_rate": 1.4343184776994288e-05,
+      "loss": 1.1915,
+      "step": 902
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.958027760944528,
+      "learning_rate": 1.433100955107008e-05,
+      "loss": 1.1494,
+      "step": 903
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.9809006109796612,
+      "learning_rate": 1.43188264175784e-05,
+      "loss": 1.1543,
+      "step": 904
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.4454683478802788,
+      "learning_rate": 1.4306635398763237e-05,
+      "loss": 1.1579,
+      "step": 905
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.7426584953486786,
+      "learning_rate": 1.4294436516882993e-05,
+      "loss": 1.1957,
+      "step": 906
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.3725883771145653,
+      "learning_rate": 1.4282229794210404e-05,
+      "loss": 1.1141,
+      "step": 907
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.472953226478673,
+      "learning_rate": 1.4270015253032536e-05,
+      "loss": 1.21,
+      "step": 908
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8042466995780081,
+      "learning_rate": 1.4257792915650728e-05,
+      "loss": 1.1372,
+      "step": 909
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.803294783727885,
+      "learning_rate": 1.4245562804380555e-05,
+      "loss": 1.2038,
+      "step": 910
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.208392917197431,
+      "learning_rate": 1.4233324941551779e-05,
+      "loss": 1.0959,
+      "step": 911
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.4350523500680334,
+      "learning_rate": 1.422107934950832e-05,
+      "loss": 1.1913,
+      "step": 912
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.5452118241134134,
+      "learning_rate": 1.4208826050608213e-05,
+      "loss": 1.0601,
+      "step": 913
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.0278336228900262,
+      "learning_rate": 1.4196565067223553e-05,
+      "loss": 1.2266,
+      "step": 914
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8266593797425065,
+      "learning_rate": 1.4184296421740476e-05,
+      "loss": 1.1214,
+      "step": 915
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.052165499903148,
+      "learning_rate": 1.4172020136559106e-05,
+      "loss": 1.2383,
+      "step": 916
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.9278264182737059,
+      "learning_rate": 1.415973623409351e-05,
+      "loss": 1.1701,
+      "step": 917
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7689848628599244,
+      "learning_rate": 1.4147444736771666e-05,
+      "loss": 1.0586,
+      "step": 918
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8514458174907514,
+      "learning_rate": 1.4135145667035426e-05,
+      "loss": 1.013,
+      "step": 919
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8880240157846245,
+      "learning_rate": 1.4122839047340451e-05,
+      "loss": 1.2184,
+      "step": 920
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.827325240123563,
+      "learning_rate": 1.4110524900156205e-05,
+      "loss": 1.1588,
+      "step": 921
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8098342272373842,
+      "learning_rate": 1.4098203247965876e-05,
+      "loss": 1.1191,
+      "step": 922
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9785092179813124,
+      "learning_rate": 1.4085874113266376e-05,
+      "loss": 1.1437,
+      "step": 923
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9452584396474497,
+      "learning_rate": 1.4073537518568262e-05,
+      "loss": 1.2109,
+      "step": 924
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6989427876469572,
+      "learning_rate": 1.4061193486395723e-05,
+      "loss": 1.126,
+      "step": 925
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8749596760839845,
+      "learning_rate": 1.4048842039286518e-05,
+      "loss": 1.1318,
+      "step": 926
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7324587107003092,
+      "learning_rate": 1.4036483199791949e-05,
+      "loss": 1.2325,
+      "step": 927
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9792910198437413,
+      "learning_rate": 1.4024116990476815e-05,
+      "loss": 1.0693,
+      "step": 928
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9251047112874022,
+      "learning_rate": 1.4011743433919369e-05,
+      "loss": 1.1443,
+      "step": 929
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8483102161980987,
+      "learning_rate": 1.3999362552711281e-05,
+      "loss": 1.1547,
+      "step": 930
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.0199384657424166,
+      "learning_rate": 1.3986974369457595e-05,
+      "loss": 1.1599,
+      "step": 931
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5985970240150686,
+      "learning_rate": 1.3974578906776683e-05,
+      "loss": 1.1449,
+      "step": 932
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6879315290974664,
+      "learning_rate": 1.3962176187300209e-05,
+      "loss": 1.2004,
+      "step": 933
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7478255212371204,
+      "learning_rate": 1.3949766233673092e-05,
+      "loss": 1.1135,
+      "step": 934
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.78839387137221,
+      "learning_rate": 1.3937349068553448e-05,
+      "loss": 1.0698,
+      "step": 935
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.889913810448143,
+      "learning_rate": 1.3924924714612572e-05,
+      "loss": 1.206,
+      "step": 936
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7052142868553809,
+      "learning_rate": 1.3912493194534876e-05,
+      "loss": 1.1877,
+      "step": 937
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9202060158327793,
+      "learning_rate": 1.3900054531017858e-05,
+      "loss": 0.983,
+      "step": 938
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.726608073632099,
+      "learning_rate": 1.3887608746772064e-05,
+      "loss": 1.2339,
+      "step": 939
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8729371184146465,
+      "learning_rate": 1.3875155864521031e-05,
+      "loss": 1.099,
+      "step": 940
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8837230590784926,
+      "learning_rate": 1.3862695907001264e-05,
+      "loss": 1.2101,
+      "step": 941
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7574816452579946,
+      "learning_rate": 1.3850228896962178e-05,
+      "loss": 1.1896,
+      "step": 942
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.796111414872043,
+      "learning_rate": 1.383775485716608e-05,
+      "loss": 1.1969,
+      "step": 943
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8788489980697319,
+      "learning_rate": 1.3825273810388089e-05,
+      "loss": 1.0655,
+      "step": 944
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7590901896545448,
+      "learning_rate": 1.3812785779416138e-05,
+      "loss": 1.2198,
+      "step": 945
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7339643421951261,
+      "learning_rate": 1.3800290787050899e-05,
+      "loss": 1.0792,
+      "step": 946
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8799459082345832,
+      "learning_rate": 1.3787788856105762e-05,
+      "loss": 1.2523,
+      "step": 947
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.797026002653295,
+      "learning_rate": 1.3775280009406777e-05,
+      "loss": 1.1629,
+      "step": 948
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.5318739842395614,
+      "learning_rate": 1.3762764269792634e-05,
+      "loss": 1.207,
+      "step": 949
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7470307437851328,
+      "learning_rate": 1.3750241660114593e-05,
+      "loss": 0.9728,
+      "step": 950
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 2.257411241531372,
+      "eval_runtime": 9.0094,
+      "eval_samples_per_second": 99.896,
+      "eval_steps_per_second": 0.222,
+      "step": 950
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8093857401423998,
+      "learning_rate": 1.3737712203236465e-05,
+      "loss": 1.2328,
+      "step": 951
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7922856356115372,
+      "learning_rate": 1.3725175922034566e-05,
+      "loss": 0.9664,
+      "step": 952
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.9648043406273213,
+      "learning_rate": 1.3712632839397664e-05,
+      "loss": 1.2252,
+      "step": 953
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8575437229606504,
+      "learning_rate": 1.370008297822695e-05,
+      "loss": 1.1684,
+      "step": 954
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8118771917495644,
+      "learning_rate": 1.3687526361435989e-05,
+      "loss": 1.1221,
+      "step": 955
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.0918113741251987,
+      "learning_rate": 1.3674963011950685e-05,
+      "loss": 1.196,
+      "step": 956
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8036974414499339,
+      "learning_rate": 1.366239295270923e-05,
+      "loss": 1.1536,
+      "step": 957
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8902921771610973,
+      "learning_rate": 1.3649816206662063e-05,
+      "loss": 1.0907,
+      "step": 958
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.772800164448925,
+      "learning_rate": 1.3637232796771845e-05,
+      "loss": 1.1938,
+      "step": 959
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.9324894813780691,
+      "learning_rate": 1.362464274601339e-05,
+      "loss": 1.2018,
+      "step": 960
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7644281765996979,
+      "learning_rate": 1.3612046077373644e-05,
+      "loss": 1.1209,
+      "step": 961
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.4496247603390529,
+      "learning_rate": 1.3599442813851633e-05,
+      "loss": 1.144,
+      "step": 962
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8108935822567168,
+      "learning_rate": 1.3586832978458428e-05,
+      "loss": 1.1983,
+      "step": 963
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7820587065312029,
+      "learning_rate": 1.3574216594217094e-05,
+      "loss": 1.1244,
+      "step": 964
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.910163607745879,
+      "learning_rate": 1.3561593684162655e-05,
+      "loss": 1.2064,
+      "step": 965
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7527179188302057,
+      "learning_rate": 1.3548964271342051e-05,
+      "loss": 1.1403,
+      "step": 966
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7740932651583183,
+      "learning_rate": 1.3536328378814094e-05,
+      "loss": 1.093,
+      "step": 967
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7599277683714987,
+      "learning_rate": 1.3523686029649423e-05,
+      "loss": 1.2585,
+      "step": 968
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7565782394718524,
+      "learning_rate": 1.3511037246930476e-05,
+      "loss": 0.9898,
+      "step": 969
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7458506461671065,
+      "learning_rate": 1.3498382053751423e-05,
+      "loss": 1.1692,
+      "step": 970
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8607280239673782,
+      "learning_rate": 1.3485720473218153e-05,
+      "loss": 1.1534,
+      "step": 971
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8315028572122712,
+      "learning_rate": 1.3473052528448203e-05,
+      "loss": 1.2169,
+      "step": 972
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7895558838371804,
+      "learning_rate": 1.3460378242570741e-05,
+      "loss": 1.1291,
+      "step": 973
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9980707107440745,
+      "learning_rate": 1.3447697638726505e-05,
+      "loss": 1.2036,
+      "step": 974
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.3056349380000984,
+      "learning_rate": 1.3435010740067773e-05,
+      "loss": 1.1035,
+      "step": 975
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.0159928078155969,
+      "learning_rate": 1.3422317569758315e-05,
+      "loss": 1.1777,
+      "step": 976
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9751023649206882,
+      "learning_rate": 1.3409618150973349e-05,
+      "loss": 1.1006,
+      "step": 977
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.975325940507271,
+      "learning_rate": 1.3396912506899508e-05,
+      "loss": 1.1906,
+      "step": 978
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5165504478487877,
+      "learning_rate": 1.3384200660734779e-05,
+      "loss": 1.1199,
+      "step": 979
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8995323228482025,
+      "learning_rate": 1.337148263568849e-05,
+      "loss": 1.1672,
+      "step": 980
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8610202830537883,
+      "learning_rate": 1.3358758454981237e-05,
+      "loss": 1.0884,
+      "step": 981
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7724910795365619,
+      "learning_rate": 1.334602814184486e-05,
+      "loss": 1.1692,
+      "step": 982
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9481829343400625,
+      "learning_rate": 1.3333291719522392e-05,
+      "loss": 1.0854,
+      "step": 983
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9572504285581873,
+      "learning_rate": 1.332054921126803e-05,
+      "loss": 1.2283,
+      "step": 984
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.4287920324734869,
+      "learning_rate": 1.3307800640347065e-05,
+      "loss": 1.0901,
+      "step": 985
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.0618482446201651,
+      "learning_rate": 1.3295046030035881e-05,
+      "loss": 1.1974,
+      "step": 986
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9907786476914261,
+      "learning_rate": 1.3282285403621864e-05,
+      "loss": 1.0789,
+      "step": 987
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.832804493441769,
+      "learning_rate": 1.32695187844034e-05,
+      "loss": 1.1846,
+      "step": 988
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9042658575889998,
+      "learning_rate": 1.3256746195689817e-05,
+      "loss": 0.9632,
+      "step": 989
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8448969745553127,
+      "learning_rate": 1.324396766080133e-05,
+      "loss": 1.1959,
+      "step": 990
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8760128287647069,
+      "learning_rate": 1.3231183203069025e-05,
+      "loss": 1.0835,
+      "step": 991
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9676380253184061,
+      "learning_rate": 1.3218392845834789e-05,
+      "loss": 1.1925,
+      "step": 992
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.0842778443995977,
+      "learning_rate": 1.3205596612451291e-05,
+      "loss": 1.165,
+      "step": 993
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9550138190219394,
+      "learning_rate": 1.319279452628192e-05,
+      "loss": 1.1511,
+      "step": 994
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7961990214400743,
+      "learning_rate": 1.3179986610700761e-05,
+      "loss": 1.1706,
+      "step": 995
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.0157480786273974,
+      "learning_rate": 1.316717288909253e-05,
+      "loss": 1.1779,
+      "step": 996
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.2155495798135376,
+      "learning_rate": 1.3154353384852559e-05,
+      "loss": 0.9958,
+      "step": 997
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7883901001563501,
+      "learning_rate": 1.3141528121386717e-05,
+      "loss": 1.1583,
+      "step": 998
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.0063197124318375,
+      "learning_rate": 1.312869712211141e-05,
+      "loss": 1.2682,
+      "step": 999
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8466757657780394,
+      "learning_rate": 1.3115860410453504e-05,
+      "loss": 1.0945,
+      "step": 1000
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 2.3478591442108154,
+      "eval_runtime": 9.9566,
+      "eval_samples_per_second": 90.392,
+      "eval_steps_per_second": 0.201,
+      "step": 1000
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.907561422565562,
+      "learning_rate": 1.3103018009850298e-05,
+      "loss": 1.0309,
+      "step": 1001
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7574283257642647,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 1.2295,
+      "step": 1002
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.2241079640492647,
+      "learning_rate": 1.3077316235609065e-05,
+      "loss": 1.2587,
+      "step": 1003
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8617233227342772,
+      "learning_rate": 1.3064456908897404e-05,
+      "loss": 1.0951,
+      "step": 1004
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7969870427575209,
+      "learning_rate": 1.3051591987093075e-05,
+      "loss": 1.1928,
+      "step": 1005
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8679000145011235,
+      "learning_rate": 1.3038721493684885e-05,
+      "loss": 1.1098,
+      "step": 1006
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.9194622069197191,
+      "learning_rate": 1.3025845452171808e-05,
+      "loss": 1.1495,
+      "step": 1007
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7992157516502462,
+      "learning_rate": 1.3012963886062958e-05,
+      "loss": 1.1051,
+      "step": 1008
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.9480643762699184,
+      "learning_rate": 1.300007681887752e-05,
+      "loss": 1.2341,
+      "step": 1009
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8180823002718888,
+      "learning_rate": 1.2987184274144742e-05,
+      "loss": 1.1158,
+      "step": 1010
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.808910501106571,
+      "learning_rate": 1.297428627540385e-05,
+      "loss": 1.1954,
+      "step": 1011
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.9436876817431267,
+      "learning_rate": 1.2961382846204056e-05,
+      "loss": 1.1498,
+      "step": 1012
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8060283058513789,
+      "learning_rate": 1.2948474010104456e-05,
+      "loss": 1.1277,
+      "step": 1013
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8589805877269069,
+      "learning_rate": 1.2935559790674045e-05,
+      "loss": 1.0684,
+      "step": 1014
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8099106164377641,
+      "learning_rate": 1.292264021149163e-05,
+      "loss": 1.1956,
+      "step": 1015
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8111802087635311,
+      "learning_rate": 1.2909715296145809e-05,
+      "loss": 1.0237,
+      "step": 1016
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7367671781501922,
+      "learning_rate": 1.2896785068234925e-05,
+      "loss": 1.2334,
+      "step": 1017
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7712269206896131,
+      "learning_rate": 1.2883849551367016e-05,
+      "loss": 1.1228,
+      "step": 1018
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7689502608311273,
+      "learning_rate": 1.2870908769159785e-05,
+      "loss": 1.1431,
+      "step": 1019
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8698743550904432,
+      "learning_rate": 1.2857962745240533e-05,
+      "loss": 1.0299,
+      "step": 1020
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8018675472117559,
+      "learning_rate": 1.2845011503246152e-05,
+      "loss": 1.2043,
+      "step": 1021
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7656390668992483,
+      "learning_rate": 1.283205506682304e-05,
+      "loss": 1.0978,
+      "step": 1022
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.957555307456909,
+      "learning_rate": 1.2819093459627099e-05,
+      "loss": 1.2109,
+      "step": 1023
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8166669182012868,
+      "learning_rate": 1.2806126705323658e-05,
+      "loss": 1.0995,
+      "step": 1024
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7296987640672141,
+      "learning_rate": 1.2793154827587441e-05,
+      "loss": 1.1743,
+      "step": 1025
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.9447411902256738,
+      "learning_rate": 1.2780177850102546e-05,
+      "loss": 1.0794,
+      "step": 1026
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7622173072731168,
+      "learning_rate": 1.2767195796562359e-05,
+      "loss": 1.2334,
+      "step": 1027
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8910074021454265,
+      "learning_rate": 1.275420869066955e-05,
+      "loss": 1.0629,
+      "step": 1028
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.90265203620373,
+      "learning_rate": 1.2741216556136e-05,
+      "loss": 1.1792,
+      "step": 1029
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8895544811753036,
+      "learning_rate": 1.2728219416682788e-05,
+      "loss": 1.0737,
+      "step": 1030
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7526720487620013,
+      "learning_rate": 1.2715217296040114e-05,
+      "loss": 1.1657,
+      "step": 1031
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.3819334324384713,
+      "learning_rate": 1.2702210217947289e-05,
+      "loss": 1.1812,
+      "step": 1032
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.032497130448781,
+      "learning_rate": 1.2689198206152657e-05,
+      "loss": 1.1849,
+      "step": 1033
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8623021595904501,
+      "learning_rate": 1.2676181284413591e-05,
+      "loss": 1.0766,
+      "step": 1034
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.9520002645483374,
+      "learning_rate": 1.2663159476496408e-05,
+      "loss": 1.1887,
+      "step": 1035
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7416630949886722,
+      "learning_rate": 1.265013280617636e-05,
+      "loss": 1.1223,
+      "step": 1036
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8223772753290941,
+      "learning_rate": 1.263710129723757e-05,
+      "loss": 1.0888,
+      "step": 1037
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8580294869418439,
+      "learning_rate": 1.2624064973473003e-05,
+      "loss": 1.1992,
+      "step": 1038
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7602795780033008,
+      "learning_rate": 1.2611023858684403e-05,
+      "loss": 1.1196,
+      "step": 1039
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7977858701113917,
+      "learning_rate": 1.2597977976682274e-05,
+      "loss": 1.0667,
+      "step": 1040
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8441398285367918,
+      "learning_rate": 1.2584927351285816e-05,
+      "loss": 1.0814,
+      "step": 1041
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8818575748617924,
+      "learning_rate": 1.257187200632289e-05,
+      "loss": 1.1637,
+      "step": 1042
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7531116423664139,
+      "learning_rate": 1.2558811965629975e-05,
+      "loss": 1.0973,
+      "step": 1043
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8561576175621228,
+      "learning_rate": 1.2545747253052127e-05,
+      "loss": 1.1938,
+      "step": 1044
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.371504772630903,
+      "learning_rate": 1.2532677892442932e-05,
+      "loss": 1.0777,
+      "step": 1045
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.279919025442849,
+      "learning_rate": 1.2519603907664453e-05,
+      "loss": 1.1606,
+      "step": 1046
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8822277803029139,
+      "learning_rate": 1.2506525322587207e-05,
+      "loss": 1.1779,
+      "step": 1047
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8772926076322269,
+      "learning_rate": 1.2493442161090101e-05,
+      "loss": 1.0677,
+      "step": 1048
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8124334247184427,
+      "learning_rate": 1.2480354447060407e-05,
+      "loss": 1.1291,
+      "step": 1049
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.0701371806304458,
+      "learning_rate": 1.2467262204393702e-05,
+      "loss": 1.2544,
+      "step": 1050
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 2.3446199893951416,
+      "eval_runtime": 9.1099,
+      "eval_samples_per_second": 98.794,
+      "eval_steps_per_second": 0.22,
+      "step": 1050
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8229697666751745,
+      "learning_rate": 1.2454165456993832e-05,
+      "loss": 1.0712,
+      "step": 1051
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.1890990596638462,
+      "learning_rate": 1.2441064228772874e-05,
+      "loss": 1.197,
+      "step": 1052
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.847757059905921,
+      "learning_rate": 1.2427958543651076e-05,
+      "loss": 1.1132,
+      "step": 1053
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7739932711778372,
+      "learning_rate": 1.2414848425556832e-05,
+      "loss": 1.1613,
+      "step": 1054
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8641421445168057,
+      "learning_rate": 1.2401733898426624e-05,
+      "loss": 1.0973,
+      "step": 1055
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7336746172489954,
+      "learning_rate": 1.2388614986204992e-05,
+      "loss": 1.1808,
+      "step": 1056
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.9199614340574109,
+      "learning_rate": 1.2375491712844472e-05,
+      "loss": 1.1285,
+      "step": 1057
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8956766574433975,
+      "learning_rate": 1.2362364102305568e-05,
+      "loss": 1.196,
+      "step": 1058
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7859975762752834,
+      "learning_rate": 1.2349232178556703e-05,
+      "loss": 1.0494,
+      "step": 1059
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7725627114176152,
+      "learning_rate": 1.2336095965574178e-05,
+      "loss": 1.2205,
+      "step": 1060
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7490573994079405,
+      "learning_rate": 1.2322955487342114e-05,
+      "loss": 1.0918,
+      "step": 1061
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7040662978358799,
+      "learning_rate": 1.2309810767852435e-05,
+      "loss": 1.1672,
+      "step": 1062
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8192083366626096,
+      "learning_rate": 1.2296661831104796e-05,
+      "loss": 1.0558,
+      "step": 1063
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8252131852072122,
+      "learning_rate": 1.2283508701106559e-05,
+      "loss": 1.0773,
+      "step": 1064
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8098936956254087,
+      "learning_rate": 1.2270351401872738e-05,
+      "loss": 1.169,
+      "step": 1065
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7167893900153028,
+      "learning_rate": 1.2257189957425964e-05,
+      "loss": 1.1557,
+      "step": 1066
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7850214489373033,
+      "learning_rate": 1.2244024391796432e-05,
+      "loss": 1.0752,
+      "step": 1067
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8012668925965657,
+      "learning_rate": 1.223085472902186e-05,
+      "loss": 1.201,
+      "step": 1068
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7944437182028166,
+      "learning_rate": 1.2217680993147454e-05,
+      "loss": 1.1443,
+      "step": 1069
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7431182385155672,
+      "learning_rate": 1.2204503208225848e-05,
+      "loss": 1.1238,
+      "step": 1070
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8329766777075324,
+      "learning_rate": 1.2191321398317078e-05,
+      "loss": 1.2134,
+      "step": 1071
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6968488527422046,
+      "learning_rate": 1.2178135587488515e-05,
+      "loss": 1.0365,
+      "step": 1072
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7304995703187777,
+      "learning_rate": 1.2164945799814856e-05,
+      "loss": 1.0769,
+      "step": 1073
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7372235149797388,
+      "learning_rate": 1.2151752059378033e-05,
+      "loss": 1.2043,
+      "step": 1074
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8378405185339788,
+      "learning_rate": 1.2138554390267219e-05,
+      "loss": 1.1268,
+      "step": 1075
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7110033009924095,
+      "learning_rate": 1.2125352816578745e-05,
+      "loss": 1.173,
+      "step": 1076
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8003712965450256,
+      "learning_rate": 1.2112147362416076e-05,
+      "loss": 1.1475,
+      "step": 1077
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8470753045858789,
+      "learning_rate": 1.2098938051889761e-05,
+      "loss": 1.1428,
+      "step": 1078
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7762246243618284,
+      "learning_rate": 1.2085724909117388e-05,
+      "loss": 1.142,
+      "step": 1079
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8662868512082409,
+      "learning_rate": 1.2072507958223551e-05,
+      "loss": 1.0944,
+      "step": 1080
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.723305350605004,
+      "learning_rate": 1.2059287223339784e-05,
+      "loss": 1.2027,
+      "step": 1081
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.718507937661056,
+      "learning_rate": 1.204606272860454e-05,
+      "loss": 1.1384,
+      "step": 1082
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.815234090582479,
+      "learning_rate": 1.2032834498163128e-05,
+      "loss": 1.1909,
+      "step": 1083
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8180065609889281,
+      "learning_rate": 1.201960255616769e-05,
+      "loss": 1.1452,
+      "step": 1084
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7347269976443872,
+      "learning_rate": 1.200636692677713e-05,
+      "loss": 1.1242,
+      "step": 1085
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.5852052896482143,
+      "learning_rate": 1.1993127634157089e-05,
+      "loss": 1.0403,
+      "step": 1086
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8114931734305559,
+      "learning_rate": 1.1979884702479909e-05,
+      "loss": 1.2556,
+      "step": 1087
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.797855040800368,
+      "learning_rate": 1.196663815592456e-05,
+      "loss": 1.0966,
+      "step": 1088
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.787738405164882,
+      "learning_rate": 1.1953388018676618e-05,
+      "loss": 1.1523,
+      "step": 1089
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.364310490953006,
+      "learning_rate": 1.1940134314928211e-05,
+      "loss": 1.1374,
+      "step": 1090
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7738819909844704,
+      "learning_rate": 1.1926877068877993e-05,
+      "loss": 1.1347,
+      "step": 1091
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7324173782000204,
+      "learning_rate": 1.1913616304731064e-05,
+      "loss": 1.002,
+      "step": 1092
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7677110324277575,
+      "learning_rate": 1.1900352046698969e-05,
+      "loss": 1.1336,
+      "step": 1093
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8211467167919759,
+      "learning_rate": 1.188708431899961e-05,
+      "loss": 1.1588,
+      "step": 1094
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7584621225298247,
+      "learning_rate": 1.187381314585725e-05,
+      "loss": 1.156,
+      "step": 1095
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.310178936405684,
+      "learning_rate": 1.1860538551502412e-05,
+      "loss": 1.103,
+      "step": 1096
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6963180178723775,
+      "learning_rate": 1.1847260560171895e-05,
+      "loss": 1.1331,
+      "step": 1097
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.0391433609712613,
+      "learning_rate": 1.183397919610868e-05,
+      "loss": 1.0832,
+      "step": 1098
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8182976572356938,
+      "learning_rate": 1.1820694483561916e-05,
+      "loss": 1.2149,
+      "step": 1099
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7621215400567337,
+      "learning_rate": 1.1807406446786857e-05,
+      "loss": 1.0988,
+      "step": 1100
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 2.367137908935547,
+      "eval_runtime": 13.1398,
+      "eval_samples_per_second": 68.494,
+      "eval_steps_per_second": 0.152,
+      "step": 1100
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8019916196911122,
+      "learning_rate": 1.1794115110044842e-05,
+      "loss": 1.1836,
+      "step": 1101
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.894722802097533,
+      "learning_rate": 1.1780820497603215e-05,
+      "loss": 1.1255,
+      "step": 1102
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8161313867377805,
+      "learning_rate": 1.176752263373532e-05,
+      "loss": 1.1864,
+      "step": 1103
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.91665495892511,
+      "learning_rate": 1.175422154272042e-05,
+      "loss": 1.1898,
+      "step": 1104
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8163572975562096,
+      "learning_rate": 1.1740917248843682e-05,
+      "loss": 1.0716,
+      "step": 1105
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8609872247036563,
+      "learning_rate": 1.1727609776396118e-05,
+      "loss": 1.0915,
+      "step": 1106
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7638368301172208,
+      "learning_rate": 1.1714299149674538e-05,
+      "loss": 1.2436,
+      "step": 1107
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.558708664039171,
+      "learning_rate": 1.1700985392981521e-05,
+      "loss": 1.1498,
+      "step": 1108
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.0604863981114014,
+      "learning_rate": 1.1687668530625356e-05,
+      "loss": 1.1002,
+      "step": 1109
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7561596475633794,
+      "learning_rate": 1.1674348586919997e-05,
+      "loss": 1.1723,
+      "step": 1110
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.1991700669825418,
+      "learning_rate": 1.1661025586185034e-05,
+      "loss": 1.0641,
+      "step": 1111
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.2722084640192937,
+      "learning_rate": 1.1647699552745628e-05,
+      "loss": 1.2043,
+      "step": 1112
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.9356323603255384,
+      "learning_rate": 1.1634370510932487e-05,
+      "loss": 1.0607,
+      "step": 1113
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8148340506321151,
+      "learning_rate": 1.1621038485081803e-05,
+      "loss": 1.0778,
+      "step": 1114
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.144023372976583,
+      "learning_rate": 1.1607703499535223e-05,
+      "loss": 1.1168,
+      "step": 1115
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8491589967076527,
+      "learning_rate": 1.1594365578639792e-05,
+      "loss": 1.1519,
+      "step": 1116
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8958573426135128,
+      "learning_rate": 1.1581024746747925e-05,
+      "loss": 1.1377,
+      "step": 1117
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8296171745831215,
+      "learning_rate": 1.1567681028217335e-05,
+      "loss": 1.1198,
+      "step": 1118
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8832373837371578,
+      "learning_rate": 1.1554334447411019e-05,
+      "loss": 1.1198,
+      "step": 1119
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.368300866842442,
+      "learning_rate": 1.1540985028697196e-05,
+      "loss": 1.1635,
+      "step": 1120
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8360908052203736,
+      "learning_rate": 1.1527632796449268e-05,
+      "loss": 1.0912,
+      "step": 1121
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.9500319802905395,
+      "learning_rate": 1.1514277775045768e-05,
+      "loss": 1.2335,
+      "step": 1122
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.9176067978363559,
+      "learning_rate": 1.150091998887033e-05,
+      "loss": 1.0487,
+      "step": 1123
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7809701670720635,
+      "learning_rate": 1.1487559462311626e-05,
+      "loss": 1.1561,
+      "step": 1124
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7983467803665704,
+      "learning_rate": 1.1474196219763345e-05,
+      "loss": 1.1287,
+      "step": 1125
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8125641241265754,
+      "learning_rate": 1.1460830285624119e-05,
+      "loss": 1.1483,
+      "step": 1126
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.0063774453125074,
+      "learning_rate": 1.1447461684297505e-05,
+      "loss": 1.1075,
+      "step": 1127
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.9649531556938787,
+      "learning_rate": 1.1434090440191927e-05,
+      "loss": 1.1779,
+      "step": 1128
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.034794575363884,
+      "learning_rate": 1.1420716577720634e-05,
+      "loss": 1.1087,
+      "step": 1129
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5486431289732978,
+      "learning_rate": 1.140734012130166e-05,
+      "loss": 1.1457,
+      "step": 1130
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.2789062524398525,
+      "learning_rate": 1.1393961095357761e-05,
+      "loss": 1.1039,
+      "step": 1131
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.9945909272174722,
+      "learning_rate": 1.1380579524316406e-05,
+      "loss": 1.1937,
+      "step": 1132
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.9809806685440203,
+      "learning_rate": 1.1367195432609687e-05,
+      "loss": 1.1093,
+      "step": 1133
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.831861388258603,
+      "learning_rate": 1.1353808844674321e-05,
+      "loss": 1.1724,
+      "step": 1134
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.0109619800641096,
+      "learning_rate": 1.1340419784951564e-05,
+      "loss": 1.0883,
+      "step": 1135
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8593792851679111,
+      "learning_rate": 1.13270282778872e-05,
+      "loss": 1.1877,
+      "step": 1136
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.749553671395514,
+      "learning_rate": 1.1313634347931466e-05,
+      "loss": 1.1016,
+      "step": 1137
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8498081421865818,
+      "learning_rate": 1.1300238019539039e-05,
+      "loss": 1.2025,
+      "step": 1138
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8994209609272072,
+      "learning_rate": 1.1286839317168958e-05,
+      "loss": 1.1086,
+      "step": 1139
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.3903838155432093,
+      "learning_rate": 1.1273438265284615e-05,
+      "loss": 0.9927,
+      "step": 1140
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8247636305647101,
+      "learning_rate": 1.1260034888353676e-05,
+      "loss": 1.1327,
+      "step": 1141
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.1720545387143053,
+      "learning_rate": 1.1246629210848062e-05,
+      "loss": 1.1602,
+      "step": 1142
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.9935419780249579,
+      "learning_rate": 1.1233221257243888e-05,
+      "loss": 1.0827,
+      "step": 1143
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8501955090522114,
+      "learning_rate": 1.1219811052021425e-05,
+      "loss": 1.155,
+      "step": 1144
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.9420243148219543,
+      "learning_rate": 1.1206398619665067e-05,
+      "loss": 1.1911,
+      "step": 1145
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.1597333251115307,
+      "learning_rate": 1.1192983984663253e-05,
+      "loss": 1.0713,
+      "step": 1146
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7937116148319214,
+      "learning_rate": 1.1179567171508463e-05,
+      "loss": 1.2574,
+      "step": 1147
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.978664047750511,
+      "learning_rate": 1.1166148204697142e-05,
+      "loss": 0.9721,
+      "step": 1148
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.9587466796446014,
+      "learning_rate": 1.1152727108729675e-05,
+      "loss": 1.0989,
+      "step": 1149
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7906190139395348,
+      "learning_rate": 1.1139303908110326e-05,
+      "loss": 1.1098,
+      "step": 1150
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 2.4341394901275635,
+      "eval_runtime": 9.3827,
+      "eval_samples_per_second": 95.922,
+      "eval_steps_per_second": 0.213,
+      "step": 1150
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.747043777260365,
+      "learning_rate": 1.112587862734721e-05,
+      "loss": 1.1816,
+      "step": 1151
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.9794381138411494,
+      "learning_rate": 1.1112451290952238e-05,
+      "loss": 1.1191,
+      "step": 1152
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.9508718473380492,
+      "learning_rate": 1.1099021923441066e-05,
+      "loss": 1.1376,
+      "step": 1153
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8097297883710646,
+      "learning_rate": 1.1085590549333074e-05,
+      "loss": 1.0638,
+      "step": 1154
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8393600500568786,
+      "learning_rate": 1.1072157193151289e-05,
+      "loss": 1.1873,
+      "step": 1155
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.0646216354707199,
+      "learning_rate": 1.1058721879422368e-05,
+      "loss": 1.1311,
+      "step": 1156
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.908173154142926,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 1.1225,
+      "step": 1157
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7875442239966034,
+      "learning_rate": 1.1031845477447554e-05,
+      "loss": 1.0384,
+      "step": 1158
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8887667314256025,
+      "learning_rate": 1.1018404438272657e-05,
+      "loss": 1.2224,
+      "step": 1159
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.9841864272582145,
+      "learning_rate": 1.1004961539692533e-05,
+      "loss": 1.1603,
+      "step": 1160
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.1653480737983506,
+      "learning_rate": 1.0991516806251252e-05,
+      "loss": 1.112,
+      "step": 1161
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.988133772567338,
+      "learning_rate": 1.0978070262496248e-05,
+      "loss": 1.0029,
+      "step": 1162
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8190267417166507,
+      "learning_rate": 1.0964621932978244e-05,
+      "loss": 1.1875,
+      "step": 1163
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8914166894777024,
+      "learning_rate": 1.0951171842251236e-05,
+      "loss": 1.0739,
+      "step": 1164
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8778158258434946,
+      "learning_rate": 1.0937720014872428e-05,
+      "loss": 1.1486,
+      "step": 1165
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7844493639332003,
+      "learning_rate": 1.0924266475402205e-05,
+      "loss": 1.1585,
+      "step": 1166
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8955842810890933,
+      "learning_rate": 1.0910811248404064e-05,
+      "loss": 1.1793,
+      "step": 1167
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.13721696661471,
+      "learning_rate": 1.0897354358444596e-05,
+      "loss": 1.1096,
+      "step": 1168
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8204604705073265,
+      "learning_rate": 1.0883895830093419e-05,
+      "loss": 1.1632,
+      "step": 1169
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7560463285674393,
+      "learning_rate": 1.0870435687923144e-05,
+      "loss": 1.0814,
+      "step": 1170
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8856840881262957,
+      "learning_rate": 1.0856973956509334e-05,
+      "loss": 1.1641,
+      "step": 1171
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8118906521493523,
+      "learning_rate": 1.0843510660430447e-05,
+      "loss": 1.1724,
+      "step": 1172
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8110115227110531,
+      "learning_rate": 1.0830045824267808e-05,
+      "loss": 1.1291,
+      "step": 1173
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8832281541484082,
+      "learning_rate": 1.0816579472605538e-05,
+      "loss": 1.1593,
+      "step": 1174
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8661269555966843,
+      "learning_rate": 1.080311163003054e-05,
+      "loss": 1.1469,
+      "step": 1175
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7302154898061033,
+      "learning_rate": 1.0789642321132427e-05,
+      "loss": 1.1554,
+      "step": 1176
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8824836986484067,
+      "learning_rate": 1.07761715705035e-05,
+      "loss": 1.1162,
+      "step": 1177
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.89099621286632,
+      "learning_rate": 1.0762699402738686e-05,
+      "loss": 1.2121,
+      "step": 1178
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8413186566016712,
+      "learning_rate": 1.0749225842435498e-05,
+      "loss": 1.045,
+      "step": 1179
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8323158390722104,
+      "learning_rate": 1.0735750914193998e-05,
+      "loss": 1.115,
+      "step": 1180
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7944192947083184,
+      "learning_rate": 1.0722274642616739e-05,
+      "loss": 1.1975,
+      "step": 1181
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.781937124707675,
+      "learning_rate": 1.070879705230873e-05,
+      "loss": 1.1834,
+      "step": 1182
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8358750916848625,
+      "learning_rate": 1.0695318167877388e-05,
+      "loss": 1.0834,
+      "step": 1183
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.9673849505783747,
+      "learning_rate": 1.0681838013932495e-05,
+      "loss": 1.1375,
+      "step": 1184
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8260023991161364,
+      "learning_rate": 1.0668356615086142e-05,
+      "loss": 1.0738,
+      "step": 1185
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.927397745554387,
+      "learning_rate": 1.0654873995952703e-05,
+      "loss": 1.1497,
+      "step": 1186
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.4630474108979477,
+      "learning_rate": 1.0641390181148772e-05,
+      "loss": 1.0789,
+      "step": 1187
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7687586800929149,
+      "learning_rate": 1.0627905195293135e-05,
+      "loss": 1.108,
+      "step": 1188
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.0298360032239149,
+      "learning_rate": 1.0614419063006708e-05,
+      "loss": 1.1185,
+      "step": 1189
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7529342529749423,
+      "learning_rate": 1.0600931808912504e-05,
+      "loss": 1.2268,
+      "step": 1190
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6636625024803321,
+      "learning_rate": 1.0587443457635583e-05,
+      "loss": 0.9264,
+      "step": 1191
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7573210173177188,
+      "learning_rate": 1.0573954033803006e-05,
+      "loss": 1.1978,
+      "step": 1192
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.9226383645357255,
+      "learning_rate": 1.0560463562043803e-05,
+      "loss": 1.1283,
+      "step": 1193
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7700304147803854,
+      "learning_rate": 1.0546972066988901e-05,
+      "loss": 1.1121,
+      "step": 1194
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7352932061576003,
+      "learning_rate": 1.0533479573271109e-05,
+      "loss": 1.1622,
+      "step": 1195
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7461944127774371,
+      "learning_rate": 1.0519986105525046e-05,
+      "loss": 1.1704,
+      "step": 1196
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7401854584820325,
+      "learning_rate": 1.0506491688387128e-05,
+      "loss": 1.0659,
+      "step": 1197
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.0961399434866708,
+      "learning_rate": 1.0492996346495485e-05,
+      "loss": 1.2204,
+      "step": 1198
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7833609666386051,
+      "learning_rate": 1.047950010448995e-05,
+      "loss": 1.0904,
+      "step": 1199
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8167593964628238,
+      "learning_rate": 1.0466002987011989e-05,
+      "loss": 1.1812,
+      "step": 1200
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 2.3048744201660156,
+      "eval_runtime": 9.0115,
+      "eval_samples_per_second": 99.873,
+      "eval_steps_per_second": 0.222,
+      "step": 1200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8748020286471742,
+      "learning_rate": 1.0452505018704677e-05,
+      "loss": 1.0757,
+      "step": 1201
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8978062250463941,
+      "learning_rate": 1.0439006224212629e-05,
+      "loss": 1.2204,
+      "step": 1202
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.9242318463143305,
+      "learning_rate": 1.042550662818198e-05,
+      "loss": 1.0698,
+      "step": 1203
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.3438185857465446,
+      "learning_rate": 1.0412006255260325e-05,
+      "loss": 1.1956,
+      "step": 1204
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8261017682830516,
+      "learning_rate": 1.0398505130096674e-05,
+      "loss": 1.0585,
+      "step": 1205
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.932150286875676,
+      "learning_rate": 1.0385003277341417e-05,
+      "loss": 1.1586,
+      "step": 1206
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7074776956946309,
+      "learning_rate": 1.037150072164626e-05,
+      "loss": 1.0731,
+      "step": 1207
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8748894554291688,
+      "learning_rate": 1.0357997487664215e-05,
+      "loss": 1.1859,
+      "step": 1208
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7665690181158847,
+      "learning_rate": 1.0344493600049509e-05,
+      "loss": 1.0743,
+      "step": 1209
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7439422828534031,
+      "learning_rate": 1.0330989083457572e-05,
+      "loss": 1.1926,
+      "step": 1210
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.766736715076788,
+      "learning_rate": 1.0317483962544986e-05,
+      "loss": 1.1277,
+      "step": 1211
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7442210599398388,
+      "learning_rate": 1.030397826196943e-05,
+      "loss": 1.125,
+      "step": 1212
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.671930124390854,
+      "learning_rate": 1.0290472006389645e-05,
+      "loss": 1.1476,
+      "step": 1213
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8622521981856319,
+      "learning_rate": 1.0276965220465382e-05,
+      "loss": 1.1519,
+      "step": 1214
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7117585184127782,
+      "learning_rate": 1.0263457928857368e-05,
+      "loss": 1.1278,
+      "step": 1215
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.794637302518549,
+      "learning_rate": 1.0249950156227245e-05,
+      "loss": 1.1097,
+      "step": 1216
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.801840122199367,
+      "learning_rate": 1.0236441927237534e-05,
+      "loss": 1.0262,
+      "step": 1217
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8963548279141618,
+      "learning_rate": 1.0222933266551594e-05,
+      "loss": 1.1372,
+      "step": 1218
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.75544861015887,
+      "learning_rate": 1.0209424198833571e-05,
+      "loss": 1.1145,
+      "step": 1219
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.9322810631881627,
+      "learning_rate": 1.0195914748748348e-05,
+      "loss": 1.196,
+      "step": 1220
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7529044582696209,
+      "learning_rate": 1.0182404940961517e-05,
+      "loss": 1.1245,
+      "step": 1221
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7126939266008184,
+      "learning_rate": 1.0168894800139311e-05,
+      "loss": 1.1228,
+      "step": 1222
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8660271380160565,
+      "learning_rate": 1.0155384350948583e-05,
+      "loss": 1.1904,
+      "step": 1223
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.0302652721889913,
+      "learning_rate": 1.0141873618056736e-05,
+      "loss": 1.0482,
+      "step": 1224
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7176854199918694,
+      "learning_rate": 1.0128362626131705e-05,
+      "loss": 1.1696,
+      "step": 1225
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7749815648252937,
+      "learning_rate": 1.0114851399841885e-05,
+      "loss": 1.1269,
+      "step": 1226
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7973772024161017,
+      "learning_rate": 1.0101339963856112e-05,
+      "loss": 1.1685,
+      "step": 1227
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7267022058785041,
+      "learning_rate": 1.0087828342843589e-05,
+      "loss": 1.0965,
+      "step": 1228
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7809880084200272,
+      "learning_rate": 1.0074316561473873e-05,
+      "loss": 1.1244,
+      "step": 1229
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7902760251386206,
+      "learning_rate": 1.0060804644416804e-05,
+      "loss": 1.0957,
+      "step": 1230
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7826217550114022,
+      "learning_rate": 1.0047292616342468e-05,
+      "loss": 1.1699,
+      "step": 1231
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8581032051749498,
+      "learning_rate": 1.0033780501921164e-05,
+      "loss": 1.1493,
+      "step": 1232
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7307698738516711,
+      "learning_rate": 1.002026832582334e-05,
+      "loss": 1.171,
+      "step": 1233
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7901099567236501,
+      "learning_rate": 1.0006756112719558e-05,
+      "loss": 1.1273,
+      "step": 1234
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7617902541786274,
+      "learning_rate": 9.993243887280445e-06,
+      "loss": 1.2025,
+      "step": 1235
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7157006669111132,
+      "learning_rate": 9.979731674176667e-06,
+      "loss": 1.0525,
+      "step": 1236
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8790192533306618,
+      "learning_rate": 9.966219498078839e-06,
+      "loss": 1.1262,
+      "step": 1237
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7128148361435578,
+      "learning_rate": 9.952707383657537e-06,
+      "loss": 1.0185,
+      "step": 1238
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7721688614552051,
+      "learning_rate": 9.939195355583199e-06,
+      "loss": 1.2056,
+      "step": 1239
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7521915434162311,
+      "learning_rate": 9.925683438526132e-06,
+      "loss": 1.1205,
+      "step": 1240
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8019395220580379,
+      "learning_rate": 9.912171657156415e-06,
+      "loss": 1.1943,
+      "step": 1241
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7024215835387069,
+      "learning_rate": 9.898660036143893e-06,
+      "loss": 1.1973,
+      "step": 1242
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8255141734864481,
+      "learning_rate": 9.885148600158116e-06,
+      "loss": 1.0459,
+      "step": 1243
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7489325464112412,
+      "learning_rate": 9.8716373738683e-06,
+      "loss": 1.0723,
+      "step": 1244
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7193846988536986,
+      "learning_rate": 9.858126381943266e-06,
+      "loss": 1.1674,
+      "step": 1245
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.89394956459997,
+      "learning_rate": 9.844615649051418e-06,
+      "loss": 1.1095,
+      "step": 1246
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8585200621887755,
+      "learning_rate": 9.83110519986069e-06,
+      "loss": 1.1998,
+      "step": 1247
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7717118958418503,
+      "learning_rate": 9.817595059038483e-06,
+      "loss": 1.2069,
+      "step": 1248
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.723939176392828,
+      "learning_rate": 9.804085251251653e-06,
+      "loss": 1.0443,
+      "step": 1249
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.774701998712394,
+      "learning_rate": 9.790575801166432e-06,
+      "loss": 1.0475,
+      "step": 1250
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 2.3323707580566406,
+      "eval_runtime": 10.032,
+      "eval_samples_per_second": 89.713,
+      "eval_steps_per_second": 0.199,
+      "step": 1250
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8458619900749724,
+      "learning_rate": 9.777066733448408e-06,
+      "loss": 1.1871,
+      "step": 1251
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8449593168894898,
+      "learning_rate": 9.763558072762467e-06,
+      "loss": 1.0452,
+      "step": 1252
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.0137757147961453,
+      "learning_rate": 9.75004984377276e-06,
+      "loss": 1.0811,
+      "step": 1253
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7582730622147237,
+      "learning_rate": 9.736542071142635e-06,
+      "loss": 1.1446,
+      "step": 1254
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8304976125747143,
+      "learning_rate": 9.723034779534621e-06,
+      "loss": 1.152,
+      "step": 1255
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8979975308219432,
+      "learning_rate": 9.709527993610359e-06,
+      "loss": 1.1532,
+      "step": 1256
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7894101272933622,
+      "learning_rate": 9.696021738030575e-06,
+      "loss": 1.1618,
+      "step": 1257
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7077267482639953,
+      "learning_rate": 9.682516037455019e-06,
+      "loss": 1.0416,
+      "step": 1258
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8520000502182744,
+      "learning_rate": 9.669010916542433e-06,
+      "loss": 1.2004,
+      "step": 1259
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.0442454283203029,
+      "learning_rate": 9.655506399950496e-06,
+      "loss": 1.1843,
+      "step": 1260
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6935202440349418,
+      "learning_rate": 9.64200251233579e-06,
+      "loss": 1.0827,
+      "step": 1261
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6961809950085467,
+      "learning_rate": 9.62849927835374e-06,
+      "loss": 1.1733,
+      "step": 1262
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.7060912145963045,
+      "learning_rate": 9.61499672265859e-06,
+      "loss": 1.1335,
+      "step": 1263
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7671221085806431,
+      "learning_rate": 9.60149486990333e-06,
+      "loss": 1.1278,
+      "step": 1264
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8522205569134896,
+      "learning_rate": 9.58799374473968e-06,
+      "loss": 1.1127,
+      "step": 1265
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7859753356130073,
+      "learning_rate": 9.574493371818021e-06,
+      "loss": 1.1689,
+      "step": 1266
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8359052600160943,
+      "learning_rate": 9.560993775787373e-06,
+      "loss": 1.0103,
+      "step": 1267
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8002686686777198,
+      "learning_rate": 9.547494981295326e-06,
+      "loss": 1.1277,
+      "step": 1268
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8385473261108477,
+      "learning_rate": 9.53399701298801e-06,
+      "loss": 1.0138,
+      "step": 1269
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8484626780557085,
+      "learning_rate": 9.520499895510051e-06,
+      "loss": 1.1785,
+      "step": 1270
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8162305988703905,
+      "learning_rate": 9.507003653504515e-06,
+      "loss": 1.1371,
+      "step": 1271
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7482558848143593,
+      "learning_rate": 9.493508311612874e-06,
+      "loss": 1.223,
+      "step": 1272
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7589010005974214,
+      "learning_rate": 9.480013894474954e-06,
+      "loss": 1.0865,
+      "step": 1273
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7691285185823166,
+      "learning_rate": 9.466520426728895e-06,
+      "loss": 1.1718,
+      "step": 1274
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8350870617046458,
+      "learning_rate": 9.453027933011099e-06,
+      "loss": 1.1084,
+      "step": 1275
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.9171530615001061,
+      "learning_rate": 9.4395364379562e-06,
+      "loss": 1.0755,
+      "step": 1276
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7966050463115607,
+      "learning_rate": 9.426045966196992e-06,
+      "loss": 1.1722,
+      "step": 1277
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8490027702251157,
+      "learning_rate": 9.412556542364419e-06,
+      "loss": 1.2033,
+      "step": 1278
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8417702149212573,
+      "learning_rate": 9.399068191087498e-06,
+      "loss": 1.063,
+      "step": 1279
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.84173140114338,
+      "learning_rate": 9.385580936993295e-06,
+      "loss": 1.1133,
+      "step": 1280
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8493584862578284,
+      "learning_rate": 9.372094804706867e-06,
+      "loss": 1.1265,
+      "step": 1281
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.9713031859982587,
+      "learning_rate": 9.35860981885123e-06,
+      "loss": 1.1485,
+      "step": 1282
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8175419043914821,
+      "learning_rate": 9.3451260040473e-06,
+      "loss": 1.1337,
+      "step": 1283
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8655413674001964,
+      "learning_rate": 9.331643384913865e-06,
+      "loss": 1.0922,
+      "step": 1284
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.9344882492771377,
+      "learning_rate": 9.318161986067508e-06,
+      "loss": 1.1383,
+      "step": 1285
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7718627713394016,
+      "learning_rate": 9.304681832122617e-06,
+      "loss": 1.0948,
+      "step": 1286
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8847800639190583,
+      "learning_rate": 9.291202947691272e-06,
+      "loss": 1.0762,
+      "step": 1287
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.2317785221986361,
+      "learning_rate": 9.277725357383263e-06,
+      "loss": 1.0709,
+      "step": 1288
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.9543780076185517,
+      "learning_rate": 9.264249085806005e-06,
+      "loss": 1.1368,
+      "step": 1289
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7799642318495636,
+      "learning_rate": 9.250774157564503e-06,
+      "loss": 1.1402,
+      "step": 1290
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7836920780096875,
+      "learning_rate": 9.237300597261316e-06,
+      "loss": 1.1191,
+      "step": 1291
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.126562180708283,
+      "learning_rate": 9.2238284294965e-06,
+      "loss": 1.1446,
+      "step": 1292
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.9683325525986611,
+      "learning_rate": 9.210357678867576e-06,
+      "loss": 1.2044,
+      "step": 1293
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8928463054173592,
+      "learning_rate": 9.196888369969462e-06,
+      "loss": 1.0328,
+      "step": 1294
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.0712745350311939,
+      "learning_rate": 9.183420527394464e-06,
+      "loss": 1.1232,
+      "step": 1295
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.9766317884734149,
+      "learning_rate": 9.169954175732192e-06,
+      "loss": 1.189,
+      "step": 1296
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8607296506503527,
+      "learning_rate": 9.156489339569555e-06,
+      "loss": 1.1461,
+      "step": 1297
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8189770615652778,
+      "learning_rate": 9.143026043490668e-06,
+      "loss": 1.0632,
+      "step": 1298
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8083974587802227,
+      "learning_rate": 9.129564312076861e-06,
+      "loss": 1.2277,
+      "step": 1299
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8904071060021848,
+      "learning_rate": 9.116104169906584e-06,
+      "loss": 1.0033,
+      "step": 1300
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 2.243335247039795,
+      "eval_runtime": 9.3051,
+      "eval_samples_per_second": 96.721,
+      "eval_steps_per_second": 0.215,
+      "step": 1300
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.9430070125181499,
+      "learning_rate": 9.10264564155541e-06,
+      "loss": 1.1988,
+      "step": 1301
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.9955243737176671,
+      "learning_rate": 9.089188751595937e-06,
+      "loss": 1.1975,
+      "step": 1302
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.0215698795945274,
+      "learning_rate": 9.075733524597801e-06,
+      "loss": 1.1091,
+      "step": 1303
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.48736719126982,
+      "learning_rate": 9.062279985127575e-06,
+      "loss": 1.102,
+      "step": 1304
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.894795192838947,
+      "learning_rate": 9.04882815774877e-06,
+      "loss": 1.2117,
+      "step": 1305
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8732936541307875,
+      "learning_rate": 9.035378067021761e-06,
+      "loss": 1.0107,
+      "step": 1306
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.778984888600085,
+      "learning_rate": 9.021929737503757e-06,
+      "loss": 1.1923,
+      "step": 1307
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.0005768220710898,
+      "learning_rate": 9.00848319374875e-06,
+      "loss": 1.2524,
+      "step": 1308
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.82880329936799,
+      "learning_rate": 8.995038460307472e-06,
+      "loss": 1.0751,
+      "step": 1309
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8625515630337257,
+      "learning_rate": 8.981595561727345e-06,
+      "loss": 1.0975,
+      "step": 1310
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.767779170578959,
+      "learning_rate": 8.968154522552448e-06,
+      "loss": 1.1986,
+      "step": 1311
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.8975162521734532,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 1.0502,
+      "step": 1312
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.328479596841404,
+      "learning_rate": 8.941278120577635e-06,
+      "loss": 1.2139,
+      "step": 1313
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8152146777725857,
+      "learning_rate": 8.927842806848714e-06,
+      "loss": 1.1965,
+      "step": 1314
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8566123632624568,
+      "learning_rate": 8.914409450666929e-06,
+      "loss": 0.9287,
+      "step": 1315
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7746175647726465,
+      "learning_rate": 8.900978076558936e-06,
+      "loss": 1.0964,
+      "step": 1316
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.1906074550088805,
+      "learning_rate": 8.887548709047765e-06,
+      "loss": 1.1825,
+      "step": 1317
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8140526952564473,
+      "learning_rate": 8.874121372652793e-06,
+      "loss": 1.0434,
+      "step": 1318
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8959855059581603,
+      "learning_rate": 8.860696091889677e-06,
+      "loss": 1.1166,
+      "step": 1319
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8996331569626936,
+      "learning_rate": 8.84727289127033e-06,
+      "loss": 1.1495,
+      "step": 1320
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.4504829649915065,
+      "learning_rate": 8.83385179530286e-06,
+      "loss": 1.1168,
+      "step": 1321
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.0721093741223704,
+      "learning_rate": 8.820432828491542e-06,
+      "loss": 1.1227,
+      "step": 1322
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.9998747949801648,
+      "learning_rate": 8.80701601533675e-06,
+      "loss": 1.1133,
+      "step": 1323
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7750735468490497,
+      "learning_rate": 8.79360138033494e-06,
+      "loss": 1.1235,
+      "step": 1324
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.9221487909421118,
+      "learning_rate": 8.780188947978576e-06,
+      "loss": 1.2042,
+      "step": 1325
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7542931748195458,
+      "learning_rate": 8.766778742756117e-06,
+      "loss": 1.175,
+      "step": 1326
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.9428364119995898,
+      "learning_rate": 8.753370789151941e-06,
+      "loss": 1.1569,
+      "step": 1327
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8360791734903594,
+      "learning_rate": 8.739965111646327e-06,
+      "loss": 1.1256,
+      "step": 1328
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.9408355886073154,
+      "learning_rate": 8.726561734715388e-06,
+      "loss": 1.1177,
+      "step": 1329
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.0551170225241941,
+      "learning_rate": 8.713160682831044e-06,
+      "loss": 1.0419,
+      "step": 1330
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.868209642816728,
+      "learning_rate": 8.699761980460966e-06,
+      "loss": 1.1655,
+      "step": 1331
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.1876661824635557,
+      "learning_rate": 8.686365652068536e-06,
+      "loss": 1.1735,
+      "step": 1332
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.0529393368221536,
+      "learning_rate": 8.672971722112805e-06,
+      "loss": 1.139,
+      "step": 1333
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.992238694330592,
+      "learning_rate": 8.659580215048436e-06,
+      "loss": 1.1338,
+      "step": 1334
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.9900938656404871,
+      "learning_rate": 8.646191155325682e-06,
+      "loss": 0.9681,
+      "step": 1335
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.472803582442396,
+      "learning_rate": 8.632804567390313e-06,
+      "loss": 1.2274,
+      "step": 1336
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8163084851133082,
+      "learning_rate": 8.619420475683597e-06,
+      "loss": 1.1452,
+      "step": 1337
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7259940216743208,
+      "learning_rate": 8.606038904642239e-06,
+      "loss": 1.1022,
+      "step": 1338
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7806613238569517,
+      "learning_rate": 8.592659878698343e-06,
+      "loss": 1.0644,
+      "step": 1339
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7280615580213574,
+      "learning_rate": 8.579283422279366e-06,
+      "loss": 1.1097,
+      "step": 1340
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8272381033999228,
+      "learning_rate": 8.565909559808076e-06,
+      "loss": 0.9943,
+      "step": 1341
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8373529852230679,
+      "learning_rate": 8.552538315702497e-06,
+      "loss": 1.1816,
+      "step": 1342
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7489671924736435,
+      "learning_rate": 8.539169714375885e-06,
+      "loss": 1.1205,
+      "step": 1343
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.0903373118280466,
+      "learning_rate": 8.52580378023666e-06,
+      "loss": 1.1528,
+      "step": 1344
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.0392234615241904,
+      "learning_rate": 8.512440537688376e-06,
+      "loss": 1.0324,
+      "step": 1345
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.9178992497326748,
+      "learning_rate": 8.499080011129674e-06,
+      "loss": 1.2115,
+      "step": 1346
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8801613534206252,
+      "learning_rate": 8.485722224954237e-06,
+      "loss": 1.1103,
+      "step": 1347
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.82361663631151,
+      "learning_rate": 8.472367203550735e-06,
+      "loss": 1.0535,
+      "step": 1348
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.5299702806421165,
+      "learning_rate": 8.459014971302808e-06,
+      "loss": 1.1181,
+      "step": 1349
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8443547101692488,
+      "learning_rate": 8.445665552588983e-06,
+      "loss": 1.1399,
+      "step": 1350
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 2.3487563133239746,
+      "eval_runtime": 9.3028,
+      "eval_samples_per_second": 96.745,
+      "eval_steps_per_second": 0.215,
+      "step": 1350
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.671867759579526,
+      "learning_rate": 8.432318971782672e-06,
+      "loss": 1.0348,
+      "step": 1351
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.1871858074800103,
+      "learning_rate": 8.418975253252079e-06,
+      "loss": 1.1278,
+      "step": 1352
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.0402985442821084,
+      "learning_rate": 8.405634421360207e-06,
+      "loss": 1.1314,
+      "step": 1353
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8257521833062307,
+      "learning_rate": 8.39229650046478e-06,
+      "loss": 1.2048,
+      "step": 1354
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8399084208105025,
+      "learning_rate": 8.378961514918199e-06,
+      "loss": 1.0847,
+      "step": 1355
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7814471605625676,
+      "learning_rate": 8.365629489067516e-06,
+      "loss": 1.1759,
+      "step": 1356
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8243018717294182,
+      "learning_rate": 8.352300447254372e-06,
+      "loss": 1.1574,
+      "step": 1357
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7350688898884422,
+      "learning_rate": 8.338974413814971e-06,
+      "loss": 1.0707,
+      "step": 1358
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.4013587481037315,
+      "learning_rate": 8.325651413080003e-06,
+      "loss": 1.0732,
+      "step": 1359
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8924755833538307,
+      "learning_rate": 8.312331469374647e-06,
+      "loss": 1.1895,
+      "step": 1360
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.3412397391612398,
+      "learning_rate": 8.299014607018479e-06,
+      "loss": 1.1012,
+      "step": 1361
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8447363795544318,
+      "learning_rate": 8.285700850325467e-06,
+      "loss": 1.0886,
+      "step": 1362
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7894367163662303,
+      "learning_rate": 8.272390223603886e-06,
+      "loss": 1.1066,
+      "step": 1363
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8792091640511538,
+      "learning_rate": 8.259082751156325e-06,
+      "loss": 1.0483,
+      "step": 1364
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.9748528193860516,
+      "learning_rate": 8.245778457279583e-06,
+      "loss": 1.2137,
+      "step": 1365
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.7926774947878112,
+      "learning_rate": 8.232477366264687e-06,
+      "loss": 1.0598,
+      "step": 1366
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8624333632798591,
+      "learning_rate": 8.219179502396786e-06,
+      "loss": 1.1106,
+      "step": 1367
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8433732521588638,
+      "learning_rate": 8.205884889955163e-06,
+      "loss": 1.237,
+      "step": 1368
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.9137021274820041,
+      "learning_rate": 8.192593553213145e-06,
+      "loss": 1.1517,
+      "step": 1369
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8788542977943267,
+      "learning_rate": 8.17930551643809e-06,
+      "loss": 1.0221,
+      "step": 1370
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.9479304859021297,
+      "learning_rate": 8.166020803891324e-06,
+      "loss": 1.142,
+      "step": 1371
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.3234903283270545,
+      "learning_rate": 8.15273943982811e-06,
+      "loss": 1.1164,
+      "step": 1372
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.920559862951049,
+      "learning_rate": 8.13946144849759e-06,
+      "loss": 1.1442,
+      "step": 1373
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.6731558753638764,
+      "learning_rate": 8.126186854142752e-06,
+      "loss": 1.168,
+      "step": 1374
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8119343774448876,
+      "learning_rate": 8.11291568100039e-06,
+      "loss": 1.1351,
+      "step": 1375
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7279613430155185,
+      "learning_rate": 8.099647953301035e-06,
+      "loss": 1.081,
+      "step": 1376
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.15510715308174,
+      "learning_rate": 8.086383695268937e-06,
+      "loss": 1.2235,
+      "step": 1377
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7819914411202705,
+      "learning_rate": 8.07312293112201e-06,
+      "loss": 1.1485,
+      "step": 1378
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.9439763568064707,
+      "learning_rate": 8.05986568507179e-06,
+      "loss": 1.0896,
+      "step": 1379
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8563252899399946,
+      "learning_rate": 8.046611981323387e-06,
+      "loss": 1.1732,
+      "step": 1380
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7847693037662328,
+      "learning_rate": 8.033361844075445e-06,
+      "loss": 1.1107,
+      "step": 1381
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8614186799268596,
+      "learning_rate": 8.020115297520093e-06,
+      "loss": 1.0884,
+      "step": 1382
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.9184142910473174,
+      "learning_rate": 8.006872365842913e-06,
+      "loss": 1.1648,
+      "step": 1383
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 6.212177062656979,
+      "learning_rate": 7.993633073222876e-06,
+      "loss": 1.1168,
+      "step": 1384
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8797489791830236,
+      "learning_rate": 7.980397443832316e-06,
+      "loss": 1.1399,
+      "step": 1385
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.5701950982225872,
+      "learning_rate": 7.967165501836873e-06,
+      "loss": 1.1382,
+      "step": 1386
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8356398406527767,
+      "learning_rate": 7.953937271395465e-06,
+      "loss": 1.1113,
+      "step": 1387
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7709366616037341,
+      "learning_rate": 7.940712776660218e-06,
+      "loss": 1.1149,
+      "step": 1388
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8101600303741683,
+      "learning_rate": 7.927492041776452e-06,
+      "loss": 1.1334,
+      "step": 1389
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.9136756414883939,
+      "learning_rate": 7.914275090882613e-06,
+      "loss": 1.1342,
+      "step": 1390
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8680149464147092,
+      "learning_rate": 7.901061948110244e-06,
+      "loss": 1.1461,
+      "step": 1391
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8996964553417984,
+      "learning_rate": 7.887852637583927e-06,
+      "loss": 0.9813,
+      "step": 1392
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7524808455356068,
+      "learning_rate": 7.874647183421259e-06,
+      "loss": 1.1648,
+      "step": 1393
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8109187398203425,
+      "learning_rate": 7.861445609732783e-06,
+      "loss": 1.139,
+      "step": 1394
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7774540401594147,
+      "learning_rate": 7.848247940621967e-06,
+      "loss": 1.1155,
+      "step": 1395
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7803548560527424,
+      "learning_rate": 7.835054200185147e-06,
+      "loss": 1.0264,
+      "step": 1396
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8271555645630937,
+      "learning_rate": 7.821864412511485e-06,
+      "loss": 1.1265,
+      "step": 1397
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8891560438571152,
+      "learning_rate": 7.808678601682927e-06,
+      "loss": 1.1601,
+      "step": 1398
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.909411683913973,
+      "learning_rate": 7.795496791774153e-06,
+      "loss": 1.1024,
+      "step": 1399
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7304775958495282,
+      "learning_rate": 7.78231900685255e-06,
+      "loss": 1.165,
+      "step": 1400
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 2.350184917449951,
+      "eval_runtime": 11.8327,
+      "eval_samples_per_second": 76.06,
+      "eval_steps_per_second": 0.169,
+      "step": 1400
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8850058929370609,
+      "learning_rate": 7.769145270978142e-06,
+      "loss": 1.0642,
+      "step": 1401
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7714396319603787,
+      "learning_rate": 7.755975608203571e-06,
+      "loss": 1.1843,
+      "step": 1402
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7958657297121172,
+      "learning_rate": 7.742810042574038e-06,
+      "loss": 1.0713,
+      "step": 1403
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7560419948640559,
+      "learning_rate": 7.729648598127263e-06,
+      "loss": 1.1403,
+      "step": 1404
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8742718809067249,
+      "learning_rate": 7.716491298893443e-06,
+      "loss": 1.1002,
+      "step": 1405
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.1694224776382038,
+      "learning_rate": 7.703338168895207e-06,
+      "loss": 1.1375,
+      "step": 1406
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.226657699814387,
+      "learning_rate": 7.690189232147566e-06,
+      "loss": 1.0974,
+      "step": 1407
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7747337019145663,
+      "learning_rate": 7.677044512657889e-06,
+      "loss": 1.1925,
+      "step": 1408
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9190877008047365,
+      "learning_rate": 7.663904034425825e-06,
+      "loss": 1.0859,
+      "step": 1409
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7939551746615824,
+      "learning_rate": 7.650767821443302e-06,
+      "loss": 1.056,
+      "step": 1410
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7881236392381286,
+      "learning_rate": 7.637635897694434e-06,
+      "loss": 1.1401,
+      "step": 1411
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6906355498451188,
+      "learning_rate": 7.624508287155534e-06,
+      "loss": 1.0748,
+      "step": 1412
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7259219108228842,
+      "learning_rate": 7.611385013795011e-06,
+      "loss": 0.9555,
+      "step": 1413
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7497301933807179,
+      "learning_rate": 7.598266101573381e-06,
+      "loss": 1.1961,
+      "step": 1414
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.9115533966081147,
+      "learning_rate": 7.58515157444317e-06,
+      "loss": 1.066,
+      "step": 1415
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7370368110550405,
+      "learning_rate": 7.572041456348925e-06,
+      "loss": 1.1738,
+      "step": 1416
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6977055982286051,
+      "learning_rate": 7.558935771227129e-06,
+      "loss": 1.0689,
+      "step": 1417
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0136163131716276,
+      "learning_rate": 7.5458345430061675e-06,
+      "loss": 1.0689,
+      "step": 1418
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.833632509201279,
+      "learning_rate": 7.532737795606299e-06,
+      "loss": 1.1202,
+      "step": 1419
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6908248078907718,
+      "learning_rate": 7.519645552939594e-06,
+      "loss": 1.189,
+      "step": 1420
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6967105697579766,
+      "learning_rate": 7.506557838909902e-06,
+      "loss": 1.0548,
+      "step": 1421
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.814007332331037,
+      "learning_rate": 7.493474677412795e-06,
+      "loss": 1.2035,
+      "step": 1422
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7978220816426238,
+      "learning_rate": 7.48039609233555e-06,
+      "loss": 1.0209,
+      "step": 1423
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.725402068992431,
+      "learning_rate": 7.467322107557069e-06,
+      "loss": 1.1279,
+      "step": 1424
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8206203016960453,
+      "learning_rate": 7.454252746947874e-06,
+      "loss": 1.0866,
+      "step": 1425
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.9299836185852272,
+      "learning_rate": 7.441188034370025e-06,
+      "loss": 1.1733,
+      "step": 1426
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7280656115096015,
+      "learning_rate": 7.428127993677116e-06,
+      "loss": 1.0319,
+      "step": 1427
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8539041934920084,
+      "learning_rate": 7.415072648714186e-06,
+      "loss": 1.1819,
+      "step": 1428
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8180539664150399,
+      "learning_rate": 7.40202202331773e-06,
+      "loss": 1.1297,
+      "step": 1429
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7916863523671094,
+      "learning_rate": 7.3889761413156e-06,
+      "loss": 1.0647,
+      "step": 1430
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7119644493769051,
+      "learning_rate": 7.3759350265270025e-06,
+      "loss": 1.1553,
+      "step": 1431
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8112569891162086,
+      "learning_rate": 7.362898702762433e-06,
+      "loss": 1.0934,
+      "step": 1432
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6851196059116562,
+      "learning_rate": 7.349867193823645e-06,
+      "loss": 1.0656,
+      "step": 1433
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7846639577745065,
+      "learning_rate": 7.336840523503597e-06,
+      "loss": 1.1883,
+      "step": 1434
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8296489707256626,
+      "learning_rate": 7.323818715586415e-06,
+      "loss": 1.1391,
+      "step": 1435
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7816258771675311,
+      "learning_rate": 7.310801793847344e-06,
+      "loss": 1.036,
+      "step": 1436
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6991610300820477,
+      "learning_rate": 7.297789782052716e-06,
+      "loss": 1.1717,
+      "step": 1437
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7774876949877576,
+      "learning_rate": 7.284782703959889e-06,
+      "loss": 1.1254,
+      "step": 1438
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8315825172288835,
+      "learning_rate": 7.271780583317214e-06,
+      "loss": 1.0766,
+      "step": 1439
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6714771526567787,
+      "learning_rate": 7.2587834438640024e-06,
+      "loss": 1.0682,
+      "step": 1440
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7167530267318645,
+      "learning_rate": 7.2457913093304545e-06,
+      "loss": 1.1373,
+      "step": 1441
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.9074384107410725,
+      "learning_rate": 7.232804203437645e-06,
+      "loss": 1.0664,
+      "step": 1442
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7087967587439991,
+      "learning_rate": 7.2198221498974565e-06,
+      "loss": 1.105,
+      "step": 1443
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.0556082466302463,
+      "learning_rate": 7.20684517241256e-06,
+      "loss": 1.1068,
+      "step": 1444
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.0618213497871007,
+      "learning_rate": 7.193873294676345e-06,
+      "loss": 1.1652,
+      "step": 1445
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.9089146343791631,
+      "learning_rate": 7.180906540372904e-06,
+      "loss": 1.0965,
+      "step": 1446
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7180728388871599,
+      "learning_rate": 7.16794493317696e-06,
+      "loss": 1.0145,
+      "step": 1447
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.728915180102743,
+      "learning_rate": 7.154988496753853e-06,
+      "loss": 1.0354,
+      "step": 1448
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7603038539721744,
+      "learning_rate": 7.142037254759469e-06,
+      "loss": 1.1496,
+      "step": 1449
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7980418926008391,
+      "learning_rate": 7.129091230840222e-06,
+      "loss": 1.1718,
+      "step": 1450
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 2.378558397293091,
+      "eval_runtime": 9.3683,
+      "eval_samples_per_second": 96.069,
+      "eval_steps_per_second": 0.213,
+      "step": 1450
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6911481262853796,
+      "learning_rate": 7.116150448632986e-06,
+      "loss": 1.1426,
+      "step": 1451
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7737695581862396,
+      "learning_rate": 7.10321493176508e-06,
+      "loss": 1.0619,
+      "step": 1452
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7592332847912123,
+      "learning_rate": 7.090284703854194e-06,
+      "loss": 1.1582,
+      "step": 1453
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.76544154603158,
+      "learning_rate": 7.077359788508375e-06,
+      "loss": 1.1036,
+      "step": 1454
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8163653073915922,
+      "learning_rate": 7.064440209325958e-06,
+      "loss": 1.1853,
+      "step": 1455
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8229404360341368,
+      "learning_rate": 7.051525989895547e-06,
+      "loss": 1.1098,
+      "step": 1456
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7407365753954236,
+      "learning_rate": 7.038617153795948e-06,
+      "loss": 1.2021,
+      "step": 1457
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8962180318571694,
+      "learning_rate": 7.025713724596151e-06,
+      "loss": 1.0338,
+      "step": 1458
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7590224732214048,
+      "learning_rate": 7.012815725855262e-06,
+      "loss": 1.1829,
+      "step": 1459
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7535024799326073,
+      "learning_rate": 6.99992318112248e-06,
+      "loss": 1.1007,
+      "step": 1460
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7252608393035173,
+      "learning_rate": 6.987036113937045e-06,
+      "loss": 0.9902,
+      "step": 1461
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7470665591540979,
+      "learning_rate": 6.974154547828191e-06,
+      "loss": 1.1058,
+      "step": 1462
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.71200145923976,
+      "learning_rate": 6.961278506315119e-06,
+      "loss": 1.0926,
+      "step": 1463
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8873335170847498,
+      "learning_rate": 6.948408012906927e-06,
+      "loss": 1.1189,
+      "step": 1464
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7071698202021497,
+      "learning_rate": 6.935543091102599e-06,
+      "loss": 1.1469,
+      "step": 1465
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8039047156950411,
+      "learning_rate": 6.922683764390934e-06,
+      "loss": 1.1934,
+      "step": 1466
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8149907836581701,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 1.077,
+      "step": 1467
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8201458207861649,
+      "learning_rate": 6.896981990149704e-06,
+      "loss": 1.1079,
+      "step": 1468
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7877666872586699,
+      "learning_rate": 6.884139589546498e-06,
+      "loss": 1.2105,
+      "step": 1469
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.9364349226718577,
+      "learning_rate": 6.87130287788859e-06,
+      "loss": 1.0997,
+      "step": 1470
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7734267703055838,
+      "learning_rate": 6.858471878613285e-06,
+      "loss": 1.0796,
+      "step": 1471
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.80342340596289,
+      "learning_rate": 6.845646615147445e-06,
+      "loss": 1.1948,
+      "step": 1472
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8429144456843146,
+      "learning_rate": 6.832827110907472e-06,
+      "loss": 1.0736,
+      "step": 1473
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8075778640419664,
+      "learning_rate": 6.8200133892992424e-06,
+      "loss": 1.1325,
+      "step": 1474
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7115898003482064,
+      "learning_rate": 6.8072054737180845e-06,
+      "loss": 1.1049,
+      "step": 1475
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8141551491515476,
+      "learning_rate": 6.794403387548713e-06,
+      "loss": 1.1301,
+      "step": 1476
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7468321713712858,
+      "learning_rate": 6.781607154165217e-06,
+      "loss": 1.1471,
+      "step": 1477
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6706123844976377,
+      "learning_rate": 6.76881679693098e-06,
+      "loss": 1.1322,
+      "step": 1478
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8868841226252564,
+      "learning_rate": 6.756032339198675e-06,
+      "loss": 1.0764,
+      "step": 1479
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.024068403894618,
+      "learning_rate": 6.743253804310186e-06,
+      "loss": 1.2088,
+      "step": 1480
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7543505642506441,
+      "learning_rate": 6.7304812155965995e-06,
+      "loss": 1.0975,
+      "step": 1481
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7143702875251675,
+      "learning_rate": 6.717714596378138e-06,
+      "loss": 1.1135,
+      "step": 1482
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7865470426290567,
+      "learning_rate": 6.704953969964121e-06,
+      "loss": 1.1201,
+      "step": 1483
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8188951880102353,
+      "learning_rate": 6.692199359652935e-06,
+      "loss": 1.1678,
+      "step": 1484
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7432843456321705,
+      "learning_rate": 6.679450788731971e-06,
+      "loss": 1.0549,
+      "step": 1485
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7737798228588817,
+      "learning_rate": 6.666708280477609e-06,
+      "loss": 1.1526,
+      "step": 1486
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7877860267341509,
+      "learning_rate": 6.65397185815514e-06,
+      "loss": 1.0925,
+      "step": 1487
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8526986222790603,
+      "learning_rate": 6.641241545018766e-06,
+      "loss": 1.1687,
+      "step": 1488
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7846629625587627,
+      "learning_rate": 6.628517364311511e-06,
+      "loss": 1.0901,
+      "step": 1489
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7301594697997263,
+      "learning_rate": 6.615799339265224e-06,
+      "loss": 1.1427,
+      "step": 1490
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6959357736660484,
+      "learning_rate": 6.603087493100495e-06,
+      "loss": 0.9753,
+      "step": 1491
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.9481138336572342,
+      "learning_rate": 6.5903818490266554e-06,
+      "loss": 1.1325,
+      "step": 1492
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7653717261360355,
+      "learning_rate": 6.5776824302416896e-06,
+      "loss": 1.0953,
+      "step": 1493
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.704210526801783,
+      "learning_rate": 6.564989259932231e-06,
+      "loss": 1.0786,
+      "step": 1494
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.074677210277518,
+      "learning_rate": 6.552302361273498e-06,
+      "loss": 1.0994,
+      "step": 1495
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.834729659766496,
+      "learning_rate": 6.539621757429263e-06,
+      "loss": 1.1691,
+      "step": 1496
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7472306846917951,
+      "learning_rate": 6.526947471551799e-06,
+      "loss": 1.137,
+      "step": 1497
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7979273835810529,
+      "learning_rate": 6.5142795267818505e-06,
+      "loss": 1.1037,
+      "step": 1498
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8727190849263229,
+      "learning_rate": 6.501617946248577e-06,
+      "loss": 1.0938,
+      "step": 1499
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8063480819822807,
+      "learning_rate": 6.488962753069528e-06,
+      "loss": 1.1013,
+      "step": 1500
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 2.449204683303833,
+      "eval_runtime": 9.8745,
+      "eval_samples_per_second": 91.144,
+      "eval_steps_per_second": 0.203,
+      "step": 1500
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6965265075090237,
+      "learning_rate": 6.476313970350578e-06,
+      "loss": 1.161,
+      "step": 1501
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8012297526167187,
+      "learning_rate": 6.4636716211859076e-06,
+      "loss": 1.1118,
+      "step": 1502
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7184355115590503,
+      "learning_rate": 6.451035728657953e-06,
+      "loss": 1.0769,
+      "step": 1503
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.9207487942633406,
+      "learning_rate": 6.438406315837348e-06,
+      "loss": 1.1872,
+      "step": 1504
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7365043694495089,
+      "learning_rate": 6.42578340578291e-06,
+      "loss": 1.0886,
+      "step": 1505
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8435071354705599,
+      "learning_rate": 6.4131670215415755e-06,
+      "loss": 1.1138,
+      "step": 1506
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7196440868153202,
+      "learning_rate": 6.400557186148371e-06,
+      "loss": 1.1929,
+      "step": 1507
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7769640223618369,
+      "learning_rate": 6.387953922626358e-06,
+      "loss": 1.0296,
+      "step": 1508
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7869200588306157,
+      "learning_rate": 6.375357253986614e-06,
+      "loss": 1.1531,
+      "step": 1509
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.1075947981481342,
+      "learning_rate": 6.362767203228158e-06,
+      "loss": 1.1428,
+      "step": 1510
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7254528970238628,
+      "learning_rate": 6.350183793337939e-06,
+      "loss": 1.1844,
+      "step": 1511
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8080991616520767,
+      "learning_rate": 6.337607047290774e-06,
+      "loss": 1.0195,
+      "step": 1512
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8117774448330095,
+      "learning_rate": 6.325036988049318e-06,
+      "loss": 1.1496,
+      "step": 1513
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8054420339757448,
+      "learning_rate": 6.3124736385640126e-06,
+      "loss": 1.1982,
+      "step": 1514
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6872160199817345,
+      "learning_rate": 6.299917021773054e-06,
+      "loss": 1.1146,
+      "step": 1515
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8191079029428923,
+      "learning_rate": 6.28736716060234e-06,
+      "loss": 1.1068,
+      "step": 1516
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7763278854470331,
+      "learning_rate": 6.274824077965438e-06,
+      "loss": 1.0916,
+      "step": 1517
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7339333966348177,
+      "learning_rate": 6.262287796763538e-06,
+      "loss": 1.0494,
+      "step": 1518
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7597029807057222,
+      "learning_rate": 6.249758339885411e-06,
+      "loss": 1.1379,
+      "step": 1519
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.9144159712922175,
+      "learning_rate": 6.23723573020737e-06,
+      "loss": 1.1622,
+      "step": 1520
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.4074750536811813,
+      "learning_rate": 6.224719990593224e-06,
+      "loss": 1.1638,
+      "step": 1521
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7437052464066358,
+      "learning_rate": 6.21221114389424e-06,
+      "loss": 0.9996,
+      "step": 1522
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7294018970709734,
+      "learning_rate": 6.199709212949101e-06,
+      "loss": 1.1848,
+      "step": 1523
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7717253289647504,
+      "learning_rate": 6.187214220583865e-06,
+      "loss": 1.0204,
+      "step": 1524
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8335216753205466,
+      "learning_rate": 6.174726189611911e-06,
+      "loss": 1.1198,
+      "step": 1525
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7714768396296743,
+      "learning_rate": 6.162245142833924e-06,
+      "loss": 1.1478,
+      "step": 1526
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6795884518373538,
+      "learning_rate": 6.149771103037821e-06,
+      "loss": 1.1036,
+      "step": 1527
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.798468605482379,
+      "learning_rate": 6.13730409299874e-06,
+      "loss": 1.1665,
+      "step": 1528
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.9014269166500948,
+      "learning_rate": 6.124844135478971e-06,
+      "loss": 1.1147,
+      "step": 1529
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8985653448579057,
+      "learning_rate": 6.112391253227939e-06,
+      "loss": 1.0616,
+      "step": 1530
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.723844022879328,
+      "learning_rate": 6.099945468982142e-06,
+      "loss": 1.1406,
+      "step": 1531
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7622061131130335,
+      "learning_rate": 6.087506805465127e-06,
+      "loss": 1.1412,
+      "step": 1532
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.0210199670890214,
+      "learning_rate": 6.075075285387431e-06,
+      "loss": 1.1165,
+      "step": 1533
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8954445040561444,
+      "learning_rate": 6.0626509314465545e-06,
+      "loss": 1.115,
+      "step": 1534
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7258131700400562,
+      "learning_rate": 6.0502337663269115e-06,
+      "loss": 1.1647,
+      "step": 1535
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7427806163604749,
+      "learning_rate": 6.037823812699792e-06,
+      "loss": 1.1007,
+      "step": 1536
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7464391031188592,
+      "learning_rate": 6.025421093223318e-06,
+      "loss": 0.9761,
+      "step": 1537
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8612231247224719,
+      "learning_rate": 6.013025630542409e-06,
+      "loss": 1.0556,
+      "step": 1538
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.696973855096318,
+      "learning_rate": 6.000637447288719e-06,
+      "loss": 1.0745,
+      "step": 1539
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.983548843106166,
+      "learning_rate": 5.988256566080636e-06,
+      "loss": 1.1585,
+      "step": 1540
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8446139930254501,
+      "learning_rate": 5.975883009523188e-06,
+      "loss": 1.1295,
+      "step": 1541
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7513713943183736,
+      "learning_rate": 5.963516800208056e-06,
+      "loss": 1.1428,
+      "step": 1542
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7685858328268531,
+      "learning_rate": 5.951157960713485e-06,
+      "loss": 0.982,
+      "step": 1543
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.9028306276366239,
+      "learning_rate": 5.938806513604278e-06,
+      "loss": 1.1958,
+      "step": 1544
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7961351089706975,
+      "learning_rate": 5.926462481431738e-06,
+      "loss": 1.077,
+      "step": 1545
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.729045273108765,
+      "learning_rate": 5.914125886733626e-06,
+      "loss": 1.1756,
+      "step": 1546
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8127817042938431,
+      "learning_rate": 5.901796752034128e-06,
+      "loss": 1.1114,
+      "step": 1547
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7363421748833503,
+      "learning_rate": 5.889475099843798e-06,
+      "loss": 1.207,
+      "step": 1548
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8584478019298438,
+      "learning_rate": 5.877160952659551e-06,
+      "loss": 1.024,
+      "step": 1549
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7166681546536106,
+      "learning_rate": 5.864854332964574e-06,
+      "loss": 1.041,
+      "step": 1550
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 2.480712413787842,
+      "eval_runtime": 10.364,
+      "eval_samples_per_second": 86.839,
+      "eval_steps_per_second": 0.193,
+      "step": 1550
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7759189052422971,
+      "learning_rate": 5.852555263228334e-06,
+      "loss": 1.0748,
+      "step": 1551
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8985309897086209,
+      "learning_rate": 5.84026376590649e-06,
+      "loss": 1.067,
+      "step": 1552
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7777733548002462,
+      "learning_rate": 5.827979863440895e-06,
+      "loss": 1.1262,
+      "step": 1553
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.729018014417819,
+      "learning_rate": 5.815703578259526e-06,
+      "loss": 1.1125,
+      "step": 1554
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7390914509595818,
+      "learning_rate": 5.803434932776452e-06,
+      "loss": 1.1406,
+      "step": 1555
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8039927272277627,
+      "learning_rate": 5.79117394939179e-06,
+      "loss": 1.0987,
+      "step": 1556
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.9686930024083484,
+      "learning_rate": 5.7789206504916815e-06,
+      "loss": 1.0963,
+      "step": 1557
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8229742025799942,
+      "learning_rate": 5.7666750584482246e-06,
+      "loss": 1.1338,
+      "step": 1558
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7818514957785645,
+      "learning_rate": 5.75443719561945e-06,
+      "loss": 1.1219,
+      "step": 1559
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8809268329343328,
+      "learning_rate": 5.742207084349274e-06,
+      "loss": 1.2068,
+      "step": 1560
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7365607346525931,
+      "learning_rate": 5.729984746967467e-06,
+      "loss": 1.0322,
+      "step": 1561
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7547236008146416,
+      "learning_rate": 5.7177702057896015e-06,
+      "loss": 1.1989,
+      "step": 1562
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7655214602019873,
+      "learning_rate": 5.705563483117014e-06,
+      "loss": 1.091,
+      "step": 1563
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7710525809942849,
+      "learning_rate": 5.693364601236762e-06,
+      "loss": 1.1718,
+      "step": 1564
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8258516403745388,
+      "learning_rate": 5.6811735824216064e-06,
+      "loss": 1.0675,
+      "step": 1565
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7729093616064575,
+      "learning_rate": 5.668990448929923e-06,
+      "loss": 1.148,
+      "step": 1566
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.768817329657577,
+      "learning_rate": 5.656815223005714e-06,
+      "loss": 1.1577,
+      "step": 1567
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6541721372619311,
+      "learning_rate": 5.644647926878537e-06,
+      "loss": 0.9552,
+      "step": 1568
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6738611667198448,
+      "learning_rate": 5.6324885827634665e-06,
+      "loss": 1.0778,
+      "step": 1569
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.860410088167586,
+      "learning_rate": 5.620337212861064e-06,
+      "loss": 1.0861,
+      "step": 1570
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8423779530932456,
+      "learning_rate": 5.608193839357325e-06,
+      "loss": 1.184,
+      "step": 1571
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7130494666596786,
+      "learning_rate": 5.5960584844236565e-06,
+      "loss": 1.0838,
+      "step": 1572
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7467419874974908,
+      "learning_rate": 5.583931170216814e-06,
+      "loss": 1.1381,
+      "step": 1573
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8951934163252196,
+      "learning_rate": 5.571811918878878e-06,
+      "loss": 1.0684,
+      "step": 1574
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.9858573152782086,
+      "learning_rate": 5.559700752537198e-06,
+      "loss": 1.1308,
+      "step": 1575
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.025962203354073,
+      "learning_rate": 5.547597693304387e-06,
+      "loss": 1.0439,
+      "step": 1576
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7612028485846755,
+      "learning_rate": 5.535502763278222e-06,
+      "loss": 1.2284,
+      "step": 1577
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7938504339055938,
+      "learning_rate": 5.5234159845416664e-06,
+      "loss": 1.0641,
+      "step": 1578
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6798520248787341,
+      "learning_rate": 5.51133737916278e-06,
+      "loss": 1.0506,
+      "step": 1579
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6967000793192533,
+      "learning_rate": 5.4992669691947235e-06,
+      "loss": 1.1271,
+      "step": 1580
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7595705878976658,
+      "learning_rate": 5.4872047766756605e-06,
+      "loss": 1.1519,
+      "step": 1581
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.285093869651487,
+      "learning_rate": 5.475150823628786e-06,
+      "loss": 1.0297,
+      "step": 1582
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.9390750879383458,
+      "learning_rate": 5.463105132062227e-06,
+      "loss": 1.1689,
+      "step": 1583
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.922255362240802,
+      "learning_rate": 5.451067723969048e-06,
+      "loss": 1.1503,
+      "step": 1584
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7783850126154288,
+      "learning_rate": 5.43903862132716e-06,
+      "loss": 1.1341,
+      "step": 1585
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8304161961504076,
+      "learning_rate": 5.427017846099338e-06,
+      "loss": 1.0619,
+      "step": 1586
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7555887081121229,
+      "learning_rate": 5.415005420233141e-06,
+      "loss": 1.1037,
+      "step": 1587
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6615398692149838,
+      "learning_rate": 5.403001365660878e-06,
+      "loss": 0.9303,
+      "step": 1588
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7019873517920403,
+      "learning_rate": 5.39100570429959e-06,
+      "loss": 1.1972,
+      "step": 1589
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7480915033159414,
+      "learning_rate": 5.379018458050974e-06,
+      "loss": 0.999,
+      "step": 1590
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7368865912799089,
+      "learning_rate": 5.367039648801386e-06,
+      "loss": 1.1416,
+      "step": 1591
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.13357242100831,
+      "learning_rate": 5.355069298421747e-06,
+      "loss": 1.1284,
+      "step": 1592
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7189081659112445,
+      "learning_rate": 5.343107428767564e-06,
+      "loss": 1.1645,
+      "step": 1593
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7737924035825609,
+      "learning_rate": 5.331154061678843e-06,
+      "loss": 1.0378,
+      "step": 1594
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8554817357677039,
+      "learning_rate": 5.3192092189800705e-06,
+      "loss": 1.2248,
+      "step": 1595
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8620879794176642,
+      "learning_rate": 5.3072729224801625e-06,
+      "loss": 1.0212,
+      "step": 1596
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.985453416166869,
+      "learning_rate": 5.295345193972445e-06,
+      "loss": 1.117,
+      "step": 1597
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.9472786963928582,
+      "learning_rate": 5.283426055234592e-06,
+      "loss": 1.1313,
+      "step": 1598
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7746405207599923,
+      "learning_rate": 5.271515528028592e-06,
+      "loss": 1.1588,
+      "step": 1599
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7520092163883141,
+      "learning_rate": 5.259613634100713e-06,
+      "loss": 1.0498,
+      "step": 1600
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 2.5218565464019775,
+      "eval_runtime": 9.2955,
+      "eval_samples_per_second": 96.821,
+      "eval_steps_per_second": 0.215,
+      "step": 1600
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.9465258584431174,
+      "learning_rate": 5.247720395181466e-06,
+      "loss": 1.1395,
+      "step": 1601
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.797725942660952,
+      "learning_rate": 5.235835832985552e-06,
+      "loss": 1.1868,
+      "step": 1602
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8139589682045474,
+      "learning_rate": 5.223959969211834e-06,
+      "loss": 1.0813,
+      "step": 1603
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.752654137152998,
+      "learning_rate": 5.2120928255432846e-06,
+      "loss": 1.1639,
+      "step": 1604
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.9232754768595844,
+      "learning_rate": 5.200234423646975e-06,
+      "loss": 1.1125,
+      "step": 1605
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7748792672988845,
+      "learning_rate": 5.188384785173995e-06,
+      "loss": 1.0161,
+      "step": 1606
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.9708336608580703,
+      "learning_rate": 5.176543931759447e-06,
+      "loss": 1.1454,
+      "step": 1607
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8871630691846137,
+      "learning_rate": 5.164711885022382e-06,
+      "loss": 1.2076,
+      "step": 1608
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7758732203860267,
+      "learning_rate": 5.152888666565789e-06,
+      "loss": 1.0429,
+      "step": 1609
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8506792292608445,
+      "learning_rate": 5.141074297976522e-06,
+      "loss": 1.1204,
+      "step": 1610
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8168938687560183,
+      "learning_rate": 5.129268800825281e-06,
+      "loss": 1.1868,
+      "step": 1611
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7228332246626927,
+      "learning_rate": 5.117472196666578e-06,
+      "loss": 1.073,
+      "step": 1612
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8710409556341027,
+      "learning_rate": 5.1056845070386754e-06,
+      "loss": 1.1701,
+      "step": 1613
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8562165157884877,
+      "learning_rate": 5.093905753463567e-06,
+      "loss": 1.2206,
+      "step": 1614
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.958688870103317,
+      "learning_rate": 5.082135957446923e-06,
+      "loss": 0.907,
+      "step": 1615
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7801942501550636,
+      "learning_rate": 5.0703751404780745e-06,
+      "loss": 1.1011,
+      "step": 1616
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7476034854037982,
+      "learning_rate": 5.058623324029944e-06,
+      "loss": 1.2111,
+      "step": 1617
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.467610597507765,
+      "learning_rate": 5.046880529559026e-06,
+      "loss": 1.0693,
+      "step": 1618
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8129997265747909,
+      "learning_rate": 5.035146778505338e-06,
+      "loss": 0.9706,
+      "step": 1619
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8428162174443055,
+      "learning_rate": 5.0234220922923985e-06,
+      "loss": 1.1671,
+      "step": 1620
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6935540228794193,
+      "learning_rate": 5.011706492327163e-06,
+      "loss": 1.0641,
+      "step": 1621
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7664487835392764,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 1.1453,
+      "step": 1622
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.1017502591150765,
+      "learning_rate": 4.988302636684654e-06,
+      "loss": 1.1307,
+      "step": 1623
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.754799039438748,
+      "learning_rate": 4.976614423738199e-06,
+      "loss": 1.1692,
+      "step": 1624
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8695410884774532,
+      "learning_rate": 4.964935382500999e-06,
+      "loss": 1.1324,
+      "step": 1625
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7557770592997468,
+      "learning_rate": 4.953265534296678e-06,
+      "loss": 1.1762,
+      "step": 1626
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8160081860859304,
+      "learning_rate": 4.941604900432065e-06,
+      "loss": 0.9935,
+      "step": 1627
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8128618301757234,
+      "learning_rate": 4.929953502197189e-06,
+      "loss": 1.1777,
+      "step": 1628
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.9591428806927773,
+      "learning_rate": 4.91831136086518e-06,
+      "loss": 1.149,
+      "step": 1629
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8234344268273015,
+      "learning_rate": 4.906678497692299e-06,
+      "loss": 1.104,
+      "step": 1630
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.0756747826502278,
+      "learning_rate": 4.895054933917856e-06,
+      "loss": 1.1195,
+      "step": 1631
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.9092023574557874,
+      "learning_rate": 4.8834406907641784e-06,
+      "loss": 1.2125,
+      "step": 1632
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8650294477299757,
+      "learning_rate": 4.87183578943658e-06,
+      "loss": 1.0325,
+      "step": 1633
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7892567509892997,
+      "learning_rate": 4.860240251123311e-06,
+      "loss": 1.0752,
+      "step": 1634
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.9787631303296512,
+      "learning_rate": 4.848654096995542e-06,
+      "loss": 1.1058,
+      "step": 1635
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.796506559976529,
+      "learning_rate": 4.837077348207296e-06,
+      "loss": 1.0583,
+      "step": 1636
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7760237757054798,
+      "learning_rate": 4.825510025895429e-06,
+      "loss": 1.1821,
+      "step": 1637
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.9087284435225723,
+      "learning_rate": 4.813952151179582e-06,
+      "loss": 1.1074,
+      "step": 1638
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7265317084017601,
+      "learning_rate": 4.802403745162159e-06,
+      "loss": 1.0351,
+      "step": 1639
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.9439452016142764,
+      "learning_rate": 4.790864828928253e-06,
+      "loss": 1.1367,
+      "step": 1640
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.9104479361422616,
+      "learning_rate": 4.779335423545658e-06,
+      "loss": 1.0046,
+      "step": 1641
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7253602115345795,
+      "learning_rate": 4.767815550064778e-06,
+      "loss": 1.0885,
+      "step": 1642
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.0882545825519045,
+      "learning_rate": 4.756305229518641e-06,
+      "loss": 1.1953,
+      "step": 1643
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8731644919723284,
+      "learning_rate": 4.744804482922799e-06,
+      "loss": 1.1256,
+      "step": 1644
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.9257964138243271,
+      "learning_rate": 4.7333133312753546e-06,
+      "loss": 1.0306,
+      "step": 1645
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8795544800640397,
+      "learning_rate": 4.721831795556873e-06,
+      "loss": 1.1408,
+      "step": 1646
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.257367025396079,
+      "learning_rate": 4.710359896730379e-06,
+      "loss": 1.0204,
+      "step": 1647
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7394052383794681,
+      "learning_rate": 4.698897655741279e-06,
+      "loss": 1.0687,
+      "step": 1648
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.0367088572576475,
+      "learning_rate": 4.68744509351737e-06,
+      "loss": 1.1935,
+      "step": 1649
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7350873299567016,
+      "learning_rate": 4.676002230968764e-06,
+      "loss": 1.1062,
+      "step": 1650
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 2.459991693496704,
+      "eval_runtime": 9.2747,
+      "eval_samples_per_second": 97.038,
+      "eval_steps_per_second": 0.216,
+      "step": 1650
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8039386188223937,
+      "learning_rate": 4.664569088987862e-06,
+      "loss": 1.1374,
+      "step": 1651
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8837082426680896,
+      "learning_rate": 4.65314568844933e-06,
+      "loss": 1.0623,
+      "step": 1652
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7675926179691429,
+      "learning_rate": 4.641732050210032e-06,
+      "loss": 1.2076,
+      "step": 1653
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7984900455562067,
+      "learning_rate": 4.630328195109026e-06,
+      "loss": 1.0972,
+      "step": 1654
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8438650096548526,
+      "learning_rate": 4.618934143967484e-06,
+      "loss": 1.1418,
+      "step": 1655
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8353476688319701,
+      "learning_rate": 4.6075499175887015e-06,
+      "loss": 1.1834,
+      "step": 1656
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.826130930118567,
+      "learning_rate": 4.596175536758024e-06,
+      "loss": 1.1774,
+      "step": 1657
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7114065821087058,
+      "learning_rate": 4.584811022242821e-06,
+      "loss": 1.0704,
+      "step": 1658
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8539947310126877,
+      "learning_rate": 4.573456394792446e-06,
+      "loss": 1.1097,
+      "step": 1659
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7692548880776193,
+      "learning_rate": 4.562111675138215e-06,
+      "loss": 1.0612,
+      "step": 1660
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7707862615734463,
+      "learning_rate": 4.5507768839933365e-06,
+      "loss": 1.1671,
+      "step": 1661
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.0641238075483097,
+      "learning_rate": 4.539452042052901e-06,
+      "loss": 1.0182,
+      "step": 1662
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7039297206922988,
+      "learning_rate": 4.528137169993827e-06,
+      "loss": 1.1321,
+      "step": 1663
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8760694517378612,
+      "learning_rate": 4.516832288474844e-06,
+      "loss": 1.1147,
+      "step": 1664
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7526973457541478,
+      "learning_rate": 4.505537418136425e-06,
+      "loss": 1.0787,
+      "step": 1665
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8272228782103522,
+      "learning_rate": 4.494252579600774e-06,
+      "loss": 1.0881,
+      "step": 1666
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8195647686041202,
+      "learning_rate": 4.482977793471769e-06,
+      "loss": 1.1847,
+      "step": 1667
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.9273937992223743,
+      "learning_rate": 4.471713080334951e-06,
+      "loss": 1.0947,
+      "step": 1668
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8206720492467162,
+      "learning_rate": 4.460458460757455e-06,
+      "loss": 1.0805,
+      "step": 1669
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8537173550444046,
+      "learning_rate": 4.449213955287992e-06,
+      "loss": 1.1523,
+      "step": 1670
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.1973697534130479,
+      "learning_rate": 4.437979584456803e-06,
+      "loss": 1.0935,
+      "step": 1671
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.0511740765917068,
+      "learning_rate": 4.426755368775637e-06,
+      "loss": 1.0548,
+      "step": 1672
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.1787786704325491,
+      "learning_rate": 4.415541328737689e-06,
+      "loss": 1.1864,
+      "step": 1673
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7962685941250545,
+      "learning_rate": 4.404337484817577e-06,
+      "loss": 1.1092,
+      "step": 1674
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.906307862460545,
+      "learning_rate": 4.393143857471312e-06,
+      "loss": 1.2065,
+      "step": 1675
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7967679377975939,
+      "learning_rate": 4.381960467136241e-06,
+      "loss": 1.0032,
+      "step": 1676
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8945768950127775,
+      "learning_rate": 4.370787334231026e-06,
+      "loss": 1.202,
+      "step": 1677
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.824717234110359,
+      "learning_rate": 4.359624479155595e-06,
+      "loss": 1.1526,
+      "step": 1678
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7349510545072296,
+      "learning_rate": 4.348471922291124e-06,
+      "loss": 1.0674,
+      "step": 1679
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.0839756492478692,
+      "learning_rate": 4.3373296839999725e-06,
+      "loss": 1.1849,
+      "step": 1680
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8347730369401857,
+      "learning_rate": 4.326197784625665e-06,
+      "loss": 1.0435,
+      "step": 1681
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8278076604879717,
+      "learning_rate": 4.315076244492847e-06,
+      "loss": 1.0258,
+      "step": 1682
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7461644397794381,
+      "learning_rate": 4.303965083907261e-06,
+      "loss": 1.1616,
+      "step": 1683
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7569852292867766,
+      "learning_rate": 4.292864323155684e-06,
+      "loss": 1.109,
+      "step": 1684
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8580496391189976,
+      "learning_rate": 4.281773982505913e-06,
+      "loss": 1.0911,
+      "step": 1685
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8845734998242154,
+      "learning_rate": 4.270694082206716e-06,
+      "loss": 1.1812,
+      "step": 1686
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8013738201680383,
+      "learning_rate": 4.2596246424878055e-06,
+      "loss": 1.0848,
+      "step": 1687
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7075617712036987,
+      "learning_rate": 4.2485656835597876e-06,
+      "loss": 1.0766,
+      "step": 1688
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8109793493428286,
+      "learning_rate": 4.237517225614138e-06,
+      "loss": 1.0932,
+      "step": 1689
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7665004066897888,
+      "learning_rate": 4.2264792888231495e-06,
+      "loss": 1.1011,
+      "step": 1690
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7414896954112714,
+      "learning_rate": 4.215451893339928e-06,
+      "loss": 1.0997,
+      "step": 1691
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6926296332933846,
+      "learning_rate": 4.204435059298303e-06,
+      "loss": 1.0384,
+      "step": 1692
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8797174778581421,
+      "learning_rate": 4.193428806812845e-06,
+      "loss": 1.0425,
+      "step": 1693
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.9056168777363915,
+      "learning_rate": 4.18243315597879e-06,
+      "loss": 1.1934,
+      "step": 1694
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7765150037541755,
+      "learning_rate": 4.1714481268720305e-06,
+      "loss": 1.0271,
+      "step": 1695
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7503866815547575,
+      "learning_rate": 4.160473739549056e-06,
+      "loss": 1.1484,
+      "step": 1696
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8179784158994647,
+      "learning_rate": 4.149510014046922e-06,
+      "loss": 1.0172,
+      "step": 1697
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8108975769153073,
+      "learning_rate": 4.138556970383235e-06,
+      "loss": 1.2159,
+      "step": 1698
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7505734075472335,
+      "learning_rate": 4.1276146285560824e-06,
+      "loss": 1.0053,
+      "step": 1699
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.9034202474925211,
+      "learning_rate": 4.116683008544019e-06,
+      "loss": 1.1999,
+      "step": 1700
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 2.4448859691619873,
+      "eval_runtime": 9.4967,
+      "eval_samples_per_second": 94.77,
+      "eval_steps_per_second": 0.211,
+      "step": 1700
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8213401969878088,
+      "learning_rate": 4.105762130306018e-06,
+      "loss": 1.0634,
+      "step": 1701
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7596811780977619,
+      "learning_rate": 4.094852013781456e-06,
+      "loss": 1.133,
+      "step": 1702
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6852754949140608,
+      "learning_rate": 4.083952678890035e-06,
+      "loss": 1.1108,
+      "step": 1703
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8341704779212652,
+      "learning_rate": 4.073064145531798e-06,
+      "loss": 1.097,
+      "step": 1704
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8723639995566047,
+      "learning_rate": 4.0621864335870456e-06,
+      "loss": 1.0955,
+      "step": 1705
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.714966801266124,
+      "learning_rate": 4.0513195629163415e-06,
+      "loss": 1.1333,
+      "step": 1706
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7734518563186547,
+      "learning_rate": 4.040463553360431e-06,
+      "loss": 1.0541,
+      "step": 1707
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.9116749703854775,
+      "learning_rate": 4.029618424740251e-06,
+      "loss": 1.1494,
+      "step": 1708
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8107506708242812,
+      "learning_rate": 4.018784196856858e-06,
+      "loss": 1.119,
+      "step": 1709
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7662299493539296,
+      "learning_rate": 4.007960889491419e-06,
+      "loss": 1.1124,
+      "step": 1710
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7240121112727814,
+      "learning_rate": 3.997148522405141e-06,
+      "loss": 1.1168,
+      "step": 1711
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7330701532571976,
+      "learning_rate": 3.986347115339281e-06,
+      "loss": 1.1261,
+      "step": 1712
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.756548767532252,
+      "learning_rate": 3.975556688015069e-06,
+      "loss": 0.8905,
+      "step": 1713
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.751951293791386,
+      "learning_rate": 3.964777260133692e-06,
+      "loss": 1.125,
+      "step": 1714
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7099513711649637,
+      "learning_rate": 3.954008851376252e-06,
+      "loss": 1.1459,
+      "step": 1715
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7611981160027421,
+      "learning_rate": 3.94325148140374e-06,
+      "loss": 1.111,
+      "step": 1716
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7719910201090303,
+      "learning_rate": 3.932505169856993e-06,
+      "loss": 1.101,
+      "step": 1717
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6947701612618595,
+      "learning_rate": 3.921769936356638e-06,
+      "loss": 1.0393,
+      "step": 1718
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7739706960527303,
+      "learning_rate": 3.911045800503101e-06,
+      "loss": 1.0986,
+      "step": 1719
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7897390779206512,
+      "learning_rate": 3.90033278187653e-06,
+      "loss": 1.1759,
+      "step": 1720
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7842884781187462,
+      "learning_rate": 3.889630900036781e-06,
+      "loss": 1.1119,
+      "step": 1721
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7683629558013947,
+      "learning_rate": 3.878940174523371e-06,
+      "loss": 1.0648,
+      "step": 1722
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7308350554116381,
+      "learning_rate": 3.86826062485546e-06,
+      "loss": 1.1575,
+      "step": 1723
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7825905771353417,
+      "learning_rate": 3.8575922705317896e-06,
+      "loss": 1.076,
+      "step": 1724
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8670462938439253,
+      "learning_rate": 3.846935131030667e-06,
+      "loss": 1.0516,
+      "step": 1725
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7901467101338434,
+      "learning_rate": 3.8362892258099196e-06,
+      "loss": 1.1417,
+      "step": 1726
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7603885232974734,
+      "learning_rate": 3.825654574306873e-06,
+      "loss": 1.0608,
+      "step": 1727
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7411302474563783,
+      "learning_rate": 3.815031195938295e-06,
+      "loss": 1.1781,
+      "step": 1728
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8849538469228431,
+      "learning_rate": 3.804419110100377e-06,
+      "loss": 1.1409,
+      "step": 1729
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.747176112556268,
+      "learning_rate": 3.7938183361686854e-06,
+      "loss": 1.0139,
+      "step": 1730
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.9923727666708938,
+      "learning_rate": 3.7832288934981453e-06,
+      "loss": 1.1936,
+      "step": 1731
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.0066338279181262,
+      "learning_rate": 3.7726508014229825e-06,
+      "loss": 1.1178,
+      "step": 1732
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7793655481626994,
+      "learning_rate": 3.7620840792567025e-06,
+      "loss": 1.1323,
+      "step": 1733
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7452379313310221,
+      "learning_rate": 3.751528746292048e-06,
+      "loss": 1.0808,
+      "step": 1734
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.759767640031121,
+      "learning_rate": 3.740984821800979e-06,
+      "loss": 1.1586,
+      "step": 1735
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8138025107651605,
+      "learning_rate": 3.7304523250346134e-06,
+      "loss": 1.0444,
+      "step": 1736
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7963252756870868,
+      "learning_rate": 3.7199312752232053e-06,
+      "loss": 1.1319,
+      "step": 1737
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8147033841751047,
+      "learning_rate": 3.7094216915761192e-06,
+      "loss": 1.1644,
+      "step": 1738
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6819669154500375,
+      "learning_rate": 3.6989235932817733e-06,
+      "loss": 1.1348,
+      "step": 1739
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.799291277524351,
+      "learning_rate": 3.6884369995076232e-06,
+      "loss": 1.0186,
+      "step": 1740
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7301202593962056,
+      "learning_rate": 3.67796192940011e-06,
+      "loss": 1.1384,
+      "step": 1741
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.2466717373115086,
+      "learning_rate": 3.6674984020846503e-06,
+      "loss": 0.9794,
+      "step": 1742
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7897435975665426,
+      "learning_rate": 3.6570464366655724e-06,
+      "loss": 1.1706,
+      "step": 1743
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8064023115399381,
+      "learning_rate": 3.6466060522261005e-06,
+      "loss": 1.0913,
+      "step": 1744
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7445537935107824,
+      "learning_rate": 3.636177267828308e-06,
+      "loss": 1.1479,
+      "step": 1745
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6612127036442935,
+      "learning_rate": 3.625760102513103e-06,
+      "loss": 1.1489,
+      "step": 1746
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.017651785490463,
+      "learning_rate": 3.6153545753001663e-06,
+      "loss": 0.9742,
+      "step": 1747
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.709290711330306,
+      "learning_rate": 3.6049607051879345e-06,
+      "loss": 1.0257,
+      "step": 1748
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.268765447580064,
+      "learning_rate": 3.5945785111535547e-06,
+      "loss": 1.1568,
+      "step": 1749
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8148420746408352,
+      "learning_rate": 3.584208012152871e-06,
+      "loss": 1.1264,
+      "step": 1750
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 2.473987579345703,
+      "eval_runtime": 9.2976,
+      "eval_samples_per_second": 96.799,
+      "eval_steps_per_second": 0.215,
+      "step": 1750
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.766645298958262,
+      "learning_rate": 3.5738492271203608e-06,
+      "loss": 1.113,
+      "step": 1751
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8302425935428667,
+      "learning_rate": 3.563502174969117e-06,
+      "loss": 1.1092,
+      "step": 1752
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7536653979538627,
+      "learning_rate": 3.553166874590809e-06,
+      "loss": 1.08,
+      "step": 1753
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.71959776314368,
+      "learning_rate": 3.542843344855664e-06,
+      "loss": 1.1127,
+      "step": 1754
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.9132133684066353,
+      "learning_rate": 3.5325316046123925e-06,
+      "loss": 1.1173,
+      "step": 1755
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7810814888697305,
+      "learning_rate": 3.5222316726882033e-06,
+      "loss": 1.116,
+      "step": 1756
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.731839639636483,
+      "learning_rate": 3.5119435678887328e-06,
+      "loss": 1.1025,
+      "step": 1757
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7294781182607571,
+      "learning_rate": 3.5016673089980304e-06,
+      "loss": 1.1885,
+      "step": 1758
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7438163438336401,
+      "learning_rate": 3.4914029147785113e-06,
+      "loss": 1.1272,
+      "step": 1759
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6430990467593201,
+      "learning_rate": 3.481150403970929e-06,
+      "loss": 1.1062,
+      "step": 1760
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.649014205614028,
+      "learning_rate": 3.4709097952943483e-06,
+      "loss": 0.9344,
+      "step": 1761
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8877699152242934,
+      "learning_rate": 3.460681107446091e-06,
+      "loss": 1.193,
+      "step": 1762
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8686111945287311,
+      "learning_rate": 3.450464359101723e-06,
+      "loss": 1.0294,
+      "step": 1763
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7167047638882469,
+      "learning_rate": 3.4402595689150008e-06,
+      "loss": 1.1498,
+      "step": 1764
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7494621832507306,
+      "learning_rate": 3.430066755517868e-06,
+      "loss": 1.1167,
+      "step": 1765
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8406437035893782,
+      "learning_rate": 3.4198859375203707e-06,
+      "loss": 1.137,
+      "step": 1766
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7632879490334314,
+      "learning_rate": 3.409717133510683e-06,
+      "loss": 1.0619,
+      "step": 1767
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7171351861494293,
+      "learning_rate": 3.399560362055022e-06,
+      "loss": 1.141,
+      "step": 1768
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7229621684555863,
+      "learning_rate": 3.3894156416976587e-06,
+      "loss": 1.0811,
+      "step": 1769
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7319332736620845,
+      "learning_rate": 3.3792829909608305e-06,
+      "loss": 1.1585,
+      "step": 1770
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.9046432885007272,
+      "learning_rate": 3.3691624283447678e-06,
+      "loss": 1.1074,
+      "step": 1771
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7413367267968383,
+      "learning_rate": 3.3590539723276083e-06,
+      "loss": 1.12,
+      "step": 1772
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.2064899450844475,
+      "learning_rate": 3.348957641365408e-06,
+      "loss": 0.9994,
+      "step": 1773
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.9030247762180355,
+      "learning_rate": 3.338873453892056e-06,
+      "loss": 1.16,
+      "step": 1774
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.727248974607304,
+      "learning_rate": 3.328801428319297e-06,
+      "loss": 1.1187,
+      "step": 1775
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8343047797847679,
+      "learning_rate": 3.318741583036655e-06,
+      "loss": 1.1389,
+      "step": 1776
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.9183739077353188,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 1.0898,
+      "step": 1777
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7265091262767231,
+      "learning_rate": 3.2986585067886066e-06,
+      "loss": 1.0781,
+      "step": 1778
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7364349375538247,
+      "learning_rate": 3.2886353124909267e-06,
+      "loss": 0.9965,
+      "step": 1779
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7607221591427598,
+      "learning_rate": 3.2786243718187592e-06,
+      "loss": 1.161,
+      "step": 1780
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7688510209712639,
+      "learning_rate": 3.2686257030500924e-06,
+      "loss": 1.1145,
+      "step": 1781
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7021093332582873,
+      "learning_rate": 3.258639324440527e-06,
+      "loss": 1.0646,
+      "step": 1782
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8366439853211688,
+      "learning_rate": 3.2486652542232134e-06,
+      "loss": 1.1827,
+      "step": 1783
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6873613082009297,
+      "learning_rate": 3.238703510608834e-06,
+      "loss": 1.0756,
+      "step": 1784
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8107410958778014,
+      "learning_rate": 3.228754111785559e-06,
+      "loss": 1.0611,
+      "step": 1785
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.9281338115265906,
+      "learning_rate": 3.218817075919033e-06,
+      "loss": 1.1042,
+      "step": 1786
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7388358350255744,
+      "learning_rate": 3.208892421152314e-06,
+      "loss": 1.1956,
+      "step": 1787
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7728176131677671,
+      "learning_rate": 3.1989801656058604e-06,
+      "loss": 1.0476,
+      "step": 1788
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7386163456111818,
+      "learning_rate": 3.189080327377491e-06,
+      "loss": 1.1485,
+      "step": 1789
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8058785877046065,
+      "learning_rate": 3.179192924542357e-06,
+      "loss": 0.9335,
+      "step": 1790
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7846477192585686,
+      "learning_rate": 3.1693179751529e-06,
+      "loss": 1.1918,
+      "step": 1791
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8904437609444843,
+      "learning_rate": 3.159455497238827e-06,
+      "loss": 0.9989,
+      "step": 1792
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.9005442558360828,
+      "learning_rate": 3.1496055088070686e-06,
+      "loss": 1.1721,
+      "step": 1793
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.9578742569828589,
+      "learning_rate": 3.139768027841764e-06,
+      "loss": 1.067,
+      "step": 1794
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7272772862392133,
+      "learning_rate": 3.1299430723042066e-06,
+      "loss": 1.1695,
+      "step": 1795
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8900889243867139,
+      "learning_rate": 3.120130660132823e-06,
+      "loss": 1.0121,
+      "step": 1796
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7159332737616142,
+      "learning_rate": 3.110330809243134e-06,
+      "loss": 1.2049,
+      "step": 1797
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8486564708324941,
+      "learning_rate": 3.1005435375277393e-06,
+      "loss": 1.0335,
+      "step": 1798
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.9348630284619027,
+      "learning_rate": 3.090768862856258e-06,
+      "loss": 1.151,
+      "step": 1799
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7802248000175145,
+      "learning_rate": 3.0810068030753114e-06,
+      "loss": 1.0564,
+      "step": 1800
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 2.474574565887451,
+      "eval_runtime": 8.9249,
+      "eval_samples_per_second": 100.841,
+      "eval_steps_per_second": 0.224,
+      "step": 1800
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8284972381162343,
+      "learning_rate": 3.0712573760084984e-06,
+      "loss": 1.1992,
+      "step": 1801
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7191909955535986,
+      "learning_rate": 3.0615205994563412e-06,
+      "loss": 1.0933,
+      "step": 1802
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.9934379523855805,
+      "learning_rate": 3.0517964911962717e-06,
+      "loss": 1.1127,
+      "step": 1803
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8433998745157927,
+      "learning_rate": 3.042085068982584e-06,
+      "loss": 1.0856,
+      "step": 1804
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7155791236419289,
+      "learning_rate": 3.0323863505464233e-06,
+      "loss": 1.1387,
+      "step": 1805
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 3.4865655741536195,
+      "learning_rate": 3.02270035359573e-06,
+      "loss": 1.0734,
+      "step": 1806
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.9224316057188197,
+      "learning_rate": 3.0130270958152196e-06,
+      "loss": 1.15,
+      "step": 1807
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8114138671292138,
+      "learning_rate": 3.003366594866345e-06,
+      "loss": 1.061,
+      "step": 1808
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6645657534593112,
+      "learning_rate": 2.9937188683872796e-06,
+      "loss": 1.1176,
+      "step": 1809
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.7102005307665529,
+      "learning_rate": 2.984083933992862e-06,
+      "loss": 1.086,
+      "step": 1810
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.2312441345649827,
+      "learning_rate": 2.974461809274577e-06,
+      "loss": 1.1687,
+      "step": 1811
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7153286808127971,
+      "learning_rate": 2.964852511800519e-06,
+      "loss": 1.1021,
+      "step": 1812
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.791105935950199,
+      "learning_rate": 2.9552560591153746e-06,
+      "loss": 1.1084,
+      "step": 1813
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.765007413982625,
+      "learning_rate": 2.945672468740365e-06,
+      "loss": 1.0345,
+      "step": 1814
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7708246412204466,
+      "learning_rate": 2.936101758173233e-06,
+      "loss": 1.1763,
+      "step": 1815
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7734043753750518,
+      "learning_rate": 2.9265439448882005e-06,
+      "loss": 1.088,
+      "step": 1816
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6520007337026515,
+      "learning_rate": 2.9169990463359556e-06,
+      "loss": 1.0432,
+      "step": 1817
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7759562285217361,
+      "learning_rate": 2.9074670799435843e-06,
+      "loss": 1.1195,
+      "step": 1818
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7122932321455965,
+      "learning_rate": 2.8979480631145817e-06,
+      "loss": 1.0856,
+      "step": 1819
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8942843255831087,
+      "learning_rate": 2.888442013228787e-06,
+      "loss": 1.1178,
+      "step": 1820
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6577093712626482,
+      "learning_rate": 2.878948947642376e-06,
+      "loss": 1.0896,
+      "step": 1821
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7985373185801131,
+      "learning_rate": 2.869468883687798e-06,
+      "loss": 1.0855,
+      "step": 1822
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7479224621889771,
+      "learning_rate": 2.8600018386737828e-06,
+      "loss": 1.072,
+      "step": 1823
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7268025915662681,
+      "learning_rate": 2.8505478298852873e-06,
+      "loss": 1.1418,
+      "step": 1824
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8222801709386265,
+      "learning_rate": 2.841106874583459e-06,
+      "loss": 1.0366,
+      "step": 1825
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.754870746181915,
+      "learning_rate": 2.8316789900056173e-06,
+      "loss": 1.1559,
+      "step": 1826
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8167080667250239,
+      "learning_rate": 2.822264193365212e-06,
+      "loss": 1.0542,
+      "step": 1827
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.721605622270174,
+      "learning_rate": 2.8128625018518107e-06,
+      "loss": 1.1048,
+      "step": 1828
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.635893193364262,
+      "learning_rate": 2.803473932631032e-06,
+      "loss": 1.0666,
+      "step": 1829
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7134213778432719,
+      "learning_rate": 2.794098502844558e-06,
+      "loss": 1.1613,
+      "step": 1830
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7475448077115445,
+      "learning_rate": 2.784736229610061e-06,
+      "loss": 1.1429,
+      "step": 1831
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7945899890859678,
+      "learning_rate": 2.775387130021214e-06,
+      "loss": 1.1427,
+      "step": 1832
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7043496562094321,
+      "learning_rate": 2.7660512211476133e-06,
+      "loss": 1.0608,
+      "step": 1833
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7540945913886055,
+      "learning_rate": 2.7567285200347893e-06,
+      "loss": 1.0848,
+      "step": 1834
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8547460889367022,
+      "learning_rate": 2.747419043704148e-06,
+      "loss": 1.1926,
+      "step": 1835
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.2699549271583306,
+      "learning_rate": 2.738122809152962e-06,
+      "loss": 1.0744,
+      "step": 1836
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6690990331805525,
+      "learning_rate": 2.7288398333543063e-06,
+      "loss": 0.9497,
+      "step": 1837
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.017024551693902,
+      "learning_rate": 2.719570133257069e-06,
+      "loss": 1.1881,
+      "step": 1838
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.715844595289102,
+      "learning_rate": 2.7103137257858867e-06,
+      "loss": 0.975,
+      "step": 1839
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7867411918104382,
+      "learning_rate": 2.7010706278411314e-06,
+      "loss": 1.1216,
+      "step": 1840
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8490487588041102,
+      "learning_rate": 2.6918408562988686e-06,
+      "loss": 1.1291,
+      "step": 1841
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.66397017634904,
+      "learning_rate": 2.6826244280108438e-06,
+      "loss": 1.108,
+      "step": 1842
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8414357380342935,
+      "learning_rate": 2.673421359804431e-06,
+      "loss": 1.0205,
+      "step": 1843
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7719538461194015,
+      "learning_rate": 2.664231668482611e-06,
+      "loss": 1.1479,
+      "step": 1844
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.052980313603225,
+      "learning_rate": 2.655055370823948e-06,
+      "loss": 1.0447,
+      "step": 1845
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7313284175620615,
+      "learning_rate": 2.645892483582544e-06,
+      "loss": 1.178,
+      "step": 1846
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.366553527775459,
+      "learning_rate": 2.6367430234880286e-06,
+      "loss": 1.1482,
+      "step": 1847
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6899713926557357,
+      "learning_rate": 2.627607007245496e-06,
+      "loss": 1.0939,
+      "step": 1848
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7940731423758448,
+      "learning_rate": 2.6184844515355146e-06,
+      "loss": 1.006,
+      "step": 1849
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6904909860530358,
+      "learning_rate": 2.609375373014065e-06,
+      "loss": 1.2012,
+      "step": 1850
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 2.553161382675171,
+      "eval_runtime": 9.3445,
+      "eval_samples_per_second": 96.313,
+      "eval_steps_per_second": 0.214,
+      "step": 1850
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7197517543448332,
+      "learning_rate": 2.600279788312524e-06,
+      "loss": 0.9646,
+      "step": 1851
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7331842128717667,
+      "learning_rate": 2.591197714037631e-06,
+      "loss": 1.1668,
+      "step": 1852
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7433794804078062,
+      "learning_rate": 2.582129166771462e-06,
+      "loss": 1.0915,
+      "step": 1853
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6608752792920786,
+      "learning_rate": 2.5730741630713896e-06,
+      "loss": 1.1366,
+      "step": 1854
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7441137271152304,
+      "learning_rate": 2.564032719470062e-06,
+      "loss": 1.1224,
+      "step": 1855
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.5268027864597205,
+      "learning_rate": 2.5550048524753645e-06,
+      "loss": 1.0339,
+      "step": 1856
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7039936821779065,
+      "learning_rate": 2.545990578570404e-06,
+      "loss": 1.0815,
+      "step": 1857
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7494274315571515,
+      "learning_rate": 2.5369899142134613e-06,
+      "loss": 1.1505,
+      "step": 1858
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.697432939789229,
+      "learning_rate": 2.528002875837968e-06,
+      "loss": 1.1646,
+      "step": 1859
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7389670345221018,
+      "learning_rate": 2.5190294798524805e-06,
+      "loss": 1.1075,
+      "step": 1860
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7804171701312074,
+      "learning_rate": 2.5100697426406496e-06,
+      "loss": 1.1648,
+      "step": 1861
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7886514408482154,
+      "learning_rate": 2.5011236805611818e-06,
+      "loss": 1.0943,
+      "step": 1862
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7133228967805424,
+      "learning_rate": 2.4921913099478203e-06,
+      "loss": 1.0363,
+      "step": 1863
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.9358076897167364,
+      "learning_rate": 2.4832726471093037e-06,
+      "loss": 1.1514,
+      "step": 1864
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7808464021439907,
+      "learning_rate": 2.4743677083293537e-06,
+      "loss": 1.1427,
+      "step": 1865
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6815589206358184,
+      "learning_rate": 2.465476509866628e-06,
+      "loss": 1.0385,
+      "step": 1866
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7478817193345602,
+      "learning_rate": 2.4565990679546913e-06,
+      "loss": 1.1839,
+      "step": 1867
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6850514920746833,
+      "learning_rate": 2.447735398802007e-06,
+      "loss": 0.9549,
+      "step": 1868
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.710407008035692,
+      "learning_rate": 2.438885518591878e-06,
+      "loss": 1.1343,
+      "step": 1869
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7766150460312187,
+      "learning_rate": 2.4300494434824373e-06,
+      "loss": 1.0768,
+      "step": 1870
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6923290675203712,
+      "learning_rate": 2.42122718960661e-06,
+      "loss": 1.1938,
+      "step": 1871
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.0187006931694613,
+      "learning_rate": 2.4124187730720916e-06,
+      "loss": 1.0755,
+      "step": 1872
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.9031120431022819,
+      "learning_rate": 2.4036242099613085e-06,
+      "loss": 1.1218,
+      "step": 1873
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.154151151823336,
+      "learning_rate": 2.3948435163313942e-06,
+      "loss": 1.0349,
+      "step": 1874
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7501466617475971,
+      "learning_rate": 2.3860767082141567e-06,
+      "loss": 1.1203,
+      "step": 1875
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.0466410768989693,
+      "learning_rate": 2.3773238016160614e-06,
+      "loss": 1.0692,
+      "step": 1876
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8121215606457608,
+      "learning_rate": 2.3685848125181843e-06,
+      "loss": 1.173,
+      "step": 1877
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7652521644200053,
+      "learning_rate": 2.3598597568761893e-06,
+      "loss": 1.047,
+      "step": 1878
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7950613972034314,
+      "learning_rate": 2.351148650620303e-06,
+      "loss": 1.1563,
+      "step": 1879
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8008696244257412,
+      "learning_rate": 2.342451509655289e-06,
+      "loss": 1.0149,
+      "step": 1880
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7036202522033553,
+      "learning_rate": 2.3337683498604045e-06,
+      "loss": 1.0728,
+      "step": 1881
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8075647768883898,
+      "learning_rate": 2.3250991870893837e-06,
+      "loss": 1.11,
+      "step": 1882
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7491985959344515,
+      "learning_rate": 2.3164440371704e-06,
+      "loss": 1.1456,
+      "step": 1883
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7032311511328233,
+      "learning_rate": 2.307802915906058e-06,
+      "loss": 1.08,
+      "step": 1884
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8480133979856301,
+      "learning_rate": 2.2991758390733243e-06,
+      "loss": 1.1901,
+      "step": 1885
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8600640608282438,
+      "learning_rate": 2.290562822423541e-06,
+      "loss": 1.0425,
+      "step": 1886
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6683230160477834,
+      "learning_rate": 2.2819638816823796e-06,
+      "loss": 1.0768,
+      "step": 1887
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8474724814381165,
+      "learning_rate": 2.2733790325498005e-06,
+      "loss": 0.9886,
+      "step": 1888
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7042345507556335,
+      "learning_rate": 2.2648082907000445e-06,
+      "loss": 1.159,
+      "step": 1889
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7235419927396182,
+      "learning_rate": 2.2562516717815874e-06,
+      "loss": 1.0807,
+      "step": 1890
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.784166276352502,
+      "learning_rate": 2.2477091914171333e-06,
+      "loss": 1.1426,
+      "step": 1891
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7366930478512308,
+      "learning_rate": 2.239180865203552e-06,
+      "loss": 1.1929,
+      "step": 1892
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.724833799860843,
+      "learning_rate": 2.2306667087118906e-06,
+      "loss": 1.0325,
+      "step": 1893
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7718263179632028,
+      "learning_rate": 2.222166737487309e-06,
+      "loss": 1.1213,
+      "step": 1894
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.9381035960922299,
+      "learning_rate": 2.2136809670490845e-06,
+      "loss": 1.1454,
+      "step": 1895
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.210392999460702,
+      "learning_rate": 2.2052094128905454e-06,
+      "loss": 1.0335,
+      "step": 1896
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.904341673013862,
+      "learning_rate": 2.196752090479083e-06,
+      "loss": 1.0914,
+      "step": 1897
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7209811777401025,
+      "learning_rate": 2.1883090152560916e-06,
+      "loss": 1.1923,
+      "step": 1898
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8390518334829505,
+      "learning_rate": 2.179880202636966e-06,
+      "loss": 1.099,
+      "step": 1899
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.824396415158213,
+      "learning_rate": 2.1714656680110413e-06,
+      "loss": 1.0464,
+      "step": 1900
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 2.570903778076172,
+      "eval_runtime": 10.2681,
+      "eval_samples_per_second": 87.65,
+      "eval_steps_per_second": 0.195,
+      "step": 1900
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8925985422985897,
+      "learning_rate": 2.163065426741603e-06,
+      "loss": 1.2199,
+      "step": 1901
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7502605482671812,
+      "learning_rate": 2.154679494165829e-06,
+      "loss": 1.1641,
+      "step": 1902
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8288214349142767,
+      "learning_rate": 2.146307885594774e-06,
+      "loss": 1.0478,
+      "step": 1903
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7457530730614095,
+      "learning_rate": 2.137950616313338e-06,
+      "loss": 1.0946,
+      "step": 1904
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.9558336777378081,
+      "learning_rate": 2.1296077015802487e-06,
+      "loss": 1.0409,
+      "step": 1905
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7595674803447695,
+      "learning_rate": 2.121279156628018e-06,
+      "loss": 1.1415,
+      "step": 1906
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.3013156561933583,
+      "learning_rate": 2.1129649966629185e-06,
+      "loss": 1.1091,
+      "step": 1907
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7620584735634007,
+      "learning_rate": 2.104665236864969e-06,
+      "loss": 1.1498,
+      "step": 1908
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6898202065642673,
+      "learning_rate": 2.0963798923878855e-06,
+      "loss": 1.0219,
+      "step": 1909
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.9220967550537517,
+      "learning_rate": 2.088108978359078e-06,
+      "loss": 1.1798,
+      "step": 1910
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7835436203495111,
+      "learning_rate": 2.0798525098795897e-06,
+      "loss": 1.085,
+      "step": 1911
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.014801356056499,
+      "learning_rate": 2.0716105020241074e-06,
+      "loss": 1.115,
+      "step": 1912
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7464543298540225,
+      "learning_rate": 2.0633829698409067e-06,
+      "loss": 1.1043,
+      "step": 1913
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7300230321807916,
+      "learning_rate": 2.0551699283518355e-06,
+      "loss": 1.0565,
+      "step": 1914
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7352858209081881,
+      "learning_rate": 2.0469713925522816e-06,
+      "loss": 1.0538,
+      "step": 1915
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7126821788645525,
+      "learning_rate": 2.0387873774111567e-06,
+      "loss": 1.1345,
+      "step": 1916
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.774904970114039,
+      "learning_rate": 2.030617897870851e-06,
+      "loss": 1.0879,
+      "step": 1917
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7575618140642042,
+      "learning_rate": 2.022462968847221e-06,
+      "loss": 1.0295,
+      "step": 1918
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.06155007297502,
+      "learning_rate": 2.0143226052295515e-06,
+      "loss": 1.0877,
+      "step": 1919
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.613511284854979,
+      "learning_rate": 2.006196821880544e-06,
+      "loss": 1.1406,
+      "step": 1920
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6532527081940871,
+      "learning_rate": 1.99808563363627e-06,
+      "loss": 1.1099,
+      "step": 1921
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.829990707205679,
+      "learning_rate": 1.9899890553061565e-06,
+      "loss": 1.0909,
+      "step": 1922
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.732305103253903,
+      "learning_rate": 1.9819071016729506e-06,
+      "loss": 1.0842,
+      "step": 1923
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.865178208534143,
+      "learning_rate": 1.9738397874927107e-06,
+      "loss": 1.1357,
+      "step": 1924
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7612567774801967,
+      "learning_rate": 1.9657871274947527e-06,
+      "loss": 1.1438,
+      "step": 1925
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7323970560244569,
+      "learning_rate": 1.957749136381645e-06,
+      "loss": 1.1076,
+      "step": 1926
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7344652012387805,
+      "learning_rate": 1.9497258288291655e-06,
+      "loss": 1.0677,
+      "step": 1927
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8418470071226791,
+      "learning_rate": 1.9417172194862955e-06,
+      "loss": 1.0989,
+      "step": 1928
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.787486064678821,
+      "learning_rate": 1.9337233229751686e-06,
+      "loss": 1.0558,
+      "step": 1929
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7462949558512176,
+      "learning_rate": 1.9257441538910583e-06,
+      "loss": 1.196,
+      "step": 1930
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7031335472029691,
+      "learning_rate": 1.9177797268023545e-06,
+      "loss": 1.1175,
+      "step": 1931
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7613561295299239,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 1.134,
+      "step": 1932
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6942119319292741,
+      "learning_rate": 1.9018951567501e-06,
+      "loss": 1.0369,
+      "step": 1933
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.9683509762938756,
+      "learning_rate": 1.8939750427886305e-06,
+      "loss": 1.1469,
+      "step": 1934
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7332394632713803,
+      "learning_rate": 1.886069728826686e-06,
+      "loss": 1.0346,
+      "step": 1935
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7264750008409483,
+      "learning_rate": 1.8781792292978062e-06,
+      "loss": 1.0531,
+      "step": 1936
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7081328581868375,
+      "learning_rate": 1.8703035586084817e-06,
+      "loss": 1.141,
+      "step": 1937
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7896622211086514,
+      "learning_rate": 1.86244273113813e-06,
+      "loss": 1.1447,
+      "step": 1938
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.750652449846152,
+      "learning_rate": 1.8545967612390725e-06,
+      "loss": 1.0505,
+      "step": 1939
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7785609019402691,
+      "learning_rate": 1.8467656632364983e-06,
+      "loss": 1.0097,
+      "step": 1940
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8752238570890374,
+      "learning_rate": 1.8389494514284422e-06,
+      "loss": 1.1693,
+      "step": 1941
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6897771475903249,
+      "learning_rate": 1.8311481400857622e-06,
+      "loss": 1.0555,
+      "step": 1942
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8214665874516756,
+      "learning_rate": 1.8233617434521144e-06,
+      "loss": 1.1537,
+      "step": 1943
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.721743369828361,
+      "learning_rate": 1.8155902757439181e-06,
+      "loss": 1.065,
+      "step": 1944
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7319724390891392,
+      "learning_rate": 1.8078337511503386e-06,
+      "loss": 1.1413,
+      "step": 1945
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8920584685211916,
+      "learning_rate": 1.800092183833252e-06,
+      "loss": 1.101,
+      "step": 1946
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6923510591226858,
+      "learning_rate": 1.7923655879272395e-06,
+      "loss": 1.0642,
+      "step": 1947
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7049630809719629,
+      "learning_rate": 1.7846539775395277e-06,
+      "loss": 1.0766,
+      "step": 1948
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8085614522181166,
+      "learning_rate": 1.7769573667500007e-06,
+      "loss": 1.1655,
+      "step": 1949
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7984508564216256,
+      "learning_rate": 1.769275769611143e-06,
+      "loss": 1.027,
+      "step": 1950
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 2.54919695854187,
+      "eval_runtime": 10.1168,
+      "eval_samples_per_second": 88.961,
+      "eval_steps_per_second": 0.198,
+      "step": 1950
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7011369411114936,
+      "learning_rate": 1.7616092001480378e-06,
+      "loss": 1.1129,
+      "step": 1951
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7819861147704675,
+      "learning_rate": 1.753957672358324e-06,
+      "loss": 1.1034,
+      "step": 1952
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7277780316461507,
+      "learning_rate": 1.7463212002121766e-06,
+      "loss": 1.1491,
+      "step": 1953
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8852498324069826,
+      "learning_rate": 1.7386997976522924e-06,
+      "loss": 1.053,
+      "step": 1954
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7005734998647326,
+      "learning_rate": 1.731093478593835e-06,
+      "loss": 1.1603,
+      "step": 1955
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.230004193868005,
+      "learning_rate": 1.7235022569244475e-06,
+      "loss": 1.182,
+      "step": 1956
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7161015708814772,
+      "learning_rate": 1.7159261465041954e-06,
+      "loss": 1.0388,
+      "step": 1957
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8097550815323062,
+      "learning_rate": 1.7083651611655672e-06,
+      "loss": 1.061,
+      "step": 1958
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7764001876116498,
+      "learning_rate": 1.7008193147134167e-06,
+      "loss": 1.1904,
+      "step": 1959
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.6626576294426323,
+      "learning_rate": 1.6932886209249767e-06,
+      "loss": 1.0702,
+      "step": 1960
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7500514188369182,
+      "learning_rate": 1.6857730935498008e-06,
+      "loss": 1.1796,
+      "step": 1961
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7382554505849426,
+      "learning_rate": 1.6782727463097626e-06,
+      "loss": 1.0404,
+      "step": 1962
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8175769463011858,
+      "learning_rate": 1.6707875928990059e-06,
+      "loss": 1.0182,
+      "step": 1963
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.9076362924945748,
+      "learning_rate": 1.6633176469839474e-06,
+      "loss": 1.1364,
+      "step": 1964
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7380902425643496,
+      "learning_rate": 1.6558629222032297e-06,
+      "loss": 1.1365,
+      "step": 1965
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8009434067111979,
+      "learning_rate": 1.6484234321677083e-06,
+      "loss": 0.9791,
+      "step": 1966
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7160729788774459,
+      "learning_rate": 1.6409991904604173e-06,
+      "loss": 1.2347,
+      "step": 1967
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7805789899559521,
+      "learning_rate": 1.633590210636561e-06,
+      "loss": 1.0849,
+      "step": 1968
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7986777554162275,
+      "learning_rate": 1.6261965062234685e-06,
+      "loss": 1.0352,
+      "step": 1969
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7880384784568544,
+      "learning_rate": 1.6188180907205852e-06,
+      "loss": 1.156,
+      "step": 1970
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8812296065094827,
+      "learning_rate": 1.6114549775994348e-06,
+      "loss": 0.9625,
+      "step": 1971
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.7554998427454955,
+      "learning_rate": 1.60410718030361e-06,
+      "loss": 1.057,
+      "step": 1972
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.905853819835457,
+      "learning_rate": 1.5967747122487408e-06,
+      "loss": 1.1819,
+      "step": 1973
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.160757500469259,
+      "learning_rate": 1.589457586822456e-06,
+      "loss": 1.1541,
+      "step": 1974
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7525111180737948,
+      "learning_rate": 1.5821558173843876e-06,
+      "loss": 1.0419,
+      "step": 1975
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7769467142564381,
+      "learning_rate": 1.5748694172661194e-06,
+      "loss": 1.1418,
+      "step": 1976
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7492589576300304,
+      "learning_rate": 1.5675983997711797e-06,
+      "loss": 1.146,
+      "step": 1977
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.0753590911048172,
+      "learning_rate": 1.5603427781750068e-06,
+      "loss": 1.1077,
+      "step": 1978
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.9778097844183965,
+      "learning_rate": 1.5531025657249366e-06,
+      "loss": 1.076,
+      "step": 1979
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7189284229853226,
+      "learning_rate": 1.5458777756401621e-06,
+      "loss": 1.1621,
+      "step": 1980
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.058500668782446,
+      "learning_rate": 1.5386684211117242e-06,
+      "loss": 1.0461,
+      "step": 1981
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7278042618855339,
+      "learning_rate": 1.5314745153024768e-06,
+      "loss": 1.0892,
+      "step": 1982
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8693664193224692,
+      "learning_rate": 1.5242960713470734e-06,
+      "loss": 1.1049,
+      "step": 1983
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.670146161240364,
+      "learning_rate": 1.5171331023519331e-06,
+      "loss": 1.1308,
+      "step": 1984
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7210051088770537,
+      "learning_rate": 1.5099856213952203e-06,
+      "loss": 1.1522,
+      "step": 1985
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8506719608033633,
+      "learning_rate": 1.5028536415268214e-06,
+      "loss": 1.1152,
+      "step": 1986
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7164263207767808,
+      "learning_rate": 1.4957371757683258e-06,
+      "loss": 1.0629,
+      "step": 1987
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.785500279158162,
+      "learning_rate": 1.4886362371129926e-06,
+      "loss": 1.0851,
+      "step": 1988
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8089499007837019,
+      "learning_rate": 1.4815508385257316e-06,
+      "loss": 1.0958,
+      "step": 1989
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7559360158927676,
+      "learning_rate": 1.474480992943077e-06,
+      "loss": 1.1037,
+      "step": 1990
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6719801718122285,
+      "learning_rate": 1.4674267132731757e-06,
+      "loss": 0.9502,
+      "step": 1991
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7154548931772431,
+      "learning_rate": 1.4603880123957448e-06,
+      "loss": 1.1545,
+      "step": 1992
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8350182906881023,
+      "learning_rate": 1.45336490316206e-06,
+      "loss": 1.074,
+      "step": 1993
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7829584249627557,
+      "learning_rate": 1.446357398394934e-06,
+      "loss": 1.1482,
+      "step": 1994
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.0801045761819394,
+      "learning_rate": 1.4393655108886839e-06,
+      "loss": 1.0947,
+      "step": 1995
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.4680064353962978,
+      "learning_rate": 1.4323892534091132e-06,
+      "loss": 1.0946,
+      "step": 1996
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7828869463675876,
+      "learning_rate": 1.425428638693489e-06,
+      "loss": 1.0582,
+      "step": 1997
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.9072736534835041,
+      "learning_rate": 1.4184836794505218e-06,
+      "loss": 1.1755,
+      "step": 1998
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.697357487566644,
+      "learning_rate": 1.4115543883603321e-06,
+      "loss": 1.0366,
+      "step": 1999
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8159034926630464,
+      "learning_rate": 1.4046407780744364e-06,
+      "loss": 1.0493,
+      "step": 2000
+    },
+    {
+      "epoch": 0.83,
+      "eval_loss": 2.5603044033050537,
+      "eval_runtime": 9.018,
+      "eval_samples_per_second": 99.8,
+      "eval_steps_per_second": 0.222,
+      "step": 2000
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8227887092386347,
+      "learning_rate": 1.3977428612157185e-06,
+      "loss": 1.1196,
+      "step": 2001
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7484370874170209,
+      "learning_rate": 1.390860650378414e-06,
+      "loss": 1.1141,
+      "step": 2002
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.668822263474963,
+      "learning_rate": 1.3839941581280803e-06,
+      "loss": 1.1132,
+      "step": 2003
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.9383794182426183,
+      "learning_rate": 1.3771433970015735e-06,
+      "loss": 1.0689,
+      "step": 2004
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7622914993108517,
+      "learning_rate": 1.3703083795070272e-06,
+      "loss": 1.1245,
+      "step": 2005
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7489961794986061,
+      "learning_rate": 1.3634891181238373e-06,
+      "loss": 1.1045,
+      "step": 2006
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.1946489688956332,
+      "learning_rate": 1.356685625302625e-06,
+      "loss": 1.1956,
+      "step": 2007
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 23.68283229720515,
+      "learning_rate": 1.3498979134652235e-06,
+      "loss": 1.0652,
+      "step": 2008
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.714993662121855,
+      "learning_rate": 1.3431259950046515e-06,
+      "loss": 1.0916,
+      "step": 2009
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8020667130474914,
+      "learning_rate": 1.3363698822851001e-06,
+      "loss": 1.0624,
+      "step": 2010
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.1612217893098749,
+      "learning_rate": 1.3296295876418862e-06,
+      "loss": 1.1467,
+      "step": 2011
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8211326898326504,
+      "learning_rate": 1.3229051233814637e-06,
+      "loss": 1.029,
+      "step": 2012
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7220897203574681,
+      "learning_rate": 1.3161965017813704e-06,
+      "loss": 0.9837,
+      "step": 2013
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7674695524080589,
+      "learning_rate": 1.3095037350902272e-06,
+      "loss": 1.076,
+      "step": 2014
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7849602152555869,
+      "learning_rate": 1.3028268355277018e-06,
+      "loss": 1.1719,
+      "step": 2015
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8305159636718255,
+      "learning_rate": 1.2961658152844902e-06,
+      "loss": 1.014,
+      "step": 2016
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7548898069823261,
+      "learning_rate": 1.2895206865223065e-06,
+      "loss": 1.0496,
+      "step": 2017
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8373163026804219,
+      "learning_rate": 1.282891461373833e-06,
+      "loss": 1.1041,
+      "step": 2018
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7167162883108626,
+      "learning_rate": 1.2762781519427324e-06,
+      "loss": 1.1896,
+      "step": 2019
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7865762355902203,
+      "learning_rate": 1.2696807703035953e-06,
+      "loss": 1.0429,
+      "step": 2020
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7863169711791046,
+      "learning_rate": 1.2630993285019432e-06,
+      "loss": 1.1963,
+      "step": 2021
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8567591189189535,
+      "learning_rate": 1.2565338385541792e-06,
+      "loss": 1.0524,
+      "step": 2022
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7667767583311532,
+      "learning_rate": 1.2499843124475974e-06,
+      "loss": 1.071,
+      "step": 2023
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.2141624293375601,
+      "learning_rate": 1.2434507621403336e-06,
+      "loss": 1.0526,
+      "step": 2024
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8419774172544316,
+      "learning_rate": 1.2369331995613664e-06,
+      "loss": 1.1685,
+      "step": 2025
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7798070638626042,
+      "learning_rate": 1.2304316366104673e-06,
+      "loss": 1.0964,
+      "step": 2026
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6699478081783663,
+      "learning_rate": 1.2239460851582118e-06,
+      "loss": 1.1099,
+      "step": 2027
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8362641830513503,
+      "learning_rate": 1.2174765570459335e-06,
+      "loss": 1.0493,
+      "step": 2028
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9578144120569076,
+      "learning_rate": 1.2110230640857123e-06,
+      "loss": 1.1187,
+      "step": 2029
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.731343560405218,
+      "learning_rate": 1.2045856180603488e-06,
+      "loss": 1.0142,
+      "step": 2030
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8482253247777753,
+      "learning_rate": 1.198164230723351e-06,
+      "loss": 1.105,
+      "step": 2031
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9135614020731123,
+      "learning_rate": 1.1917589137989006e-06,
+      "loss": 1.0859,
+      "step": 2032
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7598477867051728,
+      "learning_rate": 1.1853696789818404e-06,
+      "loss": 1.1703,
+      "step": 2033
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7150011922522083,
+      "learning_rate": 1.1789965379376488e-06,
+      "loss": 1.1375,
+      "step": 2034
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9276625104106015,
+      "learning_rate": 1.172639502302425e-06,
+      "loss": 1.0609,
+      "step": 2035
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8016341080500344,
+      "learning_rate": 1.166298583682861e-06,
+      "loss": 1.0311,
+      "step": 2036
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8704470998470023,
+      "learning_rate": 1.159973793656215e-06,
+      "loss": 1.176,
+      "step": 2037
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7389616723893224,
+      "learning_rate": 1.1536651437703095e-06,
+      "loss": 1.0541,
+      "step": 2038
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6775656620174616,
+      "learning_rate": 1.1473726455434898e-06,
+      "loss": 1.0349,
+      "step": 2039
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8271502886198017,
+      "learning_rate": 1.1410963104646144e-06,
+      "loss": 1.1806,
+      "step": 2040
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7190076899525927,
+      "learning_rate": 1.1348361499930272e-06,
+      "loss": 1.0576,
+      "step": 2041
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.7156120460750626,
+      "learning_rate": 1.1285921755585504e-06,
+      "loss": 1.0811,
+      "step": 2042
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9075653607684799,
+      "learning_rate": 1.122364398561444e-06,
+      "loss": 1.113,
+      "step": 2043
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.658026458230634,
+      "learning_rate": 1.1161528303723978e-06,
+      "loss": 1.1395,
+      "step": 2044
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6661716227778847,
+      "learning_rate": 1.1099574823325065e-06,
+      "loss": 1.0682,
+      "step": 2045
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6937146065826646,
+      "learning_rate": 1.1037783657532542e-06,
+      "loss": 1.1511,
+      "step": 2046
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7084018843202262,
+      "learning_rate": 1.097615491916485e-06,
+      "loss": 0.8856,
+      "step": 2047
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8718152218669916,
+      "learning_rate": 1.0914688720743894e-06,
+      "loss": 1.1739,
+      "step": 2048
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.2326887162861835,
+      "learning_rate": 1.0853385174494758e-06,
+      "loss": 1.1205,
+      "step": 2049
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.001256123322539,
+      "learning_rate": 1.0792244392345674e-06,
+      "loss": 1.1522,
+      "step": 2050
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 2.5704681873321533,
+      "eval_runtime": 10.4175,
+      "eval_samples_per_second": 86.393,
+      "eval_steps_per_second": 0.192,
+      "step": 2050
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.793198474752761,
+      "learning_rate": 1.0731266485927584e-06,
+      "loss": 1.0072,
+      "step": 2051
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.0941354542849584,
+      "learning_rate": 1.0670451566574102e-06,
+      "loss": 1.1661,
+      "step": 2052
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.831004319446031,
+      "learning_rate": 1.0609799745321235e-06,
+      "loss": 1.098,
+      "step": 2053
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.923176047393094,
+      "learning_rate": 1.0549311132907258e-06,
+      "loss": 1.1728,
+      "step": 2054
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8908927963862338,
+      "learning_rate": 1.0488985839772404e-06,
+      "loss": 1.1154,
+      "step": 2055
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.815944557889252,
+      "learning_rate": 1.042882397605871e-06,
+      "loss": 1.1879,
+      "step": 2056
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6996945811306433,
+      "learning_rate": 1.0368825651609893e-06,
+      "loss": 1.0023,
+      "step": 2057
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7552451150443389,
+      "learning_rate": 1.0308990975971012e-06,
+      "loss": 1.182,
+      "step": 2058
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.0588933634507476,
+      "learning_rate": 1.0249320058388367e-06,
+      "loss": 1.1269,
+      "step": 2059
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.937425720690437,
+      "learning_rate": 1.0189813007809235e-06,
+      "loss": 1.128,
+      "step": 2060
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7378626218783119,
+      "learning_rate": 1.013046993288177e-06,
+      "loss": 0.9235,
+      "step": 2061
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.745086751893708,
+      "learning_rate": 1.007129094195468e-06,
+      "loss": 1.1568,
+      "step": 2062
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7257518128758137,
+      "learning_rate": 1.0012276143077104e-06,
+      "loss": 1.0736,
+      "step": 2063
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7736619497703497,
+      "learning_rate": 9.953425643998381e-07,
+      "loss": 1.1191,
+      "step": 2064
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8054854156468928,
+      "learning_rate": 9.89473955216793e-07,
+      "loss": 1.1869,
+      "step": 2065
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7683262390872675,
+      "learning_rate": 9.836217974734941e-07,
+      "loss": 1.0388,
+      "step": 2066
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.3748371703631146,
+      "learning_rate": 9.77786101854825e-07,
+      "loss": 1.0564,
+      "step": 2067
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8770656638243364,
+      "learning_rate": 9.7196687901561e-07,
+      "loss": 1.148,
+      "step": 2068
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7191645584388003,
+      "learning_rate": 9.661641395806042e-07,
+      "loss": 1.125,
+      "step": 2069
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8173989301660586,
+      "learning_rate": 9.603778941444608e-07,
+      "loss": 1.0976,
+      "step": 2070
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8133235720647812,
+      "learning_rate": 9.546081532717199e-07,
+      "loss": 1.1148,
+      "step": 2071
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7881691726859379,
+      "learning_rate": 9.488549274967873e-07,
+      "loss": 1.1017,
+      "step": 2072
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.3116601263664847,
+      "learning_rate": 9.431182273239215e-07,
+      "loss": 1.1253,
+      "step": 2073
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8434700955354113,
+      "learning_rate": 9.373980632271951e-07,
+      "loss": 1.1097,
+      "step": 2074
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7412731683177509,
+      "learning_rate": 9.316944456505029e-07,
+      "loss": 1.1018,
+      "step": 2075
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8203717853634255,
+      "learning_rate": 9.260073850075213e-07,
+      "loss": 1.0892,
+      "step": 2076
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8094591749874104,
+      "learning_rate": 9.203368916817012e-07,
+      "loss": 1.1608,
+      "step": 2077
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.678814760566094,
+      "learning_rate": 9.146829760262388e-07,
+      "loss": 0.9275,
+      "step": 2078
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.914857044743798,
+      "learning_rate": 9.090456483640686e-07,
+      "loss": 1.159,
+      "step": 2079
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7595477667519405,
+      "learning_rate": 9.034249189878397e-07,
+      "loss": 1.0408,
+      "step": 2080
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7412010831495949,
+      "learning_rate": 8.978207981598896e-07,
+      "loss": 1.1542,
+      "step": 2081
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.815044004248336,
+      "learning_rate": 8.92233296112236e-07,
+      "loss": 1.0782,
+      "step": 2082
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7580280125472493,
+      "learning_rate": 8.866624230465526e-07,
+      "loss": 1.136,
+      "step": 2083
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7748415948816665,
+      "learning_rate": 8.811081891341567e-07,
+      "loss": 1.0405,
+      "step": 2084
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.9520599344766347,
+      "learning_rate": 8.755706045159751e-07,
+      "loss": 1.0889,
+      "step": 2085
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8350183430075032,
+      "learning_rate": 8.700496793025481e-07,
+      "loss": 1.0766,
+      "step": 2086
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7423752365732789,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 1.1199,
+      "step": 2087
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.9041855480443871,
+      "learning_rate": 8.590578473799905e-07,
+      "loss": 1.0801,
+      "step": 2088
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7813494508579112,
+      "learning_rate": 8.535869607397718e-07,
+      "loss": 1.1637,
+      "step": 2089
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7634764751784001,
+      "learning_rate": 8.481327736420952e-07,
+      "loss": 1.0038,
+      "step": 2090
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.9078117482427072,
+      "learning_rate": 8.426952960452284e-07,
+      "loss": 1.1794,
+      "step": 2091
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7761818115333792,
+      "learning_rate": 8.37274537876931e-07,
+      "loss": 0.9824,
+      "step": 2092
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7069363380296881,
+      "learning_rate": 8.318705090344337e-07,
+      "loss": 1.127,
+      "step": 2093
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8184621144548873,
+      "learning_rate": 8.264832193844274e-07,
+      "loss": 1.0999,
+      "step": 2094
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7629458092901217,
+      "learning_rate": 8.211126787630386e-07,
+      "loss": 1.1311,
+      "step": 2095
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8573173027908065,
+      "learning_rate": 8.157588969758123e-07,
+      "loss": 1.0899,
+      "step": 2096
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7334651567554206,
+      "learning_rate": 8.10421883797694e-07,
+      "loss": 1.1365,
+      "step": 2097
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7936897365930254,
+      "learning_rate": 8.05101648973019e-07,
+      "loss": 1.0218,
+      "step": 2098
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.687055307300093,
+      "learning_rate": 7.997982022154816e-07,
+      "loss": 1.1259,
+      "step": 2099
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.9383718114466909,
+      "learning_rate": 7.945115532081271e-07,
+      "loss": 1.109,
+      "step": 2100
+    },
+    {
+      "epoch": 0.88,
+      "eval_loss": 2.5815186500549316,
+      "eval_runtime": 9.7273,
+      "eval_samples_per_second": 92.523,
+      "eval_steps_per_second": 0.206,
+      "step": 2100
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7735873052789298,
+      "learning_rate": 7.892417116033346e-07,
+      "loss": 1.1788,
+      "step": 2101
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.0516150352461262,
+      "learning_rate": 7.839886870227909e-07,
+      "loss": 1.0699,
+      "step": 2102
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7818943182814504,
+      "learning_rate": 7.787524890574827e-07,
+      "loss": 1.1616,
+      "step": 2103
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.9297299560129376,
+      "learning_rate": 7.735331272676683e-07,
+      "loss": 1.0405,
+      "step": 2104
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.070108768314522,
+      "learning_rate": 7.683306111828769e-07,
+      "loss": 1.1116,
+      "step": 2105
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8316074053832151,
+      "learning_rate": 7.631449503018706e-07,
+      "loss": 1.0453,
+      "step": 2106
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7613216022262796,
+      "learning_rate": 7.579761540926434e-07,
+      "loss": 1.1547,
+      "step": 2107
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6775692629462753,
+      "learning_rate": 7.528242319923918e-07,
+      "loss": 1.0744,
+      "step": 2108
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.3769316656385213,
+      "learning_rate": 7.476891934075126e-07,
+      "loss": 1.0811,
+      "step": 2109
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8726304914624492,
+      "learning_rate": 7.425710477135684e-07,
+      "loss": 1.115,
+      "step": 2110
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.1054095770602088,
+      "learning_rate": 7.374698042552819e-07,
+      "loss": 1.0582,
+      "step": 2111
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8536168118200469,
+      "learning_rate": 7.32385472346514e-07,
+      "loss": 1.1543,
+      "step": 2112
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8305006124122815,
+      "learning_rate": 7.273180612702546e-07,
+      "loss": 1.1578,
+      "step": 2113
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7528444686235233,
+      "learning_rate": 7.222675802785906e-07,
+      "loss": 1.1179,
+      "step": 2114
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7620524820572331,
+      "learning_rate": 7.172340385927045e-07,
+      "loss": 1.1075,
+      "step": 2115
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7464602225834236,
+      "learning_rate": 7.122174454028452e-07,
+      "loss": 1.0016,
+      "step": 2116
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7765389098094241,
+      "learning_rate": 7.072178098683247e-07,
+      "loss": 1.0661,
+      "step": 2117
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7625101056184171,
+      "learning_rate": 7.022351411174866e-07,
+      "loss": 1.1039,
+      "step": 2118
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7803879580144817,
+      "learning_rate": 6.972694482477005e-07,
+      "loss": 1.1296,
+      "step": 2119
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7861118747364232,
+      "learning_rate": 6.923207403253385e-07,
+      "loss": 1.1381,
+      "step": 2120
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8178225844910723,
+      "learning_rate": 6.873890263857663e-07,
+      "loss": 1.1009,
+      "step": 2121
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8933791901333212,
+      "learning_rate": 6.824743154333157e-07,
+      "loss": 1.1548,
+      "step": 2122
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7821930852368287,
+      "learning_rate": 6.775766164412789e-07,
+      "loss": 1.0226,
+      "step": 2123
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7190745472052495,
+      "learning_rate": 6.726959383518871e-07,
+      "loss": 1.1079,
+      "step": 2124
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.9298389652560309,
+      "learning_rate": 6.678322900762934e-07,
+      "loss": 1.0727,
+      "step": 2125
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7375046288694695,
+      "learning_rate": 6.629856804945579e-07,
+      "loss": 1.1037,
+      "step": 2126
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7741320438703896,
+      "learning_rate": 6.581561184556296e-07,
+      "loss": 1.0596,
+      "step": 2127
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8484004352299219,
+      "learning_rate": 6.533436127773385e-07,
+      "loss": 1.0705,
+      "step": 2128
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6366186224309374,
+      "learning_rate": 6.485481722463671e-07,
+      "loss": 1.0308,
+      "step": 2129
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7366033325204807,
+      "learning_rate": 6.437698056182429e-07,
+      "loss": 1.1873,
+      "step": 2130
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.984251479718554,
+      "learning_rate": 6.390085216173158e-07,
+      "loss": 1.1392,
+      "step": 2131
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7270528089022632,
+      "learning_rate": 6.342643289367523e-07,
+      "loss": 1.1276,
+      "step": 2132
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8519472042958158,
+      "learning_rate": 6.295372362385111e-07,
+      "loss": 1.0334,
+      "step": 2133
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8545869813197795,
+      "learning_rate": 6.248272521533272e-07,
+      "loss": 1.1488,
+      "step": 2134
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7125370364305599,
+      "learning_rate": 6.201343852806996e-07,
+      "loss": 1.0607,
+      "step": 2135
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7944091975081062,
+      "learning_rate": 6.154586441888821e-07,
+      "loss": 1.117,
+      "step": 2136
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6310642990284465,
+      "learning_rate": 6.108000374148448e-07,
+      "loss": 0.9576,
+      "step": 2137
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8125175077195255,
+      "learning_rate": 6.061585734642894e-07,
+      "loss": 1.1718,
+      "step": 2138
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.1297784112884475,
+      "learning_rate": 6.015342608116092e-07,
+      "loss": 1.053,
+      "step": 2139
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7625103860074257,
+      "learning_rate": 5.969271078998884e-07,
+      "loss": 1.1393,
+      "step": 2140
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7571549251472189,
+      "learning_rate": 5.923371231408714e-07,
+      "loss": 1.1734,
+      "step": 2141
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7620543897533195,
+      "learning_rate": 5.877643149149669e-07,
+      "loss": 0.9563,
+      "step": 2142
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.829603251740175,
+      "learning_rate": 5.83208691571221e-07,
+      "loss": 1.1689,
+      "step": 2143
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7159102534347267,
+      "learning_rate": 5.786702614272998e-07,
+      "loss": 1.0901,
+      "step": 2144
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7386164560635057,
+      "learning_rate": 5.741490327694787e-07,
+      "loss": 1.1579,
+      "step": 2145
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7988292932252902,
+      "learning_rate": 5.696450138526277e-07,
+      "loss": 1.0862,
+      "step": 2146
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7948422834386418,
+      "learning_rate": 5.651582129001987e-07,
+      "loss": 1.182,
+      "step": 2147
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7993156324017917,
+      "learning_rate": 5.606886381041976e-07,
+      "loss": 1.0193,
+      "step": 2148
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6925091548204214,
+      "learning_rate": 5.562362976251901e-07,
+      "loss": 1.0618,
+      "step": 2149
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.0579922127029833,
+      "learning_rate": 5.518011995922656e-07,
+      "loss": 1.0962,
+      "step": 2150
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 2.5843088626861572,
+      "eval_runtime": 9.2996,
+      "eval_samples_per_second": 96.779,
+      "eval_steps_per_second": 0.215,
+      "step": 2150
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8008196208275914,
+      "learning_rate": 5.473833521030414e-07,
+      "loss": 1.0104,
+      "step": 2151
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.9095399541359837,
+      "learning_rate": 5.429827632236284e-07,
+      "loss": 1.1351,
+      "step": 2152
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7416945577821182,
+      "learning_rate": 5.385994409886364e-07,
+      "loss": 1.1153,
+      "step": 2153
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7619832086509972,
+      "learning_rate": 5.342333934011435e-07,
+      "loss": 1.09,
+      "step": 2154
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.9032317844021922,
+      "learning_rate": 5.298846284326898e-07,
+      "loss": 1.0996,
+      "step": 2155
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8572228886175396,
+      "learning_rate": 5.255531540232606e-07,
+      "loss": 1.0878,
+      "step": 2156
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7336437707892044,
+      "learning_rate": 5.212389780812733e-07,
+      "loss": 1.0745,
+      "step": 2157
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8531093658557443,
+      "learning_rate": 5.169421084835601e-07,
+      "loss": 1.1218,
+      "step": 2158
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7722900716300092,
+      "learning_rate": 5.12662553075356e-07,
+      "loss": 1.1395,
+      "step": 2159
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.950041716385223,
+      "learning_rate": 5.084003196702802e-07,
+      "loss": 0.9987,
+      "step": 2160
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7423233313967631,
+      "learning_rate": 5.041554160503325e-07,
+      "loss": 1.1495,
+      "step": 2161
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.737283468161627,
+      "learning_rate": 4.999278499658667e-07,
+      "loss": 1.1358,
+      "step": 2162
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7784686579931263,
+      "learning_rate": 4.957176291355815e-07,
+      "loss": 1.1518,
+      "step": 2163
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7644511009678028,
+      "learning_rate": 4.915247612465113e-07,
+      "loss": 1.0186,
+      "step": 2164
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7781775062776076,
+      "learning_rate": 4.873492539540015e-07,
+      "loss": 1.1952,
+      "step": 2165
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.0045644114406365,
+      "learning_rate": 4.831911148817036e-07,
+      "loss": 1.0414,
+      "step": 2166
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.353263927558617,
+      "learning_rate": 4.790503516215572e-07,
+      "loss": 1.0268,
+      "step": 2167
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.4696393048529368,
+      "learning_rate": 4.74926971733779e-07,
+      "loss": 1.0217,
+      "step": 2168
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7224035412681061,
+      "learning_rate": 4.708209827468457e-07,
+      "loss": 1.1709,
+      "step": 2169
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.1974060661453076,
+      "learning_rate": 4.667323921574818e-07,
+      "loss": 1.1509,
+      "step": 2170
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.9519661412660829,
+      "learning_rate": 4.626612074306436e-07,
+      "loss": 1.1691,
+      "step": 2171
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7167506993040328,
+      "learning_rate": 4.5860743599951186e-07,
+      "loss": 1.0569,
+      "step": 2172
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.834646753114657,
+      "learning_rate": 4.5457108526547166e-07,
+      "loss": 1.0966,
+      "step": 2173
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7829827360787306,
+      "learning_rate": 4.505521625981013e-07,
+      "loss": 1.0502,
+      "step": 2174
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7350682219692678,
+      "learning_rate": 4.4655067533515897e-07,
+      "loss": 1.0947,
+      "step": 2175
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7710541505281401,
+      "learning_rate": 4.4256663078257043e-07,
+      "loss": 1.1875,
+      "step": 2176
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7394524408202104,
+      "learning_rate": 4.3860003621441384e-07,
+      "loss": 1.0581,
+      "step": 2177
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.773824443575137,
+      "learning_rate": 4.346508988729059e-07,
+      "loss": 1.0406,
+      "step": 2178
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7352790274190177,
+      "learning_rate": 4.307192259683912e-07,
+      "loss": 1.1197,
+      "step": 2179
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7737667703812778,
+      "learning_rate": 4.268050246793276e-07,
+      "loss": 1.1534,
+      "step": 2180
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.2573154653883563,
+      "learning_rate": 4.2290830215227397e-07,
+      "loss": 1.0383,
+      "step": 2181
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8863791626769575,
+      "learning_rate": 4.190290655018736e-07,
+      "loss": 1.1014,
+      "step": 2182
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.904215990340494,
+      "learning_rate": 4.1516732181084654e-07,
+      "loss": 1.1443,
+      "step": 2183
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6665598174546291,
+      "learning_rate": 4.113230781299748e-07,
+      "loss": 1.1198,
+      "step": 2184
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7865678172139581,
+      "learning_rate": 4.074963414780864e-07,
+      "loss": 1.0621,
+      "step": 2185
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8915787830195252,
+      "learning_rate": 4.0368711884204547e-07,
+      "loss": 1.0839,
+      "step": 2186
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6665536624566356,
+      "learning_rate": 3.998954171767422e-07,
+      "loss": 0.9651,
+      "step": 2187
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7052397530833645,
+      "learning_rate": 3.9612124340507316e-07,
+      "loss": 1.1153,
+      "step": 2188
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8283521703057419,
+      "learning_rate": 3.92364604417933e-07,
+      "loss": 1.1142,
+      "step": 2189
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8237024457016731,
+      "learning_rate": 3.886255070742018e-07,
+      "loss": 1.0637,
+      "step": 2190
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8753034486637011,
+      "learning_rate": 3.849039582007341e-07,
+      "loss": 1.1272,
+      "step": 2191
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8738436864329842,
+      "learning_rate": 3.8119996459234144e-07,
+      "loss": 1.1756,
+      "step": 2192
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8022773574117077,
+      "learning_rate": 3.775135330117852e-07,
+      "loss": 1.0499,
+      "step": 2193
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7144362958585424,
+      "learning_rate": 3.73844670189758e-07,
+      "loss": 1.1467,
+      "step": 2194
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.0135865773904755,
+      "learning_rate": 3.701933828248816e-07,
+      "loss": 1.0784,
+      "step": 2195
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.5344763255393723,
+      "learning_rate": 3.6655967758368327e-07,
+      "loss": 1.0311,
+      "step": 2196
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7838064688680266,
+      "learning_rate": 3.629435611005916e-07,
+      "loss": 1.1035,
+      "step": 2197
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7904435074186833,
+      "learning_rate": 3.593450399779197e-07,
+      "loss": 1.128,
+      "step": 2198
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.773145619676842,
+      "learning_rate": 3.5576412078585755e-07,
+      "loss": 1.0511,
+      "step": 2199
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.9320538758067343,
+      "learning_rate": 3.52200810062453e-07,
+      "loss": 1.1189,
+      "step": 2200
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 2.5895583629608154,
+      "eval_runtime": 11.2481,
+      "eval_samples_per_second": 80.013,
+      "eval_steps_per_second": 0.178,
+      "step": 2200
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.0204479424144324,
+      "learning_rate": 3.4865511431360967e-07,
+      "loss": 1.1475,
+      "step": 2201
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7837453270906646,
+      "learning_rate": 3.451270400130646e-07,
+      "loss": 1.0769,
+      "step": 2202
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.0168529909127566,
+      "learning_rate": 3.4161659360238965e-07,
+      "loss": 1.0727,
+      "step": 2203
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7860021860040863,
+      "learning_rate": 3.381237814909588e-07,
+      "loss": 1.1274,
+      "step": 2204
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7716751588137782,
+      "learning_rate": 3.346486100559632e-07,
+      "loss": 0.9708,
+      "step": 2205
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7351938140354218,
+      "learning_rate": 3.311910856423739e-07,
+      "loss": 1.1615,
+      "step": 2206
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7163076787483474,
+      "learning_rate": 3.2775121456295024e-07,
+      "loss": 1.1286,
+      "step": 2207
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.749271753246435,
+      "learning_rate": 3.2432900309821715e-07,
+      "loss": 1.0363,
+      "step": 2208
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8089246348294565,
+      "learning_rate": 3.209244574964543e-07,
+      "loss": 1.0814,
+      "step": 2209
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7490854620086975,
+      "learning_rate": 3.1753758397369274e-07,
+      "loss": 1.1279,
+      "step": 2210
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7317641823242784,
+      "learning_rate": 3.1416838871368925e-07,
+      "loss": 1.0625,
+      "step": 2211
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.247667747091306,
+      "learning_rate": 3.10816877867931e-07,
+      "loss": 1.0868,
+      "step": 2212
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7633477003875692,
+      "learning_rate": 3.0748305755561425e-07,
+      "loss": 1.1764,
+      "step": 2213
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6639708056383523,
+      "learning_rate": 3.041669338636388e-07,
+      "loss": 0.9255,
+      "step": 2214
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7883244285726891,
+      "learning_rate": 3.00868512846586e-07,
+      "loss": 1.0811,
+      "step": 2215
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.988851647545823,
+      "learning_rate": 2.9758780052672296e-07,
+      "loss": 1.1766,
+      "step": 2216
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8412722477249944,
+      "learning_rate": 2.943248028939838e-07,
+      "loss": 1.0162,
+      "step": 2217
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8642895653823759,
+      "learning_rate": 2.910795259059551e-07,
+      "loss": 1.0872,
+      "step": 2218
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.9606044914844551,
+      "learning_rate": 2.878519754878706e-07,
+      "loss": 1.0954,
+      "step": 2219
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7800030487779746,
+      "learning_rate": 2.8464215753260196e-07,
+      "loss": 1.1096,
+      "step": 2220
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7160785187396002,
+      "learning_rate": 2.814500779006413e-07,
+      "loss": 1.0677,
+      "step": 2221
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.9390278324522734,
+      "learning_rate": 2.7827574242009434e-07,
+      "loss": 1.1598,
+      "step": 2222
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7931416089260662,
+      "learning_rate": 2.751191568866707e-07,
+      "loss": 1.0442,
+      "step": 2223
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7383329178537177,
+      "learning_rate": 2.7198032706367337e-07,
+      "loss": 1.0927,
+      "step": 2224
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8255383059042223,
+      "learning_rate": 2.688592586819827e-07,
+      "loss": 1.1786,
+      "step": 2225
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7448786810389773,
+      "learning_rate": 2.6575595744005476e-07,
+      "loss": 1.0436,
+      "step": 2226
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7439060474380761,
+      "learning_rate": 2.6267042900390173e-07,
+      "loss": 1.1139,
+      "step": 2227
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8181546854419012,
+      "learning_rate": 2.596026790070927e-07,
+      "loss": 1.0915,
+      "step": 2228
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6297740757251267,
+      "learning_rate": 2.565527130507295e-07,
+      "loss": 1.0961,
+      "step": 2229
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7098216586566373,
+      "learning_rate": 2.535205367034477e-07,
+      "loss": 1.1055,
+      "step": 2230
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.9383511386322725,
+      "learning_rate": 2.5050615550140323e-07,
+      "loss": 1.1528,
+      "step": 2231
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.9193231938368944,
+      "learning_rate": 2.4750957494826033e-07,
+      "loss": 1.1275,
+      "step": 2232
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.9515484726057026,
+      "learning_rate": 2.445308005151814e-07,
+      "loss": 1.1016,
+      "step": 2233
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8508052676627187,
+      "learning_rate": 2.4156983764082045e-07,
+      "loss": 0.9611,
+      "step": 2234
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.0767369385971213,
+      "learning_rate": 2.386266917313118e-07,
+      "loss": 1.1194,
+      "step": 2235
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.0123867393740733,
+      "learning_rate": 2.3570136816025714e-07,
+      "loss": 1.127,
+      "step": 2236
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.9614813180997569,
+      "learning_rate": 2.3279387226871842e-07,
+      "loss": 1.0945,
+      "step": 2237
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7686112948166423,
+      "learning_rate": 2.299042093652093e-07,
+      "loss": 1.025,
+      "step": 2238
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7549738410552538,
+      "learning_rate": 2.270323847256839e-07,
+      "loss": 1.1262,
+      "step": 2239
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7559399458480428,
+      "learning_rate": 2.2417840359352572e-07,
+      "loss": 0.9821,
+      "step": 2240
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7468548375744398,
+      "learning_rate": 2.2134227117953988e-07,
+      "loss": 1.0998,
+      "step": 2241
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6995571758565065,
+      "learning_rate": 2.1852399266194312e-07,
+      "loss": 1.1751,
+      "step": 2242
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7943916938445169,
+      "learning_rate": 2.1572357318635717e-07,
+      "loss": 1.1259,
+      "step": 2243
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.9119433935183859,
+      "learning_rate": 2.1294101786579312e-07,
+      "loss": 1.0027,
+      "step": 2244
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8744882007095287,
+      "learning_rate": 2.101763317806471e-07,
+      "loss": 1.2035,
+      "step": 2245
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.0446320367259156,
+      "learning_rate": 2.074295199786891e-07,
+      "loss": 1.0452,
+      "step": 2246
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6596015121410878,
+      "learning_rate": 2.0470058747505516e-07,
+      "loss": 1.1327,
+      "step": 2247
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8490456999699284,
+      "learning_rate": 2.019895392522375e-07,
+      "loss": 1.0833,
+      "step": 2248
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7445494400224035,
+      "learning_rate": 1.9929638026007113e-07,
+      "loss": 1.1463,
+      "step": 2249
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.2591410845990163,
+      "learning_rate": 1.9662111541573713e-07,
+      "loss": 1.0278,
+      "step": 2250
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 2.5933802127838135,
+      "eval_runtime": 9.9441,
+      "eval_samples_per_second": 90.506,
+      "eval_steps_per_second": 0.201,
+      "step": 2250
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6888024904551258,
+      "learning_rate": 1.939637496037361e-07,
+      "loss": 1.102,
+      "step": 2251
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8319674190375471,
+      "learning_rate": 1.9132428767589471e-07,
+      "loss": 1.0311,
+      "step": 2252
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7252144594621507,
+      "learning_rate": 1.8870273445134813e-07,
+      "loss": 1.0805,
+      "step": 2253
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.0435710247965124,
+      "learning_rate": 1.8609909471653643e-07,
+      "loss": 1.083,
+      "step": 2254
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6928435066553217,
+      "learning_rate": 1.8351337322518814e-07,
+      "loss": 1.1283,
+      "step": 2255
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7259652907423643,
+      "learning_rate": 1.809455746983213e-07,
+      "loss": 1.1826,
+      "step": 2256
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8225252399035999,
+      "learning_rate": 1.783957038242279e-07,
+      "loss": 1.0592,
+      "step": 2257
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.903313747412587,
+      "learning_rate": 1.7586376525846936e-07,
+      "loss": 1.046,
+      "step": 2258
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6755112692058999,
+      "learning_rate": 1.7334976362386458e-07,
+      "loss": 1.112,
+      "step": 2259
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.726128924232459,
+      "learning_rate": 1.7085370351048292e-07,
+      "loss": 1.079,
+      "step": 2260
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7836685583219845,
+      "learning_rate": 1.683755894756367e-07,
+      "loss": 1.1456,
+      "step": 2261
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7447831381000527,
+      "learning_rate": 1.6591542604387445e-07,
+      "loss": 1.0467,
+      "step": 2262
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6697257209300452,
+      "learning_rate": 1.6347321770696535e-07,
+      "loss": 1.0265,
+      "step": 2263
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.9319872208126068,
+      "learning_rate": 1.6104896892390142e-07,
+      "loss": 1.1811,
+      "step": 2264
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7504839868317256,
+      "learning_rate": 1.5864268412088102e-07,
+      "loss": 1.0179,
+      "step": 2265
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.192157815385769,
+      "learning_rate": 1.5625436769130642e-07,
+      "loss": 1.0894,
+      "step": 2266
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.708824503806437,
+      "learning_rate": 1.538840239957684e-07,
+      "loss": 1.1525,
+      "step": 2267
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7444811330729261,
+      "learning_rate": 1.5153165736204733e-07,
+      "loss": 1.1408,
+      "step": 2268
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7982842166030193,
+      "learning_rate": 1.4919727208509983e-07,
+      "loss": 0.9771,
+      "step": 2269
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7192420085205755,
+      "learning_rate": 1.4688087242705206e-07,
+      "loss": 1.0764,
+      "step": 2270
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7346814730150235,
+      "learning_rate": 1.4458246261719212e-07,
+      "loss": 0.9906,
+      "step": 2271
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7389789350438183,
+      "learning_rate": 1.4230204685196202e-07,
+      "loss": 1.1434,
+      "step": 2272
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.858666858826461,
+      "learning_rate": 1.400396292949513e-07,
+      "loss": 1.1532,
+      "step": 2273
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.9074430574253788,
+      "learning_rate": 1.3779521407688458e-07,
+      "loss": 1.1789,
+      "step": 2274
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8365799312606907,
+      "learning_rate": 1.3556880529562278e-07,
+      "loss": 0.9848,
+      "step": 2275
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7923391676596275,
+      "learning_rate": 1.3336040701614762e-07,
+      "loss": 1.1937,
+      "step": 2276
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7960703967349725,
+      "learning_rate": 1.3117002327055927e-07,
+      "loss": 1.1113,
+      "step": 2277
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6939004124301441,
+      "learning_rate": 1.2899765805806208e-07,
+      "loss": 1.0892,
+      "step": 2278
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.2521685601316541,
+      "learning_rate": 1.2684331534496886e-07,
+      "loss": 1.1006,
+      "step": 2279
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7881671111909606,
+      "learning_rate": 1.2470699906468097e-07,
+      "loss": 1.0744,
+      "step": 2280
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8160170414531394,
+      "learning_rate": 1.2258871311769282e-07,
+      "loss": 1.0793,
+      "step": 2281
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.9469349206756924,
+      "learning_rate": 1.20488461371574e-07,
+      "loss": 1.1265,
+      "step": 2282
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.73458511204271,
+      "learning_rate": 1.1840624766096931e-07,
+      "loss": 1.0086,
+      "step": 2283
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.891143674387953,
+      "learning_rate": 1.1634207578758994e-07,
+      "loss": 1.1921,
+      "step": 2284
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8370119312599639,
+      "learning_rate": 1.1429594952020451e-07,
+      "loss": 1.0368,
+      "step": 2285
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8363671096784444,
+      "learning_rate": 1.1226787259463579e-07,
+      "loss": 1.1337,
+      "step": 2286
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6980431862227701,
+      "learning_rate": 1.102578487137529e-07,
+      "loss": 1.1189,
+      "step": 2287
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7497156317093847,
+      "learning_rate": 1.082658815474602e-07,
+      "loss": 1.0216,
+      "step": 2288
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8031047798189552,
+      "learning_rate": 1.0629197473269736e-07,
+      "loss": 1.0576,
+      "step": 2289
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6752491430481092,
+      "learning_rate": 1.0433613187342706e-07,
+      "loss": 1.0604,
+      "step": 2290
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8782386262498104,
+      "learning_rate": 1.0239835654063279e-07,
+      "loss": 1.0525,
+      "step": 2291
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7572050606341723,
+      "learning_rate": 1.0047865227230891e-07,
+      "loss": 1.1205,
+      "step": 2292
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8244964463275012,
+      "learning_rate": 9.857702257345503e-08,
+      "loss": 1.1144,
+      "step": 2293
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.181924520431632,
+      "learning_rate": 9.669347091607384e-08,
+      "loss": 0.9956,
+      "step": 2294
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.5816603131170175,
+      "learning_rate": 9.482800073915554e-08,
+      "loss": 1.1195,
+      "step": 2295
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6894077024501315,
+      "learning_rate": 9.29806154486812e-08,
+      "loss": 1.097,
+      "step": 2296
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 6.6836024617463705,
+      "learning_rate": 9.11513184176116e-08,
+      "loss": 1.1011,
+      "step": 2297
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7980576521060946,
+      "learning_rate": 8.934011298588064e-08,
+      "loss": 1.1011,
+      "step": 2298
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7196069818551477,
+      "learning_rate": 8.754700246039194e-08,
+      "loss": 1.1386,
+      "step": 2299
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7871069721387942,
+      "learning_rate": 8.577199011501225e-08,
+      "loss": 0.984,
+      "step": 2300
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 2.589226722717285,
+      "eval_runtime": 9.6361,
+      "eval_samples_per_second": 93.398,
+      "eval_steps_per_second": 0.208,
+      "step": 2300
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7424446938522445,
+      "learning_rate": 8.40150791905603e-08,
+      "loss": 1.1415,
+      "step": 2301
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7578065373272728,
+      "learning_rate": 8.227627289481121e-08,
+      "loss": 1.0798,
+      "step": 2302
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7955256891550265,
+      "learning_rate": 8.055557440247996e-08,
+      "loss": 1.0977,
+      "step": 2303
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.907204072948001,
+      "learning_rate": 7.885298685522235e-08,
+      "loss": 1.0651,
+      "step": 2304
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.0054923837236334,
+      "learning_rate": 7.716851336162623e-08,
+      "loss": 1.128,
+      "step": 2305
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.0693787087412265,
+      "learning_rate": 7.550215699720808e-08,
+      "loss": 1.0772,
+      "step": 2306
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.692393774986647,
+      "learning_rate": 7.385392080440535e-08,
+      "loss": 1.1384,
+      "step": 2307
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8845697790962994,
+      "learning_rate": 7.222380779257077e-08,
+      "loss": 1.0708,
+      "step": 2308
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7388909155888638,
+      "learning_rate": 7.061182093796914e-08,
+      "loss": 1.0943,
+      "step": 2309
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.4531831220434808,
+      "learning_rate": 6.901796318376952e-08,
+      "loss": 1.0679,
+      "step": 2310
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7617954889685343,
+      "learning_rate": 6.744223744004186e-08,
+      "loss": 1.0982,
+      "step": 2311
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6733041671542612,
+      "learning_rate": 6.588464658374816e-08,
+      "loss": 0.978,
+      "step": 2312
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7284072825201411,
+      "learning_rate": 6.434519345874468e-08,
+      "loss": 1.1556,
+      "step": 2313
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6610542517820707,
+      "learning_rate": 6.282388087576529e-08,
+      "loss": 1.0407,
+      "step": 2314
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.0790242283163278,
+      "learning_rate": 6.132071161242813e-08,
+      "loss": 1.186,
+      "step": 2315
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7444989708542987,
+      "learning_rate": 5.98356884132223e-08,
+      "loss": 0.986,
+      "step": 2316
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8227873512668001,
+      "learning_rate": 5.836881398950667e-08,
+      "loss": 1.0403,
+      "step": 2317
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.9724334414596881,
+      "learning_rate": 5.6920091019504464e-08,
+      "loss": 1.0869,
+      "step": 2318
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.9192343583603083,
+      "learning_rate": 5.548952214829762e-08,
+      "loss": 1.1515,
+      "step": 2319
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6864130461717918,
+      "learning_rate": 5.407710998782123e-08,
+      "loss": 1.0811,
+      "step": 2320
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8937430045934018,
+      "learning_rate": 5.2682857116861384e-08,
+      "loss": 1.1126,
+      "step": 2321
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7730057854125026,
+      "learning_rate": 5.1306766081048456e-08,
+      "loss": 1.1067,
+      "step": 2322
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7379495202039894,
+      "learning_rate": 4.994883939285267e-08,
+      "loss": 1.033,
+      "step": 2323
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7886339128259029,
+      "learning_rate": 4.860907953157967e-08,
+      "loss": 1.0412,
+      "step": 2324
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7599180659324617,
+      "learning_rate": 4.72874889433661e-08,
+      "loss": 1.1609,
+      "step": 2325
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7324391338312866,
+      "learning_rate": 4.598407004117511e-08,
+      "loss": 1.0252,
+      "step": 2326
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.768847753069428,
+      "learning_rate": 4.469882520479196e-08,
+      "loss": 1.1656,
+      "step": 2327
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.9818147990029188,
+      "learning_rate": 4.3431756780819564e-08,
+      "loss": 1.1016,
+      "step": 2328
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7845116335974486,
+      "learning_rate": 4.218286708267516e-08,
+      "loss": 1.0294,
+      "step": 2329
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7330342418914214,
+      "learning_rate": 4.095215839058364e-08,
+      "loss": 1.1664,
+      "step": 2330
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4939057739199337,
+      "learning_rate": 3.973963295157646e-08,
+      "loss": 1.0803,
+      "step": 2331
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.059084997020599,
+      "learning_rate": 3.854529297948606e-08,
+      "loss": 1.1342,
+      "step": 2332
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8343652547465445,
+      "learning_rate": 3.736914065493924e-08,
+      "loss": 1.0909,
+      "step": 2333
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.853978724806958,
+      "learning_rate": 3.621117812535935e-08,
+      "loss": 1.1527,
+      "step": 2334
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8044064310008456,
+      "learning_rate": 3.50714075049563e-08,
+      "loss": 1.042,
+      "step": 2335
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.9581602465654037,
+      "learning_rate": 3.39498308747277e-08,
+      "loss": 1.1422,
+      "step": 2336
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.9418981866767318,
+      "learning_rate": 3.284645028244771e-08,
+      "loss": 1.1312,
+      "step": 2337
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7627528381750369,
+      "learning_rate": 3.176126774267374e-08,
+      "loss": 1.014,
+      "step": 2338
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.728705375522156,
+      "learning_rate": 3.069428523673312e-08,
+      "loss": 1.0481,
+      "step": 2339
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8537815587233133,
+      "learning_rate": 2.9645504712726424e-08,
+      "loss": 1.1048,
+      "step": 2340
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.749226991664117,
+      "learning_rate": 2.8614928085517467e-08,
+      "loss": 1.0152,
+      "step": 2341
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8381884610901567,
+      "learning_rate": 2.760255723673888e-08,
+      "loss": 1.1239,
+      "step": 2342
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.9041372160764902,
+      "learning_rate": 2.660839401477655e-08,
+      "loss": 1.0953,
+      "step": 2343
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7345849350394085,
+      "learning_rate": 2.56324402347774e-08,
+      "loss": 1.1378,
+      "step": 2344
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7354100921928703,
+      "learning_rate": 2.46746976786405e-08,
+      "loss": 1.0686,
+      "step": 2345
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7179354134814899,
+      "learning_rate": 2.373516809501375e-08,
+      "loss": 1.0256,
+      "step": 2346
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6992360753520864,
+      "learning_rate": 2.2813853199292745e-08,
+      "loss": 0.9338,
+      "step": 2347
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.762708973779458,
+      "learning_rate": 2.1910754673616363e-08,
+      "loss": 1.1439,
+      "step": 2348
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.9943290195093085,
+      "learning_rate": 2.1025874166864523e-08,
+      "loss": 1.0974,
+      "step": 2349
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6572598677874111,
+      "learning_rate": 2.0159213294652647e-08,
+      "loss": 1.0937,
+      "step": 2350
+    },
+    {
+      "epoch": 0.98,
+      "eval_loss": 2.588456869125366,
+      "eval_runtime": 9.6612,
+      "eval_samples_per_second": 93.156,
+      "eval_steps_per_second": 0.207,
+      "step": 2350
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.309400626673138,
+      "learning_rate": 1.9310773639333868e-08,
+      "loss": 1.1134,
+      "step": 2351
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.1368152536776237,
+      "learning_rate": 1.8480556749991274e-08,
+      "loss": 1.1169,
+      "step": 2352
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.266070157196926,
+      "learning_rate": 1.7668564142434563e-08,
+      "loss": 1.0612,
+      "step": 2353
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7453993652624349,
+      "learning_rate": 1.6874797299204494e-08,
+      "loss": 1.1161,
+      "step": 2354
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.6412311495113356,
+      "learning_rate": 1.6099257669561774e-08,
+      "loss": 1.1021,
+      "step": 2355
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6392266814850405,
+      "learning_rate": 1.5341946669487075e-08,
+      "loss": 1.0745,
+      "step": 2356
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7566882580796954,
+      "learning_rate": 1.4602865681682122e-08,
+      "loss": 1.0886,
+      "step": 2357
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7994588989422393,
+      "learning_rate": 1.3882016055563052e-08,
+      "loss": 1.1288,
+      "step": 2358
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7529519531033513,
+      "learning_rate": 1.3179399107257074e-08,
+      "loss": 1.1285,
+      "step": 2359
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.844460840131692,
+      "learning_rate": 1.2495016119604686e-08,
+      "loss": 0.9629,
+      "step": 2360
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7586134664777406,
+      "learning_rate": 1.1828868342154132e-08,
+      "loss": 1.0932,
+      "step": 2361
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7152904510694706,
+      "learning_rate": 1.1180956991160285e-08,
+      "loss": 1.0872,
+      "step": 2362
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8142339953455432,
+      "learning_rate": 1.0551283249579102e-08,
+      "loss": 1.1247,
+      "step": 2363
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8918235335046223,
+      "learning_rate": 9.93984826707317e-09,
+      "loss": 1.0706,
+      "step": 2364
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7396945789398953,
+      "learning_rate": 9.346653159999497e-09,
+      "loss": 1.1901,
+      "step": 2365
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7583724238271928,
+      "learning_rate": 8.771699011416169e-09,
+      "loss": 1.0064,
+      "step": 2366
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7669609841814268,
+      "learning_rate": 8.214986871076803e-09,
+      "loss": 1.0687,
+      "step": 2367
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7935905547307186,
+      "learning_rate": 7.676517755426105e-09,
+      "loss": 1.137,
+      "step": 2368
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7896500615151392,
+      "learning_rate": 7.156292647602092e-09,
+      "loss": 1.1495,
+      "step": 2369
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.9133147727465644,
+      "learning_rate": 6.654312497434978e-09,
+      "loss": 1.0955,
+      "step": 2370
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7199628654669107,
+      "learning_rate": 6.170578221439405e-09,
+      "loss": 1.148,
+      "step": 2371
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.09388902636575,
+      "learning_rate": 5.705090702819993e-09,
+      "loss": 1.0211,
+      "step": 2372
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7517221891868777,
+      "learning_rate": 5.25785079146357e-09,
+      "loss": 1.1731,
+      "step": 2373
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7757273945846392,
+      "learning_rate": 4.828859303942501e-09,
+      "loss": 1.0793,
+      "step": 2374
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8207681961599385,
+      "learning_rate": 4.4181170235102485e-09,
+      "loss": 1.1323,
+      "step": 2375
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.956915407698254,
+      "learning_rate": 4.025624700101371e-09,
+      "loss": 1.1203,
+      "step": 2376
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7000078828124026,
+      "learning_rate": 3.6513830503293047e-09,
+      "loss": 1.0847,
+      "step": 2377
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.006587094866184,
+      "learning_rate": 3.2953927574852494e-09,
+      "loss": 0.9742,
+      "step": 2378
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7893467658935743,
+      "learning_rate": 2.9576544715370636e-09,
+      "loss": 1.1908,
+      "step": 2379
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8769650035417567,
+      "learning_rate": 2.63816880912815e-09,
+      "loss": 1.1657,
+      "step": 2380
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6866987121112214,
+      "learning_rate": 2.336936353576347e-09,
+      "loss": 1.0213,
+      "step": 2381
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.94374597750828,
+      "learning_rate": 2.053957654871708e-09,
+      "loss": 1.1424,
+      "step": 2382
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7455300388782728,
+      "learning_rate": 1.7892332296776117e-09,
+      "loss": 1.1425,
+      "step": 2383
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8273702370805089,
+      "learning_rate": 1.5427635613296522e-09,
+      "loss": 1.0483,
+      "step": 2384
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8031738913532693,
+      "learning_rate": 1.3145490998323074e-09,
+      "loss": 1.1094,
+      "step": 2385
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.759846012305489,
+      "learning_rate": 1.1045902618600502e-09,
+      "loss": 1.1185,
+      "step": 2386
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.801679239444427,
+      "learning_rate": 9.128874307551273e-10,
+      "loss": 1.1107,
+      "step": 2387
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.8195031748059199,
+      "learning_rate": 7.394409565308902e-10,
+      "loss": 1.073,
+      "step": 2388
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.8035369541515524,
+      "learning_rate": 5.842511558651342e-10,
+      "loss": 1.005,
+      "step": 2389
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.1654835536135948,
+      "learning_rate": 4.4731831210342856e-10,
+      "loss": 1.0705,
+      "step": 2390
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7791913345249419,
+      "learning_rate": 3.286426752580063e-10,
+      "loss": 1.1625,
+      "step": 2391
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.784728420527114,
+      "learning_rate": 2.282244620088747e-10,
+      "loss": 1.0815,
+      "step": 2392
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.8170492279058171,
+      "learning_rate": 1.460638556982641e-10,
+      "loss": 1.0656,
+      "step": 2393
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.8898555829229974,
+      "learning_rate": 8.216100633617884e-11,
+      "loss": 1.1103,
+      "step": 2394
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.754553879445816,
+      "learning_rate": 3.6516030595956654e-11,
+      "loss": 1.1209,
+      "step": 2395
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7492116513237558,
+      "learning_rate": 9.129011815378619e-12,
+      "loss": 1.1737,
+      "step": 2396
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.9740716626454127,
+      "learning_rate": 0.0,
+      "loss": 1.0452,
+      "step": 2397
+    },
+    {
+      "epoch": 1.0,
+      "step": 2397,
+      "total_flos": 0.0,
+      "train_loss": 0.0669286791662599,
+      "train_runtime": 9878.7178,
+      "train_samples_per_second": 498.047,
+      "train_steps_per_second": 0.243
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2397,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50,
+  "total_flos": 0.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}