diff --git "a/checkpoint-630/trainer_state.json" "b/checkpoint-630/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-630/trainer_state.json"
@@ -0,0 +1,4443 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 630,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0031746031746031746,
+      "grad_norm": 5.5625,
+      "learning_rate": 2e-05,
+      "loss": 0.4882,
+      "step": 1
+    },
+    {
+      "epoch": 0.006349206349206349,
+      "grad_norm": 4.84375,
+      "learning_rate": 4e-05,
+      "loss": 0.5092,
+      "step": 2
+    },
+    {
+      "epoch": 0.009523809523809525,
+      "grad_norm": 10.5625,
+      "learning_rate": 6e-05,
+      "loss": 0.584,
+      "step": 3
+    },
+    {
+      "epoch": 0.012698412698412698,
+      "grad_norm": 5.71875,
+      "learning_rate": 8e-05,
+      "loss": 0.4811,
+      "step": 4
+    },
+    {
+      "epoch": 0.015873015873015872,
+      "grad_norm": 8.3125,
+      "learning_rate": 0.0001,
+      "loss": 0.6836,
+      "step": 5
+    },
+    {
+      "epoch": 0.01904761904761905,
+      "grad_norm": 19.5,
+      "learning_rate": 0.00012,
+      "loss": 0.9705,
+      "step": 6
+    },
+    {
+      "epoch": 0.022222222222222223,
+      "grad_norm": 35.5,
+      "learning_rate": 0.00014,
+      "loss": 1.1903,
+      "step": 7
+    },
+    {
+      "epoch": 0.025396825396825397,
+      "grad_norm": 17.875,
+      "learning_rate": 0.00016,
+      "loss": 1.1362,
+      "step": 8
+    },
+    {
+      "epoch": 0.02857142857142857,
+      "grad_norm": 19.875,
+      "learning_rate": 0.00018,
+      "loss": 1.4816,
+      "step": 9
+    },
+    {
+      "epoch": 0.031746031746031744,
+      "grad_norm": 29.125,
+      "learning_rate": 0.0002,
+      "loss": 1.7281,
+      "step": 10
+    },
+    {
+      "epoch": 0.03492063492063492,
+      "grad_norm": 102.0,
+      "learning_rate": 0.00019999968417282543,
+      "loss": 3.067,
+      "step": 11
+    },
+    {
+      "epoch": 0.0380952380952381,
+      "grad_norm": 71.5,
+      "learning_rate": 0.0001999987366932966,
+      "loss": 2.223,
+      "step": 12
+    },
+    {
+      "epoch": 0.04126984126984127,
+      "grad_norm": 11.875,
+      "learning_rate": 0.00019999715756739833,
+      "loss": 1.6384,
+      "step": 13
+    },
+    {
+      "epoch": 0.044444444444444446,
+      "grad_norm": 19.875,
+      "learning_rate": 0.00019999494680510518,
+      "loss": 1.6407,
+      "step": 14
+    },
+    {
+      "epoch": 0.047619047619047616,
+      "grad_norm": 15.25,
+      "learning_rate": 0.00019999210442038162,
+      "loss": 1.5053,
+      "step": 15
+    },
+    {
+      "epoch": 0.050793650793650794,
+      "grad_norm": 77.0,
+      "learning_rate": 0.00019998863043118163,
+      "loss": 1.9586,
+      "step": 16
+    },
+    {
+      "epoch": 0.05396825396825397,
+      "grad_norm": 33.0,
+      "learning_rate": 0.00019998452485944887,
+      "loss": 2.3313,
+      "step": 17
+    },
+    {
+      "epoch": 0.05714285714285714,
+      "grad_norm": 14.6875,
+      "learning_rate": 0.0001999797877311163,
+      "loss": 1.5517,
+      "step": 18
+    },
+    {
+      "epoch": 0.06031746031746032,
+      "grad_norm": 24.375,
+      "learning_rate": 0.00019997441907610624,
+      "loss": 1.7327,
+      "step": 19
+    },
+    {
+      "epoch": 0.06349206349206349,
+      "grad_norm": 11.0625,
+      "learning_rate": 0.00019996841892833,
+      "loss": 1.4351,
+      "step": 20
+    },
+    {
+      "epoch": 0.06666666666666667,
+      "grad_norm": 11.3125,
+      "learning_rate": 0.00019996178732568782,
+      "loss": 1.4974,
+      "step": 21
+    },
+    {
+      "epoch": 0.06984126984126984,
+      "grad_norm": 10.0625,
+      "learning_rate": 0.00019995452431006844,
+      "loss": 1.3888,
+      "step": 22
+    },
+    {
+      "epoch": 0.07301587301587302,
+      "grad_norm": 8.3125,
+      "learning_rate": 0.0001999466299273491,
+      "loss": 1.4097,
+      "step": 23
+    },
+    {
+      "epoch": 0.0761904761904762,
+      "grad_norm": 8.0625,
+      "learning_rate": 0.00019993810422739495,
+      "loss": 1.4168,
+      "step": 24
+    },
+    {
+      "epoch": 0.07936507936507936,
+      "grad_norm": 10.125,
+      "learning_rate": 0.00019992894726405893,
+      "loss": 1.6777,
+      "step": 25
+    },
+    {
+      "epoch": 0.08253968253968254,
+      "grad_norm": 14.25,
+      "learning_rate": 0.00019991915909518143,
+      "loss": 1.6031,
+      "step": 26
+    },
+    {
+      "epoch": 0.08571428571428572,
+      "grad_norm": 7.75,
+      "learning_rate": 0.0001999087397825899,
+      "loss": 1.4771,
+      "step": 27
+    },
+    {
+      "epoch": 0.08888888888888889,
+      "grad_norm": 6.90625,
+      "learning_rate": 0.00019989768939209824,
+      "loss": 1.2748,
+      "step": 28
+    },
+    {
+      "epoch": 0.09206349206349207,
+      "grad_norm": 9.625,
+      "learning_rate": 0.00019988600799350685,
+      "loss": 1.4211,
+      "step": 29
+    },
+    {
+      "epoch": 0.09523809523809523,
+      "grad_norm": 9.4375,
+      "learning_rate": 0.00019987369566060176,
+      "loss": 1.6158,
+      "step": 30
+    },
+    {
+      "epoch": 0.09841269841269841,
+      "grad_norm": 4.75,
+      "learning_rate": 0.00019986075247115432,
+      "loss": 0.9995,
+      "step": 31
+    },
+    {
+      "epoch": 0.10158730158730159,
+      "grad_norm": 5.8125,
+      "learning_rate": 0.0001998471785069208,
+      "loss": 0.9024,
+      "step": 32
+    },
+    {
+      "epoch": 0.10476190476190476,
+      "grad_norm": 7.96875,
+      "learning_rate": 0.00019983297385364166,
+      "loss": 1.6809,
+      "step": 33
+    },
+    {
+      "epoch": 0.10793650793650794,
+      "grad_norm": 8.0625,
+      "learning_rate": 0.00019981813860104128,
+      "loss": 1.2161,
+      "step": 34
+    },
+    {
+      "epoch": 0.1111111111111111,
+      "grad_norm": 6.125,
+      "learning_rate": 0.00019980267284282717,
+      "loss": 1.2645,
+      "step": 35
+    },
+    {
+      "epoch": 0.11428571428571428,
+      "grad_norm": 7.59375,
+      "learning_rate": 0.00019978657667668942,
+      "loss": 1.7419,
+      "step": 36
+    },
+    {
+      "epoch": 0.11746031746031746,
+      "grad_norm": 7.21875,
+      "learning_rate": 0.00019976985020430022,
+      "loss": 1.465,
+      "step": 37
+    },
+    {
+      "epoch": 0.12063492063492064,
+      "grad_norm": 13.1875,
+      "learning_rate": 0.00019975249353131305,
+      "loss": 1.8324,
+      "step": 38
+    },
+    {
+      "epoch": 0.12380952380952381,
+      "grad_norm": 11.0625,
+      "learning_rate": 0.00019973450676736204,
+      "loss": 1.6745,
+      "step": 39
+    },
+    {
+      "epoch": 0.12698412698412698,
+      "grad_norm": 5.03125,
+      "learning_rate": 0.0001997158900260614,
+      "loss": 1.3337,
+      "step": 40
+    },
+    {
+      "epoch": 0.13015873015873017,
+      "grad_norm": 4.75,
+      "learning_rate": 0.0001996966434250046,
+      "loss": 1.151,
+      "step": 41
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "grad_norm": 6.0,
+      "learning_rate": 0.00019967676708576362,
+      "loss": 1.1229,
+      "step": 42
+    },
+    {
+      "epoch": 0.1365079365079365,
+      "grad_norm": 7.90625,
+      "learning_rate": 0.00019965626113388824,
+      "loss": 1.7332,
+      "step": 43
+    },
+    {
+      "epoch": 0.13968253968253969,
+      "grad_norm": 5.1875,
+      "learning_rate": 0.00019963512569890512,
+      "loss": 1.5288,
+      "step": 44
+    },
+    {
+      "epoch": 0.14285714285714285,
+      "grad_norm": 5.09375,
+      "learning_rate": 0.00019961336091431727,
+      "loss": 1.4823,
+      "step": 45
+    },
+    {
+      "epoch": 0.14603174603174604,
+      "grad_norm": 5.75,
+      "learning_rate": 0.00019959096691760282,
+      "loss": 1.7105,
+      "step": 46
+    },
+    {
+      "epoch": 0.1492063492063492,
+      "grad_norm": 5.8125,
+      "learning_rate": 0.00019956794385021442,
+      "loss": 1.4352,
+      "step": 47
+    },
+    {
+      "epoch": 0.1523809523809524,
+      "grad_norm": 10.1875,
+      "learning_rate": 0.00019954429185757832,
+      "loss": 1.5014,
+      "step": 48
+    },
+    {
+      "epoch": 0.15555555555555556,
+      "grad_norm": 8.625,
+      "learning_rate": 0.00019952001108909336,
+      "loss": 1.3925,
+      "step": 49
+    },
+    {
+      "epoch": 0.15873015873015872,
+      "grad_norm": 4.21875,
+      "learning_rate": 0.00019949510169813003,
+      "loss": 1.1968,
+      "step": 50
+    },
+    {
+      "epoch": 0.1619047619047619,
+      "grad_norm": 5.75,
+      "learning_rate": 0.0001994695638420296,
+      "loss": 1.6516,
+      "step": 51
+    },
+    {
+      "epoch": 0.16507936507936508,
+      "grad_norm": 7.625,
+      "learning_rate": 0.00019944339768210306,
+      "loss": 1.5957,
+      "step": 52
+    },
+    {
+      "epoch": 0.16825396825396827,
+      "grad_norm": 5.90625,
+      "learning_rate": 0.0001994166033836301,
+      "loss": 1.347,
+      "step": 53
+    },
+    {
+      "epoch": 0.17142857142857143,
+      "grad_norm": 7.75,
+      "learning_rate": 0.00019938918111585805,
+      "loss": 1.4683,
+      "step": 54
+    },
+    {
+      "epoch": 0.1746031746031746,
+      "grad_norm": 3.84375,
+      "learning_rate": 0.00019936113105200085,
+      "loss": 1.0574,
+      "step": 55
+    },
+    {
+      "epoch": 0.17777777777777778,
+      "grad_norm": 4.5,
+      "learning_rate": 0.000199332453369238,
+      "loss": 1.4662,
+      "step": 56
+    },
+    {
+      "epoch": 0.18095238095238095,
+      "grad_norm": 5.71875,
+      "learning_rate": 0.00019930314824871326,
+      "loss": 1.1714,
+      "step": 57
+    },
+    {
+      "epoch": 0.18412698412698414,
+      "grad_norm": 4.875,
+      "learning_rate": 0.00019927321587553376,
+      "loss": 1.2882,
+      "step": 58
+    },
+    {
+      "epoch": 0.1873015873015873,
+      "grad_norm": 9.0,
+      "learning_rate": 0.0001992426564387686,
+      "loss": 1.31,
+      "step": 59
+    },
+    {
+      "epoch": 0.19047619047619047,
+      "grad_norm": 5.125,
+      "learning_rate": 0.0001992114701314478,
+      "loss": 1.4787,
+      "step": 60
+    },
+    {
+      "epoch": 0.19365079365079366,
+      "grad_norm": 4.53125,
+      "learning_rate": 0.00019917965715056103,
+      "loss": 1.3568,
+      "step": 61
+    },
+    {
+      "epoch": 0.19682539682539682,
+      "grad_norm": 5.03125,
+      "learning_rate": 0.00019914721769705634,
+      "loss": 1.2429,
+      "step": 62
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.71875,
+      "learning_rate": 0.00019911415197583905,
+      "loss": 1.1885,
+      "step": 63
+    },
+    {
+      "epoch": 0.20317460317460317,
+      "grad_norm": 4.75,
+      "learning_rate": 0.00019908046019577008,
+      "loss": 1.465,
+      "step": 64
+    },
+    {
+      "epoch": 0.20634920634920634,
+      "grad_norm": 5.59375,
+      "learning_rate": 0.00019904614256966512,
+      "loss": 1.6841,
+      "step": 65
+    },
+    {
+      "epoch": 0.20952380952380953,
+      "grad_norm": 6.28125,
+      "learning_rate": 0.00019901119931429296,
+      "loss": 1.3299,
+      "step": 66
+    },
+    {
+      "epoch": 0.2126984126984127,
+      "grad_norm": 5.03125,
+      "learning_rate": 0.0001989756306503741,
+      "loss": 1.3073,
+      "step": 67
+    },
+    {
+      "epoch": 0.21587301587301588,
+      "grad_norm": 3.5625,
+      "learning_rate": 0.00019893943680257962,
+      "loss": 1.3444,
+      "step": 68
+    },
+    {
+      "epoch": 0.21904761904761905,
+      "grad_norm": 9.1875,
+      "learning_rate": 0.00019890261799952955,
+      "loss": 1.5232,
+      "step": 69
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 5.03125,
+      "learning_rate": 0.0001988651744737914,
+      "loss": 1.3022,
+      "step": 70
+    },
+    {
+      "epoch": 0.2253968253968254,
+      "grad_norm": 6.8125,
+      "learning_rate": 0.00019882710646187886,
+      "loss": 1.5566,
+      "step": 71
+    },
+    {
+      "epoch": 0.22857142857142856,
+      "grad_norm": 5.0,
+      "learning_rate": 0.00019878841420425023,
+      "loss": 1.1856,
+      "step": 72
+    },
+    {
+      "epoch": 0.23174603174603176,
+      "grad_norm": 5.875,
+      "learning_rate": 0.00019874909794530675,
+      "loss": 1.4443,
+      "step": 73
+    },
+    {
+      "epoch": 0.23492063492063492,
+      "grad_norm": 4.34375,
+      "learning_rate": 0.00019870915793339137,
+      "loss": 1.3829,
+      "step": 74
+    },
+    {
+      "epoch": 0.23809523809523808,
+      "grad_norm": 3.921875,
+      "learning_rate": 0.0001986685944207868,
+      "loss": 1.2855,
+      "step": 75
+    },
+    {
+      "epoch": 0.24126984126984127,
+      "grad_norm": 6.125,
+      "learning_rate": 0.00019862740766371436,
+      "loss": 1.429,
+      "step": 76
+    },
+    {
+      "epoch": 0.24444444444444444,
+      "grad_norm": 4.5,
+      "learning_rate": 0.00019858559792233188,
+      "loss": 1.3491,
+      "step": 77
+    },
+    {
+      "epoch": 0.24761904761904763,
+      "grad_norm": 5.84375,
+      "learning_rate": 0.0001985431654607325,
+      "loss": 1.2587,
+      "step": 78
+    },
+    {
+      "epoch": 0.2507936507936508,
+      "grad_norm": 3.890625,
+      "learning_rate": 0.0001985001105469426,
+      "loss": 0.9128,
+      "step": 79
+    },
+    {
+      "epoch": 0.25396825396825395,
+      "grad_norm": 5.3125,
+      "learning_rate": 0.00019845643345292054,
+      "loss": 1.2216,
+      "step": 80
+    },
+    {
+      "epoch": 0.2571428571428571,
+      "grad_norm": 5.71875,
+      "learning_rate": 0.0001984121344545545,
+      "loss": 1.2427,
+      "step": 81
+    },
+    {
+      "epoch": 0.26031746031746034,
+      "grad_norm": 7.625,
+      "learning_rate": 0.0001983672138316611,
+      "loss": 1.0946,
+      "step": 82
+    },
+    {
+      "epoch": 0.2634920634920635,
+      "grad_norm": 6.75,
+      "learning_rate": 0.00019832167186798331,
+      "loss": 1.7476,
+      "step": 83
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 4.875,
+      "learning_rate": 0.00019827550885118903,
+      "loss": 1.4327,
+      "step": 84
+    },
+    {
+      "epoch": 0.2698412698412698,
+      "grad_norm": 7.5625,
+      "learning_rate": 0.0001982287250728689,
+      "loss": 1.5031,
+      "step": 85
+    },
+    {
+      "epoch": 0.273015873015873,
+      "grad_norm": 6.6875,
+      "learning_rate": 0.00019818132082853466,
+      "loss": 1.7595,
+      "step": 86
+    },
+    {
+      "epoch": 0.2761904761904762,
+      "grad_norm": 4.5625,
+      "learning_rate": 0.0001981332964176174,
+      "loss": 1.3127,
+      "step": 87
+    },
+    {
+      "epoch": 0.27936507936507937,
+      "grad_norm": 4.8125,
+      "learning_rate": 0.00019808465214346524,
+      "loss": 1.0837,
+      "step": 88
+    },
+    {
+      "epoch": 0.28253968253968254,
+      "grad_norm": 3.625,
+      "learning_rate": 0.00019803538831334198,
+      "loss": 1.236,
+      "step": 89
+    },
+    {
+      "epoch": 0.2857142857142857,
+      "grad_norm": 8.5,
+      "learning_rate": 0.0001979855052384247,
+      "loss": 1.4098,
+      "step": 90
+    },
+    {
+      "epoch": 0.28888888888888886,
+      "grad_norm": 6.59375,
+      "learning_rate": 0.000197935003233802,
+      "loss": 1.4328,
+      "step": 91
+    },
+    {
+      "epoch": 0.2920634920634921,
+      "grad_norm": 5.90625,
+      "learning_rate": 0.00019788388261847203,
+      "loss": 1.2988,
+      "step": 92
+    },
+    {
+      "epoch": 0.29523809523809524,
+      "grad_norm": 3.3125,
+      "learning_rate": 0.00019783214371534035,
+      "loss": 1.2953,
+      "step": 93
+    },
+    {
+      "epoch": 0.2984126984126984,
+      "grad_norm": 4.84375,
+      "learning_rate": 0.00019777978685121798,
+      "loss": 1.5484,
+      "step": 94
+    },
+    {
+      "epoch": 0.30158730158730157,
+      "grad_norm": 4.59375,
+      "learning_rate": 0.00019772681235681936,
+      "loss": 1.4477,
+      "step": 95
+    },
+    {
+      "epoch": 0.3047619047619048,
+      "grad_norm": 5.0625,
+      "learning_rate": 0.00019767322056676016,
+      "loss": 1.5552,
+      "step": 96
+    },
+    {
+      "epoch": 0.30793650793650795,
+      "grad_norm": 3.921875,
+      "learning_rate": 0.0001976190118195553,
+      "loss": 1.0938,
+      "step": 97
+    },
+    {
+      "epoch": 0.3111111111111111,
+      "grad_norm": 4.3125,
+      "learning_rate": 0.0001975641864576166,
+      "loss": 1.3172,
+      "step": 98
+    },
+    {
+      "epoch": 0.3142857142857143,
+      "grad_norm": 5.625,
+      "learning_rate": 0.00019750874482725092,
+      "loss": 1.4208,
+      "step": 99
+    },
+    {
+      "epoch": 0.31746031746031744,
+      "grad_norm": 3.6875,
+      "learning_rate": 0.00019745268727865774,
+      "loss": 1.1331,
+      "step": 100
+    },
+    {
+      "epoch": 0.32063492063492066,
+      "grad_norm": 4.65625,
+      "learning_rate": 0.00019739601416592691,
+      "loss": 1.4513,
+      "step": 101
+    },
+    {
+      "epoch": 0.3238095238095238,
+      "grad_norm": 3.890625,
+      "learning_rate": 0.00019733872584703672,
+      "loss": 1.0613,
+      "step": 102
+    },
+    {
+      "epoch": 0.326984126984127,
+      "grad_norm": 6.03125,
+      "learning_rate": 0.00019728082268385125,
+      "loss": 1.2647,
+      "step": 103
+    },
+    {
+      "epoch": 0.33015873015873015,
+      "grad_norm": 4.90625,
+      "learning_rate": 0.00019722230504211843,
+      "loss": 1.2359,
+      "step": 104
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 5.09375,
+      "learning_rate": 0.0001971631732914674,
+      "loss": 1.4131,
+      "step": 105
+    },
+    {
+      "epoch": 0.33650793650793653,
+      "grad_norm": 5.8125,
+      "learning_rate": 0.00019710342780540648,
+      "loss": 1.4764,
+      "step": 106
+    },
+    {
+      "epoch": 0.3396825396825397,
+      "grad_norm": 3.5625,
+      "learning_rate": 0.00019704306896132064,
+      "loss": 1.1251,
+      "step": 107
+    },
+    {
+      "epoch": 0.34285714285714286,
+      "grad_norm": 3.609375,
+      "learning_rate": 0.0001969820971404691,
+      "loss": 1.2523,
+      "step": 108
+    },
+    {
+      "epoch": 0.346031746031746,
+      "grad_norm": 4.25,
+      "learning_rate": 0.00019692051272798304,
+      "loss": 1.31,
+      "step": 109
+    },
+    {
+      "epoch": 0.3492063492063492,
+      "grad_norm": 4.625,
+      "learning_rate": 0.0001968583161128631,
+      "loss": 1.4099,
+      "step": 110
+    },
+    {
+      "epoch": 0.3523809523809524,
+      "grad_norm": 4.125,
+      "learning_rate": 0.00019679550768797693,
+      "loss": 1.1298,
+      "step": 111
+    },
+    {
+      "epoch": 0.35555555555555557,
+      "grad_norm": 3.859375,
+      "learning_rate": 0.0001967320878500566,
+      "loss": 1.0828,
+      "step": 112
+    },
+    {
+      "epoch": 0.35873015873015873,
+      "grad_norm": 5.5,
+      "learning_rate": 0.0001966680569996963,
+      "loss": 1.4486,
+      "step": 113
+    },
+    {
+      "epoch": 0.3619047619047619,
+      "grad_norm": 5.46875,
+      "learning_rate": 0.0001966034155413497,
+      "loss": 1.3412,
+      "step": 114
+    },
+    {
+      "epoch": 0.36507936507936506,
+      "grad_norm": 4.75,
+      "learning_rate": 0.0001965381638833274,
+      "loss": 1.2534,
+      "step": 115
+    },
+    {
+      "epoch": 0.3682539682539683,
+      "grad_norm": 4.1875,
+      "learning_rate": 0.00019647230243779432,
+      "loss": 1.4807,
+      "step": 116
+    },
+    {
+      "epoch": 0.37142857142857144,
+      "grad_norm": 5.78125,
+      "learning_rate": 0.0001964058316207671,
+      "loss": 1.6319,
+      "step": 117
+    },
+    {
+      "epoch": 0.3746031746031746,
+      "grad_norm": 4.65625,
+      "learning_rate": 0.00019633875185211158,
+      "loss": 1.3659,
+      "step": 118
+    },
+    {
+      "epoch": 0.37777777777777777,
+      "grad_norm": 6.0,
+      "learning_rate": 0.00019627106355554007,
+      "loss": 1.2221,
+      "step": 119
+    },
+    {
+      "epoch": 0.38095238095238093,
+      "grad_norm": 5.125,
+      "learning_rate": 0.0001962027671586086,
+      "loss": 1.1326,
+      "step": 120
+    },
+    {
+      "epoch": 0.38412698412698415,
+      "grad_norm": 4.59375,
+      "learning_rate": 0.00019613386309271435,
+      "loss": 1.2941,
+      "step": 121
+    },
+    {
+      "epoch": 0.3873015873015873,
+      "grad_norm": 5.78125,
+      "learning_rate": 0.0001960643517930928,
+      "loss": 1.4116,
+      "step": 122
+    },
+    {
+      "epoch": 0.3904761904761905,
+      "grad_norm": 4.4375,
+      "learning_rate": 0.0001959942336988152,
+      "loss": 1.4073,
+      "step": 123
+    },
+    {
+      "epoch": 0.39365079365079364,
+      "grad_norm": 3.484375,
+      "learning_rate": 0.00019592350925278545,
+      "loss": 1.2554,
+      "step": 124
+    },
+    {
+      "epoch": 0.3968253968253968,
+      "grad_norm": 3.90625,
+      "learning_rate": 0.0001958521789017376,
+      "loss": 1.5309,
+      "step": 125
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.90625,
+      "learning_rate": 0.00019578024309623295,
+      "loss": 1.2464,
+      "step": 126
+    },
+    {
+      "epoch": 0.4031746031746032,
+      "grad_norm": 4.34375,
+      "learning_rate": 0.00019570770229065715,
+      "loss": 1.1707,
+      "step": 127
+    },
+    {
+      "epoch": 0.40634920634920635,
+      "grad_norm": 6.125,
+      "learning_rate": 0.0001956345569432173,
+      "loss": 1.274,
+      "step": 128
+    },
+    {
+      "epoch": 0.4095238095238095,
+      "grad_norm": 4.8125,
+      "learning_rate": 0.00019556080751593918,
+      "loss": 1.28,
+      "step": 129
+    },
+    {
+      "epoch": 0.4126984126984127,
+      "grad_norm": 6.9375,
+      "learning_rate": 0.00019548645447466431,
+      "loss": 1.6019,
+      "step": 130
+    },
+    {
+      "epoch": 0.4158730158730159,
+      "grad_norm": 4.1875,
+      "learning_rate": 0.00019541149828904687,
+      "loss": 1.286,
+      "step": 131
+    },
+    {
+      "epoch": 0.41904761904761906,
+      "grad_norm": 5.53125,
+      "learning_rate": 0.00019533593943255086,
+      "loss": 1.2265,
+      "step": 132
+    },
+    {
+      "epoch": 0.4222222222222222,
+      "grad_norm": 4.4375,
+      "learning_rate": 0.0001952597783824471,
+      "loss": 1.1624,
+      "step": 133
+    },
+    {
+      "epoch": 0.4253968253968254,
+      "grad_norm": 4.125,
+      "learning_rate": 0.00019518301561981014,
+      "loss": 1.5051,
+      "step": 134
+    },
+    {
+      "epoch": 0.42857142857142855,
+      "grad_norm": 2.671875,
+      "learning_rate": 0.00019510565162951537,
+      "loss": 0.7997,
+      "step": 135
+    },
+    {
+      "epoch": 0.43174603174603177,
+      "grad_norm": 4.03125,
+      "learning_rate": 0.00019502768690023574,
+      "loss": 1.3338,
+      "step": 136
+    },
+    {
+      "epoch": 0.43492063492063493,
+      "grad_norm": 7.0,
+      "learning_rate": 0.0001949491219244389,
+      "loss": 1.4556,
+      "step": 137
+    },
+    {
+      "epoch": 0.4380952380952381,
+      "grad_norm": 3.796875,
+      "learning_rate": 0.00019486995719838392,
+      "loss": 1.2246,
+      "step": 138
+    },
+    {
+      "epoch": 0.44126984126984126,
+      "grad_norm": 3.3125,
+      "learning_rate": 0.00019479019322211823,
+      "loss": 1.0053,
+      "step": 139
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 6.25,
+      "learning_rate": 0.00019470983049947444,
+      "loss": 1.3433,
+      "step": 140
+    },
+    {
+      "epoch": 0.44761904761904764,
+      "grad_norm": 8.625,
+      "learning_rate": 0.0001946288695380672,
+      "loss": 1.4911,
+      "step": 141
+    },
+    {
+      "epoch": 0.4507936507936508,
+      "grad_norm": 4.1875,
+      "learning_rate": 0.00019454731084928993,
+      "loss": 1.2431,
+      "step": 142
+    },
+    {
+      "epoch": 0.45396825396825397,
+      "grad_norm": 3.421875,
+      "learning_rate": 0.00019446515494831167,
+      "loss": 1.1637,
+      "step": 143
+    },
+    {
+      "epoch": 0.45714285714285713,
+      "grad_norm": 4.21875,
+      "learning_rate": 0.00019438240235407375,
+      "loss": 1.1885,
+      "step": 144
+    },
+    {
+      "epoch": 0.4603174603174603,
+      "grad_norm": 3.96875,
+      "learning_rate": 0.00019429905358928646,
+      "loss": 1.3931,
+      "step": 145
+    },
+    {
+      "epoch": 0.4634920634920635,
+      "grad_norm": 4.65625,
+      "learning_rate": 0.00019421510918042593,
+      "loss": 1.1286,
+      "step": 146
+    },
+    {
+      "epoch": 0.4666666666666667,
+      "grad_norm": 3.25,
+      "learning_rate": 0.0001941305696577307,
+      "loss": 1.0868,
+      "step": 147
+    },
+    {
+      "epoch": 0.46984126984126984,
+      "grad_norm": 3.140625,
+      "learning_rate": 0.00019404543555519833,
+      "loss": 1.1095,
+      "step": 148
+    },
+    {
+      "epoch": 0.473015873015873,
+      "grad_norm": 4.28125,
+      "learning_rate": 0.00019395970741058202,
+      "loss": 1.316,
+      "step": 149
+    },
+    {
+      "epoch": 0.47619047619047616,
+      "grad_norm": 4.5,
+      "learning_rate": 0.00019387338576538744,
+      "loss": 0.8492,
+      "step": 150
+    },
+    {
+      "epoch": 0.4793650793650794,
+      "grad_norm": 2.875,
+      "learning_rate": 0.0001937864711648689,
+      "loss": 0.906,
+      "step": 151
+    },
+    {
+      "epoch": 0.48253968253968255,
+      "grad_norm": 4.59375,
+      "learning_rate": 0.0001936989641580263,
+      "loss": 1.3716,
+      "step": 152
+    },
+    {
+      "epoch": 0.4857142857142857,
+      "grad_norm": 4.53125,
+      "learning_rate": 0.0001936108652976015,
+      "loss": 1.2317,
+      "step": 153
+    },
+    {
+      "epoch": 0.4888888888888889,
+      "grad_norm": 5.53125,
+      "learning_rate": 0.0001935221751400747,
+      "loss": 1.3703,
+      "step": 154
+    },
+    {
+      "epoch": 0.49206349206349204,
+      "grad_norm": 5.34375,
+      "learning_rate": 0.00019343289424566122,
+      "loss": 1.1271,
+      "step": 155
+    },
+    {
+      "epoch": 0.49523809523809526,
+      "grad_norm": 3.875,
+      "learning_rate": 0.00019334302317830764,
+      "loss": 1.1003,
+      "step": 156
+    },
+    {
+      "epoch": 0.4984126984126984,
+      "grad_norm": 4.46875,
+      "learning_rate": 0.0001932525625056885,
+      "loss": 1.4605,
+      "step": 157
+    },
+    {
+      "epoch": 0.5015873015873016,
+      "grad_norm": 5.0,
+      "learning_rate": 0.0001931615127992026,
+      "loss": 1.4922,
+      "step": 158
+    },
+    {
+      "epoch": 0.5047619047619047,
+      "grad_norm": 3.09375,
+      "learning_rate": 0.00019306987463396934,
+      "loss": 1.1037,
+      "step": 159
+    },
+    {
+      "epoch": 0.5079365079365079,
+      "grad_norm": 3.546875,
+      "learning_rate": 0.00019297764858882514,
+      "loss": 1.1211,
+      "step": 160
+    },
+    {
+      "epoch": 0.5111111111111111,
+      "grad_norm": 5.125,
+      "learning_rate": 0.0001928848352463199,
+      "loss": 1.2307,
+      "step": 161
+    },
+    {
+      "epoch": 0.5142857142857142,
+      "grad_norm": 3.6875,
+      "learning_rate": 0.00019279143519271307,
+      "loss": 1.0421,
+      "step": 162
+    },
+    {
+      "epoch": 0.5174603174603175,
+      "grad_norm": 3.25,
+      "learning_rate": 0.0001926974490179702,
+      "loss": 1.303,
+      "step": 163
+    },
+    {
+      "epoch": 0.5206349206349207,
+      "grad_norm": 3.203125,
+      "learning_rate": 0.00019260287731575902,
+      "loss": 1.1132,
+      "step": 164
+    },
+    {
+      "epoch": 0.5238095238095238,
+      "grad_norm": 3.578125,
+      "learning_rate": 0.0001925077206834458,
+      "loss": 1.144,
+      "step": 165
+    },
+    {
+      "epoch": 0.526984126984127,
+      "grad_norm": 5.15625,
+      "learning_rate": 0.00019241197972209157,
+      "loss": 1.1203,
+      "step": 166
+    },
+    {
+      "epoch": 0.5301587301587302,
+      "grad_norm": 4.3125,
+      "learning_rate": 0.00019231565503644826,
+      "loss": 1.2756,
+      "step": 167
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 4.46875,
+      "learning_rate": 0.00019221874723495495,
+      "loss": 1.2898,
+      "step": 168
+    },
+    {
+      "epoch": 0.5365079365079365,
+      "grad_norm": 4.09375,
+      "learning_rate": 0.00019212125692973396,
+      "loss": 1.3759,
+      "step": 169
+    },
+    {
+      "epoch": 0.5396825396825397,
+      "grad_norm": 3.34375,
+      "learning_rate": 0.00019202318473658705,
+      "loss": 1.113,
+      "step": 170
+    },
+    {
+      "epoch": 0.5428571428571428,
+      "grad_norm": 3.09375,
+      "learning_rate": 0.0001919245312749915,
+      "loss": 0.9661,
+      "step": 171
+    },
+    {
+      "epoch": 0.546031746031746,
+      "grad_norm": 2.921875,
+      "learning_rate": 0.0001918252971680962,
+      "loss": 1.1019,
+      "step": 172
+    },
+    {
+      "epoch": 0.5492063492063493,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.00019172548304271768,
+      "loss": 0.773,
+      "step": 173
+    },
+    {
+      "epoch": 0.5523809523809524,
+      "grad_norm": 3.359375,
+      "learning_rate": 0.0001916250895293362,
+      "loss": 1.2015,
+      "step": 174
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 4.40625,
+      "learning_rate": 0.00019152411726209176,
+      "loss": 1.3768,
+      "step": 175
+    },
+    {
+      "epoch": 0.5587301587301587,
+      "grad_norm": 3.484375,
+      "learning_rate": 0.0001914225668787801,
+      "loss": 0.9422,
+      "step": 176
+    },
+    {
+      "epoch": 0.5619047619047619,
+      "grad_norm": 5.78125,
+      "learning_rate": 0.00019132043902084865,
+      "loss": 1.4228,
+      "step": 177
+    },
+    {
+      "epoch": 0.5650793650793651,
+      "grad_norm": 3.703125,
+      "learning_rate": 0.0001912177343333924,
+      "loss": 1.2504,
+      "step": 178
+    },
+    {
+      "epoch": 0.5682539682539682,
+      "grad_norm": 4.09375,
+      "learning_rate": 0.00019111445346515002,
+      "loss": 1.2305,
+      "step": 179
+    },
+    {
+      "epoch": 0.5714285714285714,
+      "grad_norm": 3.234375,
+      "learning_rate": 0.00019101059706849957,
+      "loss": 1.1119,
+      "step": 180
+    },
+    {
+      "epoch": 0.5746031746031746,
+      "grad_norm": 3.0625,
+      "learning_rate": 0.00019090616579945456,
+      "loss": 1.0028,
+      "step": 181
+    },
+    {
+      "epoch": 0.5777777777777777,
+      "grad_norm": 3.890625,
+      "learning_rate": 0.00019080116031765958,
+      "loss": 1.1326,
+      "step": 182
+    },
+    {
+      "epoch": 0.580952380952381,
+      "grad_norm": 6.25,
+      "learning_rate": 0.00019069558128638635,
+      "loss": 1.582,
+      "step": 183
+    },
+    {
+      "epoch": 0.5841269841269842,
+      "grad_norm": 5.59375,
+      "learning_rate": 0.00019058942937252943,
+      "loss": 1.4948,
+      "step": 184
+    },
+    {
+      "epoch": 0.5873015873015873,
+      "grad_norm": 3.109375,
+      "learning_rate": 0.00019048270524660196,
+      "loss": 0.9625,
+      "step": 185
+    },
+    {
+      "epoch": 0.5904761904761905,
+      "grad_norm": 3.75,
+      "learning_rate": 0.00019037540958273156,
+      "loss": 1.4566,
+      "step": 186
+    },
+    {
+      "epoch": 0.5936507936507937,
+      "grad_norm": 4.28125,
+      "learning_rate": 0.00019026754305865592,
+      "loss": 1.1914,
+      "step": 187
+    },
+    {
+      "epoch": 0.5968253968253968,
+      "grad_norm": 4.15625,
+      "learning_rate": 0.00019015910635571869,
+      "loss": 1.3762,
+      "step": 188
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.875,
+      "learning_rate": 0.00019005010015886496,
+      "loss": 1.3227,
+      "step": 189
+    },
+    {
+      "epoch": 0.6031746031746031,
+      "grad_norm": 2.984375,
+      "learning_rate": 0.0001899405251566371,
+      "loss": 1.0795,
+      "step": 190
+    },
+    {
+      "epoch": 0.6063492063492063,
+      "grad_norm": 6.09375,
+      "learning_rate": 0.00018983038204117044,
+      "loss": 1.5507,
+      "step": 191
+    },
+    {
+      "epoch": 0.6095238095238096,
+      "grad_norm": 4.3125,
+      "learning_rate": 0.00018971967150818877,
+      "loss": 1.2823,
+      "step": 192
+    },
+    {
+      "epoch": 0.6126984126984127,
+      "grad_norm": 5.34375,
+      "learning_rate": 0.0001896083942569999,
+      "loss": 1.5036,
+      "step": 193
+    },
+    {
+      "epoch": 0.6158730158730159,
+      "grad_norm": 3.484375,
+      "learning_rate": 0.00018949655099049148,
+      "loss": 1.08,
+      "step": 194
+    },
+    {
+      "epoch": 0.6190476190476191,
+      "grad_norm": 3.34375,
+      "learning_rate": 0.0001893841424151264,
+      "loss": 1.2002,
+      "step": 195
+    },
+    {
+      "epoch": 0.6222222222222222,
+      "grad_norm": 2.9375,
+      "learning_rate": 0.00018927116924093824,
+      "loss": 1.1845,
+      "step": 196
+    },
+    {
+      "epoch": 0.6253968253968254,
+      "grad_norm": 2.65625,
+      "learning_rate": 0.00018915763218152704,
+      "loss": 0.9349,
+      "step": 197
+    },
+    {
+      "epoch": 0.6285714285714286,
+      "grad_norm": 4.0625,
+      "learning_rate": 0.0001890435319540545,
+      "loss": 1.4206,
+      "step": 198
+    },
+    {
+      "epoch": 0.6317460317460317,
+      "grad_norm": 2.25,
+      "learning_rate": 0.00018892886927923973,
+      "loss": 1.016,
+      "step": 199
+    },
+    {
+      "epoch": 0.6349206349206349,
+      "grad_norm": 4.90625,
+      "learning_rate": 0.00018881364488135448,
+      "loss": 1.3077,
+      "step": 200
+    },
+    {
+      "epoch": 0.638095238095238,
+      "grad_norm": 3.28125,
+      "learning_rate": 0.00018869785948821865,
+      "loss": 1.3698,
+      "step": 201
+    },
+    {
+      "epoch": 0.6412698412698413,
+      "grad_norm": 5.5625,
+      "learning_rate": 0.00018858151383119575,
+      "loss": 1.5526,
+      "step": 202
+    },
+    {
+      "epoch": 0.6444444444444445,
+      "grad_norm": 3.515625,
+      "learning_rate": 0.00018846460864518818,
+      "loss": 1.2174,
+      "step": 203
+    },
+    {
+      "epoch": 0.6476190476190476,
+      "grad_norm": 3.234375,
+      "learning_rate": 0.0001883471446686326,
+      "loss": 1.3522,
+      "step": 204
+    },
+    {
+      "epoch": 0.6507936507936508,
+      "grad_norm": 4.09375,
+      "learning_rate": 0.00018822912264349534,
+      "loss": 1.4896,
+      "step": 205
+    },
+    {
+      "epoch": 0.653968253968254,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.00018811054331526768,
+      "loss": 1.0023,
+      "step": 206
+    },
+    {
+      "epoch": 0.6571428571428571,
+      "grad_norm": 8.375,
+      "learning_rate": 0.00018799140743296105,
+      "loss": 1.495,
+      "step": 207
+    },
+    {
+      "epoch": 0.6603174603174603,
+      "grad_norm": 4.375,
+      "learning_rate": 0.00018787171574910247,
+      "loss": 1.1232,
+      "step": 208
+    },
+    {
+      "epoch": 0.6634920634920635,
+      "grad_norm": 4.09375,
+      "learning_rate": 0.00018775146901972968,
+      "loss": 1.2766,
+      "step": 209
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 2.5,
+      "learning_rate": 0.00018763066800438636,
+      "loss": 1.2216,
+      "step": 210
+    },
+    {
+      "epoch": 0.6698412698412698,
+      "grad_norm": 3.96875,
+      "learning_rate": 0.00018750931346611739,
+      "loss": 1.2713,
+      "step": 211
+    },
+    {
+      "epoch": 0.6730158730158731,
+      "grad_norm": 2.953125,
+      "learning_rate": 0.00018738740617146396,
+      "loss": 1.1207,
+      "step": 212
+    },
+    {
+      "epoch": 0.6761904761904762,
+      "grad_norm": 4.34375,
+      "learning_rate": 0.00018726494689045876,
+      "loss": 1.609,
+      "step": 213
+    },
+    {
+      "epoch": 0.6793650793650794,
+      "grad_norm": 2.53125,
+      "learning_rate": 0.00018714193639662128,
+      "loss": 0.8925,
+      "step": 214
+    },
+    {
+      "epoch": 0.6825396825396826,
+      "grad_norm": 2.984375,
+      "learning_rate": 0.0001870183754669526,
+      "loss": 1.1183,
+      "step": 215
+    },
+    {
+      "epoch": 0.6857142857142857,
+      "grad_norm": 4.46875,
+      "learning_rate": 0.00018689426488193066,
+      "loss": 1.3495,
+      "step": 216
+    },
+    {
+      "epoch": 0.6888888888888889,
+      "grad_norm": 3.421875,
+      "learning_rate": 0.0001867696054255054,
+      "loss": 1.0495,
+      "step": 217
+    },
+    {
+      "epoch": 0.692063492063492,
+      "grad_norm": 7.09375,
+      "learning_rate": 0.0001866443978850937,
+      "loss": 1.5438,
+      "step": 218
+    },
+    {
+      "epoch": 0.6952380952380952,
+      "grad_norm": 4.3125,
+      "learning_rate": 0.0001865186430515745,
+      "loss": 1.3319,
+      "step": 219
+    },
+    {
+      "epoch": 0.6984126984126984,
+      "grad_norm": 3.796875,
+      "learning_rate": 0.00018639234171928353,
+      "loss": 1.1643,
+      "step": 220
+    },
+    {
+      "epoch": 0.7015873015873015,
+      "grad_norm": 2.828125,
+      "learning_rate": 0.00018626549468600878,
+      "loss": 0.9681,
+      "step": 221
+    },
+    {
+      "epoch": 0.7047619047619048,
+      "grad_norm": 3.84375,
+      "learning_rate": 0.000186138102752985,
+      "loss": 1.1597,
+      "step": 222
+    },
+    {
+      "epoch": 0.707936507936508,
+      "grad_norm": 3.234375,
+      "learning_rate": 0.00018601016672488888,
+      "loss": 1.1647,
+      "step": 223
+    },
+    {
+      "epoch": 0.7111111111111111,
+      "grad_norm": 4.25,
+      "learning_rate": 0.00018588168740983388,
+      "loss": 1.15,
+      "step": 224
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 4.5,
+      "learning_rate": 0.00018575266561936523,
+      "loss": 1.2461,
+      "step": 225
+    },
+    {
+      "epoch": 0.7174603174603175,
+      "grad_norm": 3.59375,
+      "learning_rate": 0.00018562310216845465,
+      "loss": 1.108,
+      "step": 226
+    },
+    {
+      "epoch": 0.7206349206349206,
+      "grad_norm": 4.40625,
+      "learning_rate": 0.00018549299787549533,
+      "loss": 1.2701,
+      "step": 227
+    },
+    {
+      "epoch": 0.7238095238095238,
+      "grad_norm": 3.453125,
+      "learning_rate": 0.0001853623535622967,
+      "loss": 1.2034,
+      "step": 228
+    },
+    {
+      "epoch": 0.726984126984127,
+      "grad_norm": 4.0625,
+      "learning_rate": 0.0001852311700540792,
+      "loss": 1.1526,
+      "step": 229
+    },
+    {
+      "epoch": 0.7301587301587301,
+      "grad_norm": 3.296875,
+      "learning_rate": 0.00018509944817946922,
+      "loss": 1.3439,
+      "step": 230
+    },
+    {
+      "epoch": 0.7333333333333333,
+      "grad_norm": 3.734375,
+      "learning_rate": 0.00018496718877049367,
+      "loss": 1.3328,
+      "step": 231
+    },
+    {
+      "epoch": 0.7365079365079366,
+      "grad_norm": 3.515625,
+      "learning_rate": 0.00018483439266257486,
+      "loss": 1.2361,
+      "step": 232
+    },
+    {
+      "epoch": 0.7396825396825397,
+      "grad_norm": 2.875,
+      "learning_rate": 0.0001847010606945252,
+      "loss": 0.9363,
+      "step": 233
+    },
+    {
+      "epoch": 0.7428571428571429,
+      "grad_norm": 5.46875,
+      "learning_rate": 0.0001845671937085419,
+      "loss": 1.2527,
+      "step": 234
+    },
+    {
+      "epoch": 0.746031746031746,
+      "grad_norm": 2.90625,
+      "learning_rate": 0.00018443279255020152,
+      "loss": 1.2294,
+      "step": 235
+    },
+    {
+      "epoch": 0.7492063492063492,
+      "grad_norm": 2.828125,
+      "learning_rate": 0.00018429785806845487,
+      "loss": 1.0708,
+      "step": 236
+    },
+    {
+      "epoch": 0.7523809523809524,
+      "grad_norm": 3.046875,
+      "learning_rate": 0.00018416239111562147,
+      "loss": 1.1407,
+      "step": 237
+    },
+    {
+      "epoch": 0.7555555555555555,
+      "grad_norm": 2.609375,
+      "learning_rate": 0.00018402639254738422,
+      "loss": 0.7489,
+      "step": 238
+    },
+    {
+      "epoch": 0.7587301587301587,
+      "grad_norm": 5.34375,
+      "learning_rate": 0.00018388986322278398,
+      "loss": 1.1833,
+      "step": 239
+    },
+    {
+      "epoch": 0.7619047619047619,
+      "grad_norm": 3.875,
+      "learning_rate": 0.0001837528040042142,
+      "loss": 1.3504,
+      "step": 240
+    },
+    {
+      "epoch": 0.765079365079365,
+      "grad_norm": 3.21875,
+      "learning_rate": 0.0001836152157574153,
+      "loss": 1.5183,
+      "step": 241
+    },
+    {
+      "epoch": 0.7682539682539683,
+      "grad_norm": 2.9375,
+      "learning_rate": 0.00018347709935146958,
+      "loss": 1.0679,
+      "step": 242
+    },
+    {
+      "epoch": 0.7714285714285715,
+      "grad_norm": 3.421875,
+      "learning_rate": 0.00018333845565879516,
+      "loss": 1.2424,
+      "step": 243
+    },
+    {
+      "epoch": 0.7746031746031746,
+      "grad_norm": 5.875,
+      "learning_rate": 0.00018319928555514108,
+      "loss": 1.3007,
+      "step": 244
+    },
+    {
+      "epoch": 0.7777777777777778,
+      "grad_norm": 2.6875,
+      "learning_rate": 0.00018305958991958127,
+      "loss": 1.0213,
+      "step": 245
+    },
+    {
+      "epoch": 0.780952380952381,
+      "grad_norm": 3.375,
+      "learning_rate": 0.00018291936963450934,
+      "loss": 1.3583,
+      "step": 246
+    },
+    {
+      "epoch": 0.7841269841269841,
+      "grad_norm": 3.59375,
+      "learning_rate": 0.00018277862558563278,
+      "loss": 1.1568,
+      "step": 247
+    },
+    {
+      "epoch": 0.7873015873015873,
+      "grad_norm": 6.53125,
+      "learning_rate": 0.00018263735866196756,
+      "loss": 1.3523,
+      "step": 248
+    },
+    {
+      "epoch": 0.7904761904761904,
+      "grad_norm": 3.21875,
+      "learning_rate": 0.0001824955697558323,
+      "loss": 1.1265,
+      "step": 249
+    },
+    {
+      "epoch": 0.7936507936507936,
+      "grad_norm": 4.59375,
+      "learning_rate": 0.00018235325976284275,
+      "loss": 1.2869,
+      "step": 250
+    },
+    {
+      "epoch": 0.7968253968253968,
+      "grad_norm": 3.0,
+      "learning_rate": 0.00018221042958190627,
+      "loss": 1.1097,
+      "step": 251
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.765625,
+      "learning_rate": 0.00018206708011521582,
+      "loss": 1.3495,
+      "step": 252
+    },
+    {
+      "epoch": 0.8031746031746032,
+      "grad_norm": 2.8125,
+      "learning_rate": 0.00018192321226824455,
+      "loss": 1.0061,
+      "step": 253
+    },
+    {
+      "epoch": 0.8063492063492064,
+      "grad_norm": 3.328125,
+      "learning_rate": 0.00018177882694974006,
+      "loss": 1.0327,
+      "step": 254
+    },
+    {
+      "epoch": 0.8095238095238095,
+      "grad_norm": 3.015625,
+      "learning_rate": 0.00018163392507171842,
+      "loss": 0.9958,
+      "step": 255
+    },
+    {
+      "epoch": 0.8126984126984127,
+      "grad_norm": 2.28125,
+      "learning_rate": 0.00018148850754945862,
+      "loss": 1.2192,
+      "step": 256
+    },
+    {
+      "epoch": 0.8158730158730159,
+      "grad_norm": 4.90625,
+      "learning_rate": 0.00018134257530149684,
+      "loss": 1.1918,
+      "step": 257
+    },
+    {
+      "epoch": 0.819047619047619,
+      "grad_norm": 2.78125,
+      "learning_rate": 0.00018119612924962045,
+      "loss": 1.0672,
+      "step": 258
+    },
+    {
+      "epoch": 0.8222222222222222,
+      "grad_norm": 4.40625,
+      "learning_rate": 0.00018104917031886222,
+      "loss": 1.318,
+      "step": 259
+    },
+    {
+      "epoch": 0.8253968253968254,
+      "grad_norm": 3.890625,
+      "learning_rate": 0.00018090169943749476,
+      "loss": 1.2472,
+      "step": 260
+    },
+    {
+      "epoch": 0.8285714285714286,
+      "grad_norm": 2.71875,
+      "learning_rate": 0.0001807537175370242,
+      "loss": 0.7784,
+      "step": 261
+    },
+    {
+      "epoch": 0.8317460317460318,
+      "grad_norm": 2.765625,
+      "learning_rate": 0.0001806052255521847,
+      "loss": 0.9956,
+      "step": 262
+    },
+    {
+      "epoch": 0.834920634920635,
+      "grad_norm": 3.203125,
+      "learning_rate": 0.00018045622442093237,
+      "loss": 1.0368,
+      "step": 263
+    },
+    {
+      "epoch": 0.8380952380952381,
+      "grad_norm": 3.921875,
+      "learning_rate": 0.00018030671508443929,
+      "loss": 1.4863,
+      "step": 264
+    },
+    {
+      "epoch": 0.8412698412698413,
+      "grad_norm": 4.03125,
+      "learning_rate": 0.00018015669848708767,
+      "loss": 1.3369,
+      "step": 265
+    },
+    {
+      "epoch": 0.8444444444444444,
+      "grad_norm": 2.328125,
+      "learning_rate": 0.0001800061755764639,
+      "loss": 1.0004,
+      "step": 266
+    },
+    {
+      "epoch": 0.8476190476190476,
+      "grad_norm": 4.0625,
+      "learning_rate": 0.00017985514730335252,
+      "loss": 1.3371,
+      "step": 267
+    },
+    {
+      "epoch": 0.8507936507936508,
+      "grad_norm": 2.84375,
+      "learning_rate": 0.0001797036146217301,
+      "loss": 1.1897,
+      "step": 268
+    },
+    {
+      "epoch": 0.8539682539682539,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.0001795515784887595,
+      "loss": 0.7096,
+      "step": 269
+    },
+    {
+      "epoch": 0.8571428571428571,
+      "grad_norm": 3.34375,
+      "learning_rate": 0.00017939903986478355,
+      "loss": 1.0215,
+      "step": 270
+    },
+    {
+      "epoch": 0.8603174603174604,
+      "grad_norm": 4.0625,
+      "learning_rate": 0.00017924599971331907,
+      "loss": 1.0234,
+      "step": 271
+    },
+    {
+      "epoch": 0.8634920634920635,
+      "grad_norm": 2.765625,
+      "learning_rate": 0.00017909245900105084,
+      "loss": 1.1715,
+      "step": 272
+    },
+    {
+      "epoch": 0.8666666666666667,
+      "grad_norm": 2.765625,
+      "learning_rate": 0.00017893841869782547,
+      "loss": 1.2362,
+      "step": 273
+    },
+    {
+      "epoch": 0.8698412698412699,
+      "grad_norm": 2.546875,
+      "learning_rate": 0.0001787838797766452,
+      "loss": 1.0232,
+      "step": 274
+    },
+    {
+      "epoch": 0.873015873015873,
+      "grad_norm": 3.40625,
+      "learning_rate": 0.00017862884321366188,
+      "loss": 1.2369,
+      "step": 275
+    },
+    {
+      "epoch": 0.8761904761904762,
+      "grad_norm": 2.65625,
+      "learning_rate": 0.00017847330998817072,
+      "loss": 1.1853,
+      "step": 276
+    },
+    {
+      "epoch": 0.8793650793650793,
+      "grad_norm": 4.25,
+      "learning_rate": 0.00017831728108260407,
+      "loss": 1.0452,
+      "step": 277
+    },
+    {
+      "epoch": 0.8825396825396825,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.00017816075748252526,
+      "loss": 0.9878,
+      "step": 278
+    },
+    {
+      "epoch": 0.8857142857142857,
+      "grad_norm": 3.96875,
+      "learning_rate": 0.0001780037401766225,
+      "loss": 1.4861,
+      "step": 279
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 4.28125,
+      "learning_rate": 0.00017784623015670238,
+      "loss": 1.4139,
+      "step": 280
+    },
+    {
+      "epoch": 0.8920634920634921,
+      "grad_norm": 2.296875,
+      "learning_rate": 0.00017768822841768379,
+      "loss": 1.0084,
+      "step": 281
+    },
+    {
+      "epoch": 0.8952380952380953,
+      "grad_norm": 3.203125,
+      "learning_rate": 0.00017752973595759158,
+      "loss": 1.2464,
+      "step": 282
+    },
+    {
+      "epoch": 0.8984126984126984,
+      "grad_norm": 7.25,
+      "learning_rate": 0.00017737075377755032,
+      "loss": 1.4547,
+      "step": 283
+    },
+    {
+      "epoch": 0.9015873015873016,
+      "grad_norm": 3.46875,
+      "learning_rate": 0.00017721128288177782,
+      "loss": 1.3467,
+      "step": 284
+    },
+    {
+      "epoch": 0.9047619047619048,
+      "grad_norm": 2.375,
+      "learning_rate": 0.00017705132427757895,
+      "loss": 1.0596,
+      "step": 285
+    },
+    {
+      "epoch": 0.9079365079365079,
+      "grad_norm": 3.140625,
+      "learning_rate": 0.00017689087897533915,
+      "loss": 1.3504,
+      "step": 286
+    },
+    {
+      "epoch": 0.9111111111111111,
+      "grad_norm": 2.640625,
+      "learning_rate": 0.0001767299479885182,
+      "loss": 1.206,
+      "step": 287
+    },
+    {
+      "epoch": 0.9142857142857143,
+      "grad_norm": 3.5625,
+      "learning_rate": 0.00017656853233364369,
+      "loss": 1.0524,
+      "step": 288
+    },
+    {
+      "epoch": 0.9174603174603174,
+      "grad_norm": 3.75,
+      "learning_rate": 0.00017640663303030452,
+      "loss": 1.1057,
+      "step": 289
+    },
+    {
+      "epoch": 0.9206349206349206,
+      "grad_norm": 4.96875,
+      "learning_rate": 0.0001762442511011448,
+      "loss": 1.2462,
+      "step": 290
+    },
+    {
+      "epoch": 0.9238095238095239,
+      "grad_norm": 2.515625,
+      "learning_rate": 0.00017608138757185698,
+      "loss": 0.9525,
+      "step": 291
+    },
+    {
+      "epoch": 0.926984126984127,
+      "grad_norm": 3.203125,
+      "learning_rate": 0.00017591804347117567,
+      "loss": 0.9553,
+      "step": 292
+    },
+    {
+      "epoch": 0.9301587301587302,
+      "grad_norm": 4.1875,
+      "learning_rate": 0.00017575421983087095,
+      "loss": 1.2396,
+      "step": 293
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "grad_norm": 4.5625,
+      "learning_rate": 0.000175589917685742,
+      "loss": 1.0318,
+      "step": 294
+    },
+    {
+      "epoch": 0.9365079365079365,
+      "grad_norm": 2.65625,
+      "learning_rate": 0.00017542513807361037,
+      "loss": 1.0792,
+      "step": 295
+    },
+    {
+      "epoch": 0.9396825396825397,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.0001752598820353138,
+      "loss": 0.6749,
+      "step": 296
+    },
+    {
+      "epoch": 0.9428571428571428,
+      "grad_norm": 3.46875,
+      "learning_rate": 0.00017509415061469917,
+      "loss": 1.5,
+      "step": 297
+    },
+    {
+      "epoch": 0.946031746031746,
+      "grad_norm": 2.46875,
+      "learning_rate": 0.00017492794485861617,
+      "loss": 1.0022,
+      "step": 298
+    },
+    {
+      "epoch": 0.9492063492063492,
+      "grad_norm": 2.90625,
+      "learning_rate": 0.0001747612658169107,
+      "loss": 0.9312,
+      "step": 299
+    },
+    {
+      "epoch": 0.9523809523809523,
+      "grad_norm": 4.84375,
+      "learning_rate": 0.00017459411454241822,
+      "loss": 1.1697,
+      "step": 300
+    },
+    {
+      "epoch": 0.9555555555555556,
+      "grad_norm": 3.390625,
+      "learning_rate": 0.000174426492090957,
+      "loss": 1.1664,
+      "step": 301
+    },
+    {
+      "epoch": 0.9587301587301588,
+      "grad_norm": 2.78125,
+      "learning_rate": 0.00017425839952132157,
+      "loss": 1.2322,
+      "step": 302
+    },
+    {
+      "epoch": 0.9619047619047619,
+      "grad_norm": 2.6875,
+      "learning_rate": 0.0001740898378952759,
+      "loss": 1.1428,
+      "step": 303
+    },
+    {
+      "epoch": 0.9650793650793651,
+      "grad_norm": 3.609375,
+      "learning_rate": 0.00017392080827754688,
+      "loss": 1.2497,
+      "step": 304
+    },
+    {
+      "epoch": 0.9682539682539683,
+      "grad_norm": 2.390625,
+      "learning_rate": 0.0001737513117358174,
+      "loss": 0.9671,
+      "step": 305
+    },
+    {
+      "epoch": 0.9714285714285714,
+      "grad_norm": 4.53125,
+      "learning_rate": 0.00017358134934071978,
+      "loss": 1.1203,
+      "step": 306
+    },
+    {
+      "epoch": 0.9746031746031746,
+      "grad_norm": 2.53125,
+      "learning_rate": 0.00017341092216582887,
+      "loss": 1.0647,
+      "step": 307
+    },
+    {
+      "epoch": 0.9777777777777777,
+      "grad_norm": 2.65625,
+      "learning_rate": 0.00017324003128765534,
+      "loss": 1.1387,
+      "step": 308
+    },
+    {
+      "epoch": 0.9809523809523809,
+      "grad_norm": 2.640625,
+      "learning_rate": 0.0001730686777856388,
+      "loss": 1.1033,
+      "step": 309
+    },
+    {
+      "epoch": 0.9841269841269841,
+      "grad_norm": 4.0625,
+      "learning_rate": 0.00017289686274214118,
+      "loss": 1.2105,
+      "step": 310
+    },
+    {
+      "epoch": 0.9873015873015873,
+      "grad_norm": 3.015625,
+      "learning_rate": 0.0001727245872424396,
+      "loss": 0.9702,
+      "step": 311
+    },
+    {
+      "epoch": 0.9904761904761905,
+      "grad_norm": 2.59375,
+      "learning_rate": 0.00017255185237471978,
+      "loss": 0.8773,
+      "step": 312
+    },
+    {
+      "epoch": 0.9936507936507937,
+      "grad_norm": 2.90625,
+      "learning_rate": 0.00017237865923006905,
+      "loss": 0.954,
+      "step": 313
+    },
+    {
+      "epoch": 0.9968253968253968,
+      "grad_norm": 3.375,
+      "learning_rate": 0.00017220500890246943,
+      "loss": 1.3194,
+      "step": 314
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.28125,
+      "learning_rate": 0.0001720309024887907,
+      "loss": 1.4559,
+      "step": 315
+    },
+    {
+      "epoch": 1.0031746031746032,
+      "grad_norm": 1.828125,
+      "learning_rate": 0.00017185634108878367,
+      "loss": 0.6536,
+      "step": 316
+    },
+    {
+      "epoch": 1.0063492063492063,
+      "grad_norm": 4.3125,
+      "learning_rate": 0.000171681325805073,
+      "loss": 0.8967,
+      "step": 317
+    },
+    {
+      "epoch": 1.0095238095238095,
+      "grad_norm": 2.84375,
+      "learning_rate": 0.00017150585774315032,
+      "loss": 0.9844,
+      "step": 318
+    },
+    {
+      "epoch": 1.0126984126984127,
+      "grad_norm": 2.25,
+      "learning_rate": 0.00017132993801136727,
+      "loss": 0.8631,
+      "step": 319
+    },
+    {
+      "epoch": 1.0158730158730158,
+      "grad_norm": 3.5625,
+      "learning_rate": 0.00017115356772092857,
+      "loss": 0.9714,
+      "step": 320
+    },
+    {
+      "epoch": 1.019047619047619,
+      "grad_norm": 3.0,
+      "learning_rate": 0.0001709767479858847,
+      "loss": 0.8505,
+      "step": 321
+    },
+    {
+      "epoch": 1.0222222222222221,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.0001707994799231253,
+      "loss": 0.6619,
+      "step": 322
+    },
+    {
+      "epoch": 1.0253968253968253,
+      "grad_norm": 2.78125,
+      "learning_rate": 0.00017062176465237175,
+      "loss": 0.8414,
+      "step": 323
+    },
+    {
+      "epoch": 1.0285714285714285,
+      "grad_norm": 3.3125,
+      "learning_rate": 0.0001704436032961703,
+      "loss": 0.8343,
+      "step": 324
+    },
+    {
+      "epoch": 1.0317460317460316,
+      "grad_norm": 2.84375,
+      "learning_rate": 0.00017026499697988493,
+      "loss": 1.1863,
+      "step": 325
+    },
+    {
+      "epoch": 1.034920634920635,
+      "grad_norm": 3.125,
+      "learning_rate": 0.00017008594683169017,
+      "loss": 0.9269,
+      "step": 326
+    },
+    {
+      "epoch": 1.0380952380952382,
+      "grad_norm": 1.6953125,
+      "learning_rate": 0.0001699064539825641,
+      "loss": 0.6899,
+      "step": 327
+    },
+    {
+      "epoch": 1.0412698412698413,
+      "grad_norm": 3.0,
+      "learning_rate": 0.00016972651956628107,
+      "loss": 0.7347,
+      "step": 328
+    },
+    {
+      "epoch": 1.0444444444444445,
+      "grad_norm": 1.984375,
+      "learning_rate": 0.00016954614471940472,
+      "loss": 0.8288,
+      "step": 329
+    },
+    {
+      "epoch": 1.0476190476190477,
+      "grad_norm": 2.875,
+      "learning_rate": 0.0001693653305812805,
+      "loss": 0.8383,
+      "step": 330
+    },
+    {
+      "epoch": 1.0507936507936508,
+      "grad_norm": 2.90625,
+      "learning_rate": 0.00016918407829402888,
+      "loss": 0.7976,
+      "step": 331
+    },
+    {
+      "epoch": 1.053968253968254,
+      "grad_norm": 1.875,
+      "learning_rate": 0.00016900238900253775,
+      "loss": 0.679,
+      "step": 332
+    },
+    {
+      "epoch": 1.0571428571428572,
+      "grad_norm": 2.375,
+      "learning_rate": 0.00016882026385445546,
+      "loss": 0.6791,
+      "step": 333
+    },
+    {
+      "epoch": 1.0603174603174603,
+      "grad_norm": 2.28125,
+      "learning_rate": 0.00016863770400018342,
+      "loss": 0.749,
+      "step": 334
+    },
+    {
+      "epoch": 1.0634920634920635,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.00016845471059286887,
+      "loss": 0.5896,
+      "step": 335
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "grad_norm": 2.484375,
+      "learning_rate": 0.00016827128478839767,
+      "loss": 0.8746,
+      "step": 336
+    },
+    {
+      "epoch": 1.0698412698412698,
+      "grad_norm": 2.625,
+      "learning_rate": 0.0001680874277453868,
+      "loss": 0.8057,
+      "step": 337
+    },
+    {
+      "epoch": 1.073015873015873,
+      "grad_norm": 3.46875,
+      "learning_rate": 0.0001679031406251774,
+      "loss": 0.6726,
+      "step": 338
+    },
+    {
+      "epoch": 1.0761904761904761,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.000167718424591827,
+      "loss": 0.6049,
+      "step": 339
+    },
+    {
+      "epoch": 1.0793650793650793,
+      "grad_norm": 3.03125,
+      "learning_rate": 0.00016753328081210245,
+      "loss": 0.902,
+      "step": 340
+    },
+    {
+      "epoch": 1.0825396825396825,
+      "grad_norm": 3.453125,
+      "learning_rate": 0.0001673477104554725,
+      "loss": 1.1903,
+      "step": 341
+    },
+    {
+      "epoch": 1.0857142857142856,
+      "grad_norm": 4.4375,
+      "learning_rate": 0.00016716171469410043,
+      "loss": 0.8901,
+      "step": 342
+    },
+    {
+      "epoch": 1.0888888888888888,
+      "grad_norm": 2.921875,
+      "learning_rate": 0.00016697529470283647,
+      "loss": 0.8613,
+      "step": 343
+    },
+    {
+      "epoch": 1.0920634920634922,
+      "grad_norm": 4.65625,
+      "learning_rate": 0.00016678845165921065,
+      "loss": 1.1702,
+      "step": 344
+    },
+    {
+      "epoch": 1.0952380952380953,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.00016660118674342517,
+      "loss": 0.7437,
+      "step": 345
+    },
+    {
+      "epoch": 1.0984126984126985,
+      "grad_norm": 1.84375,
+      "learning_rate": 0.00016641350113834704,
+      "loss": 0.6215,
+      "step": 346
+    },
+    {
+      "epoch": 1.1015873015873017,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.00016622539602950052,
+      "loss": 0.8003,
+      "step": 347
+    },
+    {
+      "epoch": 1.1047619047619048,
+      "grad_norm": 2.359375,
+      "learning_rate": 0.0001660368726050597,
+      "loss": 0.9519,
+      "step": 348
+    },
+    {
+      "epoch": 1.107936507936508,
+      "grad_norm": 2.421875,
+      "learning_rate": 0.00016584793205584098,
+      "loss": 0.8744,
+      "step": 349
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 2.328125,
+      "learning_rate": 0.00016565857557529566,
+      "loss": 0.6514,
+      "step": 350
+    },
+    {
+      "epoch": 1.1142857142857143,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00016546880435950207,
+      "loss": 0.7064,
+      "step": 351
+    },
+    {
+      "epoch": 1.1174603174603175,
+      "grad_norm": 2.828125,
+      "learning_rate": 0.00016527861960715837,
+      "loss": 0.8101,
+      "step": 352
+    },
+    {
+      "epoch": 1.1206349206349207,
+      "grad_norm": 2.546875,
+      "learning_rate": 0.00016508802251957486,
+      "loss": 0.8317,
+      "step": 353
+    },
+    {
+      "epoch": 1.1238095238095238,
+      "grad_norm": 2.671875,
+      "learning_rate": 0.0001648970143006663,
+      "loss": 0.9023,
+      "step": 354
+    },
+    {
+      "epoch": 1.126984126984127,
+      "grad_norm": 2.28125,
+      "learning_rate": 0.00016470559615694446,
+      "loss": 0.7692,
+      "step": 355
+    },
+    {
+      "epoch": 1.1301587301587301,
+      "grad_norm": 1.9453125,
+      "learning_rate": 0.00016451376929751027,
+      "loss": 0.6903,
+      "step": 356
+    },
+    {
+      "epoch": 1.1333333333333333,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.0001643215349340465,
+      "loss": 0.7706,
+      "step": 357
+    },
+    {
+      "epoch": 1.1365079365079365,
+      "grad_norm": 1.6953125,
+      "learning_rate": 0.0001641288942808099,
+      "loss": 0.7576,
+      "step": 358
+    },
+    {
+      "epoch": 1.1396825396825396,
+      "grad_norm": 2.59375,
+      "learning_rate": 0.0001639358485546235,
+      "loss": 1.0061,
+      "step": 359
+    },
+    {
+      "epoch": 1.1428571428571428,
+      "grad_norm": 3.5625,
+      "learning_rate": 0.000163742398974869,
+      "loss": 0.9229,
+      "step": 360
+    },
+    {
+      "epoch": 1.146031746031746,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.00016354854676347907,
+      "loss": 0.7506,
+      "step": 361
+    },
+    {
+      "epoch": 1.1492063492063491,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.0001633542931449297,
+      "loss": 0.9113,
+      "step": 362
+    },
+    {
+      "epoch": 1.1523809523809523,
+      "grad_norm": 3.265625,
+      "learning_rate": 0.00016315963934623228,
+      "loss": 0.7125,
+      "step": 363
+    },
+    {
+      "epoch": 1.1555555555555554,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.00016296458659692602,
+      "loss": 0.6727,
+      "step": 364
+    },
+    {
+      "epoch": 1.1587301587301586,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00016276913612907007,
+      "loss": 0.8072,
+      "step": 365
+    },
+    {
+      "epoch": 1.161904761904762,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.0001625732891772358,
+      "loss": 0.8393,
+      "step": 366
+    },
+    {
+      "epoch": 1.1650793650793652,
+      "grad_norm": 1.953125,
+      "learning_rate": 0.00016237704697849903,
+      "loss": 0.692,
+      "step": 367
+    },
+    {
+      "epoch": 1.1682539682539683,
+      "grad_norm": 2.875,
+      "learning_rate": 0.00016218041077243212,
+      "loss": 0.9549,
+      "step": 368
+    },
+    {
+      "epoch": 1.1714285714285715,
+      "grad_norm": 2.25,
+      "learning_rate": 0.00016198338180109626,
+      "loss": 0.7114,
+      "step": 369
+    },
+    {
+      "epoch": 1.1746031746031746,
+      "grad_norm": 2.953125,
+      "learning_rate": 0.00016178596130903344,
+      "loss": 0.907,
+      "step": 370
+    },
+    {
+      "epoch": 1.1777777777777778,
+      "grad_norm": 2.3125,
+      "learning_rate": 0.00016158815054325887,
+      "loss": 0.749,
+      "step": 371
+    },
+    {
+      "epoch": 1.180952380952381,
+      "grad_norm": 1.7578125,
+      "learning_rate": 0.00016138995075325277,
+      "loss": 0.6943,
+      "step": 372
+    },
+    {
+      "epoch": 1.1841269841269841,
+      "grad_norm": 2.140625,
+      "learning_rate": 0.0001611913631909528,
+      "loss": 0.9494,
+      "step": 373
+    },
+    {
+      "epoch": 1.1873015873015873,
+      "grad_norm": 2.15625,
+      "learning_rate": 0.00016099238911074592,
+      "loss": 0.8223,
+      "step": 374
+    },
+    {
+      "epoch": 1.1904761904761905,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.00016079302976946055,
+      "loss": 0.7776,
+      "step": 375
+    },
+    {
+      "epoch": 1.1936507936507936,
+      "grad_norm": 2.765625,
+      "learning_rate": 0.00016059328642635862,
+      "loss": 0.8218,
+      "step": 376
+    },
+    {
+      "epoch": 1.1968253968253968,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.00016039316034312769,
+      "loss": 0.61,
+      "step": 377
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.625,
+      "learning_rate": 0.00016019265278387286,
+      "loss": 0.8364,
+      "step": 378
+    },
+    {
+      "epoch": 1.2031746031746031,
+      "grad_norm": 2.328125,
+      "learning_rate": 0.00015999176501510883,
+      "loss": 0.795,
+      "step": 379
+    },
+    {
+      "epoch": 1.2063492063492063,
+      "grad_norm": 1.8828125,
+      "learning_rate": 0.0001597904983057519,
+      "loss": 0.6461,
+      "step": 380
+    },
+    {
+      "epoch": 1.2095238095238094,
+      "grad_norm": 3.46875,
+      "learning_rate": 0.00015958885392711202,
+      "loss": 1.0827,
+      "step": 381
+    },
+    {
+      "epoch": 1.2126984126984126,
+      "grad_norm": 2.484375,
+      "learning_rate": 0.00015938683315288473,
+      "loss": 0.9734,
+      "step": 382
+    },
+    {
+      "epoch": 1.215873015873016,
+      "grad_norm": 2.3125,
+      "learning_rate": 0.00015918443725914297,
+      "loss": 0.8806,
+      "step": 383
+    },
+    {
+      "epoch": 1.2190476190476192,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.0001589816675243292,
+      "loss": 0.8709,
+      "step": 384
+    },
+    {
+      "epoch": 1.2222222222222223,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.00015877852522924732,
+      "loss": 0.6492,
+      "step": 385
+    },
+    {
+      "epoch": 1.2253968253968255,
+      "grad_norm": 2.125,
+      "learning_rate": 0.00015857501165705444,
+      "loss": 0.8271,
+      "step": 386
+    },
+    {
+      "epoch": 1.2285714285714286,
+      "grad_norm": 3.84375,
+      "learning_rate": 0.0001583711280932529,
+      "loss": 0.8211,
+      "step": 387
+    },
+    {
+      "epoch": 1.2317460317460318,
+      "grad_norm": 1.7265625,
+      "learning_rate": 0.00015816687582568208,
+      "loss": 0.5792,
+      "step": 388
+    },
+    {
+      "epoch": 1.234920634920635,
+      "grad_norm": 3.109375,
+      "learning_rate": 0.00015796225614451032,
+      "loss": 1.1246,
+      "step": 389
+    },
+    {
+      "epoch": 1.2380952380952381,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.00015775727034222675,
+      "loss": 0.5635,
+      "step": 390
+    },
+    {
+      "epoch": 1.2412698412698413,
+      "grad_norm": 6.8125,
+      "learning_rate": 0.00015755191971363313,
+      "loss": 0.7416,
+      "step": 391
+    },
+    {
+      "epoch": 1.2444444444444445,
+      "grad_norm": 3.265625,
+      "learning_rate": 0.00015734620555583556,
+      "loss": 0.9866,
+      "step": 392
+    },
+    {
+      "epoch": 1.2476190476190476,
+      "grad_norm": 3.234375,
+      "learning_rate": 0.00015714012916823652,
+      "loss": 0.7828,
+      "step": 393
+    },
+    {
+      "epoch": 1.2507936507936508,
+      "grad_norm": 2.359375,
+      "learning_rate": 0.00015693369185252646,
+      "loss": 0.9174,
+      "step": 394
+    },
+    {
+      "epoch": 1.253968253968254,
+      "grad_norm": 2.46875,
+      "learning_rate": 0.00015672689491267567,
+      "loss": 0.809,
+      "step": 395
+    },
+    {
+      "epoch": 1.2571428571428571,
+      "grad_norm": 1.90625,
+      "learning_rate": 0.000156519739654926,
+      "loss": 0.6508,
+      "step": 396
+    },
+    {
+      "epoch": 1.2603174603174603,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.00015631222738778265,
+      "loss": 0.6718,
+      "step": 397
+    },
+    {
+      "epoch": 1.2634920634920634,
+      "grad_norm": 2.375,
+      "learning_rate": 0.0001561043594220059,
+      "loss": 0.853,
+      "step": 398
+    },
+    {
+      "epoch": 1.2666666666666666,
+      "grad_norm": 3.21875,
+      "learning_rate": 0.00015589613707060277,
+      "loss": 1.0597,
+      "step": 399
+    },
+    {
+      "epoch": 1.2698412698412698,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.00015568756164881882,
+      "loss": 0.7873,
+      "step": 400
+    },
+    {
+      "epoch": 1.273015873015873,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00015547863447412975,
+      "loss": 0.6155,
+      "step": 401
+    },
+    {
+      "epoch": 1.276190476190476,
+      "grad_norm": 1.9296875,
+      "learning_rate": 0.00015526935686623316,
+      "loss": 0.7287,
+      "step": 402
+    },
+    {
+      "epoch": 1.2793650793650793,
+      "grad_norm": 2.484375,
+      "learning_rate": 0.00015505973014704017,
+      "loss": 0.9382,
+      "step": 403
+    },
+    {
+      "epoch": 1.2825396825396824,
+      "grad_norm": 1.84375,
+      "learning_rate": 0.00015484975564066704,
+      "loss": 0.8442,
+      "step": 404
+    },
+    {
+      "epoch": 1.2857142857142856,
+      "grad_norm": 3.421875,
+      "learning_rate": 0.00015463943467342693,
+      "loss": 1.0277,
+      "step": 405
+    },
+    {
+      "epoch": 1.2888888888888888,
+      "grad_norm": 2.78125,
+      "learning_rate": 0.0001544287685738213,
+      "loss": 1.1096,
+      "step": 406
+    },
+    {
+      "epoch": 1.2920634920634921,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.00015421775867253178,
+      "loss": 0.8502,
+      "step": 407
+    },
+    {
+      "epoch": 1.2952380952380953,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.0001540064063024116,
+      "loss": 0.6553,
+      "step": 408
+    },
+    {
+      "epoch": 1.2984126984126985,
+      "grad_norm": 2.796875,
+      "learning_rate": 0.00015379471279847713,
+      "loss": 0.7656,
+      "step": 409
+    },
+    {
+      "epoch": 1.3015873015873016,
+      "grad_norm": 3.09375,
+      "learning_rate": 0.00015358267949789966,
+      "loss": 0.6839,
+      "step": 410
+    },
+    {
+      "epoch": 1.3047619047619048,
+      "grad_norm": 2.5,
+      "learning_rate": 0.00015337030773999674,
+      "loss": 0.9232,
+      "step": 411
+    },
+    {
+      "epoch": 1.307936507936508,
+      "grad_norm": 2.65625,
+      "learning_rate": 0.0001531575988662238,
+      "loss": 1.0738,
+      "step": 412
+    },
+    {
+      "epoch": 1.3111111111111111,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.00015294455422016574,
+      "loss": 0.5921,
+      "step": 413
+    },
+    {
+      "epoch": 1.3142857142857143,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00015273117514752826,
+      "loss": 0.7161,
+      "step": 414
+    },
+    {
+      "epoch": 1.3174603174603174,
+      "grad_norm": 3.234375,
+      "learning_rate": 0.0001525174629961296,
+      "loss": 1.0948,
+      "step": 415
+    },
+    {
+      "epoch": 1.3206349206349206,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00015230341911589182,
+      "loss": 0.48,
+      "step": 416
+    },
+    {
+      "epoch": 1.3238095238095238,
+      "grad_norm": 1.7734375,
+      "learning_rate": 0.00015208904485883244,
+      "loss": 0.7662,
+      "step": 417
+    },
+    {
+      "epoch": 1.326984126984127,
+      "grad_norm": 2.34375,
+      "learning_rate": 0.00015187434157905576,
+      "loss": 0.6982,
+      "step": 418
+    },
+    {
+      "epoch": 1.33015873015873,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00015165931063274441,
+      "loss": 0.5967,
+      "step": 419
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 2.5,
+      "learning_rate": 0.00015144395337815064,
+      "loss": 0.9773,
+      "step": 420
+    },
+    {
+      "epoch": 1.3365079365079366,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.00015122827117558803,
+      "loss": 0.8937,
+      "step": 421
+    },
+    {
+      "epoch": 1.3396825396825398,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.00015101226538742247,
+      "loss": 0.7282,
+      "step": 422
+    },
+    {
+      "epoch": 1.342857142857143,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.00015079593737806399,
+      "loss": 0.7094,
+      "step": 423
+    },
+    {
+      "epoch": 1.3460317460317461,
+      "grad_norm": 2.96875,
+      "learning_rate": 0.00015057928851395792,
+      "loss": 0.9899,
+      "step": 424
+    },
+    {
+      "epoch": 1.3492063492063493,
+      "grad_norm": 2.671875,
+      "learning_rate": 0.0001503623201635761,
+      "loss": 1.0624,
+      "step": 425
+    },
+    {
+      "epoch": 1.3523809523809525,
+      "grad_norm": 2.828125,
+      "learning_rate": 0.00015014503369740865,
+      "loss": 0.9886,
+      "step": 426
+    },
+    {
+      "epoch": 1.3555555555555556,
+      "grad_norm": 2.671875,
+      "learning_rate": 0.0001499274304879549,
+      "loss": 0.8696,
+      "step": 427
+    },
+    {
+      "epoch": 1.3587301587301588,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.00014970951190971512,
+      "loss": 0.8644,
+      "step": 428
+    },
+    {
+      "epoch": 1.361904761904762,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.00014949127933918134,
+      "loss": 0.6432,
+      "step": 429
+    },
+    {
+      "epoch": 1.3650793650793651,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.00014927273415482915,
+      "loss": 0.6992,
+      "step": 430
+    },
+    {
+      "epoch": 1.3682539682539683,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00014905387773710876,
+      "loss": 0.7194,
+      "step": 431
+    },
+    {
+      "epoch": 1.3714285714285714,
+      "grad_norm": 2.53125,
+      "learning_rate": 0.00014883471146843617,
+      "loss": 0.8825,
+      "step": 432
+    },
+    {
+      "epoch": 1.3746031746031746,
+      "grad_norm": 1.9921875,
+      "learning_rate": 0.0001486152367331847,
+      "loss": 0.8407,
+      "step": 433
+    },
+    {
+      "epoch": 1.3777777777777778,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.000148395454917676,
+      "loss": 0.5399,
+      "step": 434
+    },
+    {
+      "epoch": 1.380952380952381,
+      "grad_norm": 1.9375,
+      "learning_rate": 0.00014817536741017152,
+      "loss": 0.7145,
+      "step": 435
+    },
+    {
+      "epoch": 1.384126984126984,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.00014795497560086358,
+      "loss": 0.8827,
+      "step": 436
+    },
+    {
+      "epoch": 1.3873015873015873,
+      "grad_norm": 2.125,
+      "learning_rate": 0.0001477342808818666,
+      "loss": 0.9019,
+      "step": 437
+    },
+    {
+      "epoch": 1.3904761904761904,
+      "grad_norm": 1.6640625,
+      "learning_rate": 0.00014751328464720841,
+      "loss": 0.733,
+      "step": 438
+    },
+    {
+      "epoch": 1.3936507936507936,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.00014729198829282127,
+      "loss": 0.8707,
+      "step": 439
+    },
+    {
+      "epoch": 1.3968253968253967,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0001470703932165333,
+      "loss": 0.4564,
+      "step": 440
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.96875,
+      "learning_rate": 0.00014684850081805935,
+      "loss": 0.7087,
+      "step": 441
+    },
+    {
+      "epoch": 1.403174603174603,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.0001466263124989925,
+      "loss": 0.7329,
+      "step": 442
+    },
+    {
+      "epoch": 1.4063492063492062,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.00014640382966279485,
+      "loss": 0.8311,
+      "step": 443
+    },
+    {
+      "epoch": 1.4095238095238094,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.00014618105371478897,
+      "loss": 0.8004,
+      "step": 444
+    },
+    {
+      "epoch": 1.4126984126984126,
+      "grad_norm": 2.0,
+      "learning_rate": 0.00014595798606214882,
+      "loss": 0.8008,
+      "step": 445
+    },
+    {
+      "epoch": 1.415873015873016,
+      "grad_norm": 1.90625,
+      "learning_rate": 0.00014573462811389088,
+      "loss": 0.7214,
+      "step": 446
+    },
+    {
+      "epoch": 1.4190476190476191,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00014551098128086538,
+      "loss": 0.6169,
+      "step": 447
+    },
+    {
+      "epoch": 1.4222222222222223,
+      "grad_norm": 3.140625,
+      "learning_rate": 0.0001452870469757473,
+      "loss": 1.0013,
+      "step": 448
+    },
+    {
+      "epoch": 1.4253968253968254,
+      "grad_norm": 1.859375,
+      "learning_rate": 0.00014506282661302735,
+      "loss": 0.818,
+      "step": 449
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.00014483832160900326,
+      "loss": 0.6719,
+      "step": 450
+    },
+    {
+      "epoch": 1.4317460317460318,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.0001446135333817706,
+      "loss": 0.5583,
+      "step": 451
+    },
+    {
+      "epoch": 1.434920634920635,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.00014438846335121402,
+      "loss": 0.7985,
+      "step": 452
+    },
+    {
+      "epoch": 1.438095238095238,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.00014416311293899816,
+      "loss": 0.6771,
+      "step": 453
+    },
+    {
+      "epoch": 1.4412698412698413,
+      "grad_norm": 2.53125,
+      "learning_rate": 0.00014393748356855866,
+      "loss": 0.8936,
+      "step": 454
+    },
+    {
+      "epoch": 1.4444444444444444,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.0001437115766650933,
+      "loss": 1.0636,
+      "step": 455
+    },
+    {
+      "epoch": 1.4476190476190476,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00014348539365555282,
+      "loss": 0.7859,
+      "step": 456
+    },
+    {
+      "epoch": 1.4507936507936507,
+      "grad_norm": 1.8828125,
+      "learning_rate": 0.00014325893596863208,
+      "loss": 0.6875,
+      "step": 457
+    },
+    {
+      "epoch": 1.453968253968254,
+      "grad_norm": 1.703125,
+      "learning_rate": 0.0001430322050347609,
+      "loss": 0.7126,
+      "step": 458
+    },
+    {
+      "epoch": 1.457142857142857,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.00014280520228609502,
+      "loss": 0.9058,
+      "step": 459
+    },
+    {
+      "epoch": 1.4603174603174602,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.00014257792915650728,
+      "loss": 0.8914,
+      "step": 460
+    },
+    {
+      "epoch": 1.4634920634920636,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.0001423503870815782,
+      "loss": 0.7137,
+      "step": 461
+    },
+    {
+      "epoch": 1.4666666666666668,
+      "grad_norm": 2.046875,
+      "learning_rate": 0.00014212257749858726,
+      "loss": 0.7943,
+      "step": 462
+    },
+    {
+      "epoch": 1.46984126984127,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.00014189450184650353,
+      "loss": 0.5577,
+      "step": 463
+    },
+    {
+      "epoch": 1.4730158730158731,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.0001416661615659768,
+      "loss": 0.6697,
+      "step": 464
+    },
+    {
+      "epoch": 1.4761904761904763,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.00014143755809932845,
+      "loss": 0.9063,
+      "step": 465
+    },
+    {
+      "epoch": 1.4793650793650794,
+      "grad_norm": 2.28125,
+      "learning_rate": 0.0001412086928905421,
+      "loss": 0.9391,
+      "step": 466
+    },
+    {
+      "epoch": 1.4825396825396826,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.00014097956738525494,
+      "loss": 0.6403,
+      "step": 467
+    },
+    {
+      "epoch": 1.4857142857142858,
+      "grad_norm": 1.9140625,
+      "learning_rate": 0.00014075018303074806,
+      "loss": 0.6846,
+      "step": 468
+    },
+    {
+      "epoch": 1.488888888888889,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.00014052054127593782,
+      "loss": 0.7981,
+      "step": 469
+    },
+    {
+      "epoch": 1.492063492063492,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.00014029064357136628,
+      "loss": 0.7516,
+      "step": 470
+    },
+    {
+      "epoch": 1.4952380952380953,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00014006049136919228,
+      "loss": 0.8163,
+      "step": 471
+    },
+    {
+      "epoch": 1.4984126984126984,
+      "grad_norm": 1.7421875,
+      "learning_rate": 0.00013983008612318228,
+      "loss": 0.7525,
+      "step": 472
+    },
+    {
+      "epoch": 1.5015873015873016,
+      "grad_norm": 2.546875,
+      "learning_rate": 0.000139599429288701,
+      "loss": 0.8414,
+      "step": 473
+    },
+    {
+      "epoch": 1.5047619047619047,
+      "grad_norm": 1.84375,
+      "learning_rate": 0.00013936852232270235,
+      "loss": 0.6444,
+      "step": 474
+    },
+    {
+      "epoch": 1.507936507936508,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00013913736668372026,
+      "loss": 0.667,
+      "step": 475
+    },
+    {
+      "epoch": 1.511111111111111,
+      "grad_norm": 2.203125,
+      "learning_rate": 0.00013890596383185933,
+      "loss": 0.6999,
+      "step": 476
+    },
+    {
+      "epoch": 1.5142857142857142,
+      "grad_norm": 1.9140625,
+      "learning_rate": 0.0001386743152287858,
+      "loss": 0.7879,
+      "step": 477
+    },
+    {
+      "epoch": 1.5174603174603174,
+      "grad_norm": 1.7265625,
+      "learning_rate": 0.0001384424223377181,
+      "loss": 0.5932,
+      "step": 478
+    },
+    {
+      "epoch": 1.5206349206349206,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00013821028662341776,
+      "loss": 0.5998,
+      "step": 479
+    },
+    {
+      "epoch": 1.5238095238095237,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00013797790955218014,
+      "loss": 0.6416,
+      "step": 480
+    },
+    {
+      "epoch": 1.5269841269841269,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.00013774529259182508,
+      "loss": 0.8084,
+      "step": 481
+    },
+    {
+      "epoch": 1.53015873015873,
+      "grad_norm": 3.65625,
+      "learning_rate": 0.0001375124372116878,
+      "loss": 0.9774,
+      "step": 482
+    },
+    {
+      "epoch": 1.5333333333333332,
+      "grad_norm": 1.7890625,
+      "learning_rate": 0.00013727934488260934,
+      "loss": 0.6791,
+      "step": 483
+    },
+    {
+      "epoch": 1.5365079365079364,
+      "grad_norm": 1.7109375,
+      "learning_rate": 0.00013704601707692762,
+      "loss": 0.7585,
+      "step": 484
+    },
+    {
+      "epoch": 1.5396825396825395,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.00013681245526846783,
+      "loss": 0.9826,
+      "step": 485
+    },
+    {
+      "epoch": 1.5428571428571427,
+      "grad_norm": 1.6953125,
+      "learning_rate": 0.00013657866093253328,
+      "loss": 0.6662,
+      "step": 486
+    },
+    {
+      "epoch": 1.5460317460317459,
+      "grad_norm": 1.9140625,
+      "learning_rate": 0.00013634463554589607,
+      "loss": 0.6444,
+      "step": 487
+    },
+    {
+      "epoch": 1.5492063492063493,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00013611038058678775,
+      "loss": 0.5516,
+      "step": 488
+    },
+    {
+      "epoch": 1.5523809523809524,
+      "grad_norm": 2.0625,
+      "learning_rate": 0.00013587589753488998,
+      "loss": 0.9073,
+      "step": 489
+    },
+    {
+      "epoch": 1.5555555555555556,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.00013564118787132506,
+      "loss": 0.7338,
+      "step": 490
+    },
+    {
+      "epoch": 1.5587301587301587,
+      "grad_norm": 1.6484375,
+      "learning_rate": 0.0001354062530786469,
+      "loss": 0.7179,
+      "step": 491
+    },
+    {
+      "epoch": 1.561904761904762,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.0001351710946408313,
+      "loss": 0.7993,
+      "step": 492
+    },
+    {
+      "epoch": 1.565079365079365,
+      "grad_norm": 2.0,
+      "learning_rate": 0.00013493571404326672,
+      "loss": 0.8476,
+      "step": 493
+    },
+    {
+      "epoch": 1.5682539682539682,
+      "grad_norm": 1.859375,
+      "learning_rate": 0.00013470011277274497,
+      "loss": 0.5682,
+      "step": 494
+    },
+    {
+      "epoch": 1.5714285714285714,
+      "grad_norm": 1.9375,
+      "learning_rate": 0.0001344642923174517,
+      "loss": 0.7588,
+      "step": 495
+    },
+    {
+      "epoch": 1.5746031746031746,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.0001342282541669571,
+      "loss": 0.6224,
+      "step": 496
+    },
+    {
+      "epoch": 1.5777777777777777,
+      "grad_norm": 1.703125,
+      "learning_rate": 0.00013399199981220647,
+      "loss": 0.7694,
+      "step": 497
+    },
+    {
+      "epoch": 1.580952380952381,
+      "grad_norm": 1.5,
+      "learning_rate": 0.0001337555307455106,
+      "loss": 0.7373,
+      "step": 498
+    },
+    {
+      "epoch": 1.5841269841269843,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00013351884846053667,
+      "loss": 0.7266,
+      "step": 499
+    },
+    {
+      "epoch": 1.5873015873015874,
+      "grad_norm": 2.140625,
+      "learning_rate": 0.00013328195445229868,
+      "loss": 0.7802,
+      "step": 500
+    },
+    {
+      "epoch": 1.5904761904761906,
+      "grad_norm": 1.7890625,
+      "learning_rate": 0.0001330448502171479,
+      "loss": 0.6952,
+      "step": 501
+    },
+    {
+      "epoch": 1.5936507936507938,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.00013280753725276352,
+      "loss": 0.6439,
+      "step": 502
+    },
+    {
+      "epoch": 1.596825396825397,
+      "grad_norm": 1.625,
+      "learning_rate": 0.00013257001705814321,
+      "loss": 0.7534,
+      "step": 503
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.15625,
+      "learning_rate": 0.00013233229113359367,
+      "loss": 0.7949,
+      "step": 504
+    },
+    {
+      "epoch": 1.6031746031746033,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.00013209436098072095,
+      "loss": 0.7768,
+      "step": 505
+    },
+    {
+      "epoch": 1.6063492063492064,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.00013185622810242128,
+      "loss": 0.8952,
+      "step": 506
+    },
+    {
+      "epoch": 1.6095238095238096,
+      "grad_norm": 1.953125,
+      "learning_rate": 0.0001316178940028713,
+      "loss": 0.8196,
+      "step": 507
+    },
+    {
+      "epoch": 1.6126984126984127,
+      "grad_norm": 1.8671875,
+      "learning_rate": 0.00013137936018751874,
+      "loss": 0.8563,
+      "step": 508
+    },
+    {
+      "epoch": 1.615873015873016,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00013114062816307285,
+      "loss": 0.559,
+      "step": 509
+    },
+    {
+      "epoch": 1.619047619047619,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.00013090169943749476,
+      "loss": 0.5877,
+      "step": 510
+    },
+    {
+      "epoch": 1.6222222222222222,
+      "grad_norm": 1.7734375,
+      "learning_rate": 0.00013066257551998822,
+      "loss": 0.7936,
+      "step": 511
+    },
+    {
+      "epoch": 1.6253968253968254,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.0001304232579209898,
+      "loss": 0.7107,
+      "step": 512
+    },
+    {
+      "epoch": 1.6285714285714286,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.0001301837481521596,
+      "loss": 0.838,
+      "step": 513
+    },
+    {
+      "epoch": 1.6317460317460317,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.00012994404772637147,
+      "loss": 0.6927,
+      "step": 514
+    },
+    {
+      "epoch": 1.6349206349206349,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.0001297041581577035,
+      "loss": 0.7573,
+      "step": 515
+    },
+    {
+      "epoch": 1.638095238095238,
+      "grad_norm": 2.0625,
+      "learning_rate": 0.00012946408096142865,
+      "loss": 0.7616,
+      "step": 516
+    },
+    {
+      "epoch": 1.6412698412698412,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.000129223817654005,
+      "loss": 0.6402,
+      "step": 517
+    },
+    {
+      "epoch": 1.6444444444444444,
+      "grad_norm": 2.75,
+      "learning_rate": 0.0001289833697530661,
+      "loss": 1.0633,
+      "step": 518
+    },
+    {
+      "epoch": 1.6476190476190475,
+      "grad_norm": 1.8046875,
+      "learning_rate": 0.00012874273877741166,
+      "loss": 0.6274,
+      "step": 519
+    },
+    {
+      "epoch": 1.6507936507936507,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.0001285019262469976,
+      "loss": 0.6651,
+      "step": 520
+    },
+    {
+      "epoch": 1.6539682539682539,
+      "grad_norm": 2.46875,
+      "learning_rate": 0.00012826093368292687,
+      "loss": 0.9522,
+      "step": 521
+    },
+    {
+      "epoch": 1.657142857142857,
+      "grad_norm": 2.046875,
+      "learning_rate": 0.00012801976260743938,
+      "loss": 0.7304,
+      "step": 522
+    },
+    {
+      "epoch": 1.6603174603174602,
+      "grad_norm": 2.34375,
+      "learning_rate": 0.00012777841454390275,
+      "loss": 0.854,
+      "step": 523
+    },
+    {
+      "epoch": 1.6634920634920634,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.0001275368910168025,
+      "loss": 0.5274,
+      "step": 524
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 1.984375,
+      "learning_rate": 0.00012729519355173254,
+      "loss": 0.8092,
+      "step": 525
+    },
+    {
+      "epoch": 1.6698412698412697,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.0001270533236753854,
+      "loss": 0.7393,
+      "step": 526
+    },
+    {
+      "epoch": 1.673015873015873,
+      "grad_norm": 6.8125,
+      "learning_rate": 0.00012681128291554264,
+      "loss": 0.9574,
+      "step": 527
+    },
+    {
+      "epoch": 1.6761904761904762,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00012656907280106527,
+      "loss": 0.813,
+      "step": 528
+    },
+    {
+      "epoch": 1.6793650793650794,
+      "grad_norm": 2.65625,
+      "learning_rate": 0.00012632669486188402,
+      "loss": 0.7876,
+      "step": 529
+    },
+    {
+      "epoch": 1.6825396825396826,
+      "grad_norm": 1.84375,
+      "learning_rate": 0.00012608415062898972,
+      "loss": 0.6211,
+      "step": 530
+    },
+    {
+      "epoch": 1.6857142857142857,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.00012584144163442347,
+      "loss": 0.7254,
+      "step": 531
+    },
+    {
+      "epoch": 1.6888888888888889,
+      "grad_norm": 1.65625,
+      "learning_rate": 0.00012559856941126728,
+      "loss": 0.5994,
+      "step": 532
+    },
+    {
+      "epoch": 1.692063492063492,
+      "grad_norm": 1.7265625,
+      "learning_rate": 0.00012535553549363406,
+      "loss": 0.7605,
+      "step": 533
+    },
+    {
+      "epoch": 1.6952380952380952,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.00012511234141665816,
+      "loss": 1.0384,
+      "step": 534
+    },
+    {
+      "epoch": 1.6984126984126984,
+      "grad_norm": 2.546875,
+      "learning_rate": 0.0001248689887164855,
+      "loss": 0.9192,
+      "step": 535
+    },
+    {
+      "epoch": 1.7015873015873015,
+      "grad_norm": 1.8671875,
+      "learning_rate": 0.00012462547893026403,
+      "loss": 0.736,
+      "step": 536
+    },
+    {
+      "epoch": 1.704761904761905,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.00012438181359613386,
+      "loss": 0.6576,
+      "step": 537
+    },
+    {
+      "epoch": 1.707936507936508,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.00012413799425321773,
+      "loss": 0.7015,
+      "step": 538
+    },
+    {
+      "epoch": 1.7111111111111112,
+      "grad_norm": 3.515625,
+      "learning_rate": 0.00012389402244161106,
+      "loss": 0.7308,
+      "step": 539
+    },
+    {
+      "epoch": 1.7142857142857144,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00012364989970237248,
+      "loss": 0.5963,
+      "step": 540
+    },
+    {
+      "epoch": 1.7174603174603176,
+      "grad_norm": 3.25,
+      "learning_rate": 0.00012340562757751384,
+      "loss": 0.8596,
+      "step": 541
+    },
+    {
+      "epoch": 1.7206349206349207,
+      "grad_norm": 2.53125,
+      "learning_rate": 0.00012316120760999066,
+      "loss": 0.7596,
+      "step": 542
+    },
+    {
+      "epoch": 1.723809523809524,
+      "grad_norm": 2.984375,
+      "learning_rate": 0.00012291664134369228,
+      "loss": 0.6814,
+      "step": 543
+    },
+    {
+      "epoch": 1.726984126984127,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.00012267193032343217,
+      "loss": 0.872,
+      "step": 544
+    },
+    {
+      "epoch": 1.7301587301587302,
+      "grad_norm": 2.28125,
+      "learning_rate": 0.00012242707609493814,
+      "loss": 0.9329,
+      "step": 545
+    },
+    {
+      "epoch": 1.7333333333333334,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00012218208020484255,
+      "loss": 0.6596,
+      "step": 546
+    },
+    {
+      "epoch": 1.7365079365079366,
+      "grad_norm": 1.9921875,
+      "learning_rate": 0.00012193694420067261,
+      "loss": 0.6615,
+      "step": 547
+    },
+    {
+      "epoch": 1.7396825396825397,
+      "grad_norm": 2.0625,
+      "learning_rate": 0.00012169166963084055,
+      "loss": 0.8363,
+      "step": 548
+    },
+    {
+      "epoch": 1.7428571428571429,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00012144625804463383,
+      "loss": 0.5201,
+      "step": 549
+    },
+    {
+      "epoch": 1.746031746031746,
+      "grad_norm": 1.703125,
+      "learning_rate": 0.00012120071099220549,
+      "loss": 0.8297,
+      "step": 550
+    },
+    {
+      "epoch": 1.7492063492063492,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.00012095503002456405,
+      "loss": 0.5218,
+      "step": 551
+    },
+    {
+      "epoch": 1.7523809523809524,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00012070921669356415,
+      "loss": 0.5978,
+      "step": 552
+    },
+    {
+      "epoch": 1.7555555555555555,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.00012046327255189627,
+      "loss": 0.6536,
+      "step": 553
+    },
+    {
+      "epoch": 1.7587301587301587,
+      "grad_norm": 1.7578125,
+      "learning_rate": 0.00012021719915307736,
+      "loss": 0.7272,
+      "step": 554
+    },
+    {
+      "epoch": 1.7619047619047619,
+      "grad_norm": 3.28125,
+      "learning_rate": 0.00011997099805144069,
+      "loss": 0.6669,
+      "step": 555
+    },
+    {
+      "epoch": 1.765079365079365,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.00011972467080212631,
+      "loss": 0.5241,
+      "step": 556
+    },
+    {
+      "epoch": 1.7682539682539682,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0001194782189610709,
+      "loss": 0.4608,
+      "step": 557
+    },
+    {
+      "epoch": 1.7714285714285714,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.00011923164408499829,
+      "loss": 0.675,
+      "step": 558
+    },
+    {
+      "epoch": 1.7746031746031745,
+      "grad_norm": 1.7578125,
+      "learning_rate": 0.00011898494773140942,
+      "loss": 0.8407,
+      "step": 559
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00011873813145857249,
+      "loss": 0.6221,
+      "step": 560
+    },
+    {
+      "epoch": 1.7809523809523808,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.00011849119682551322,
+      "loss": 0.6679,
+      "step": 561
+    },
+    {
+      "epoch": 1.784126984126984,
+      "grad_norm": 1.9140625,
+      "learning_rate": 0.00011824414539200504,
+      "loss": 0.7782,
+      "step": 562
+    },
+    {
+      "epoch": 1.7873015873015872,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.000117996978718559,
+      "loss": 0.9411,
+      "step": 563
+    },
+    {
+      "epoch": 1.7904761904761903,
+      "grad_norm": 2.46875,
+      "learning_rate": 0.00011774969836641416,
+      "loss": 0.8041,
+      "step": 564
+    },
+    {
+      "epoch": 1.7936507936507935,
+      "grad_norm": 2.5625,
+      "learning_rate": 0.00011750230589752762,
+      "loss": 0.6918,
+      "step": 565
+    },
+    {
+      "epoch": 1.7968253968253967,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.00011725480287456467,
+      "loss": 0.7215,
+      "step": 566
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00011700719086088891,
+      "loss": 0.504,
+      "step": 567
+    },
+    {
+      "epoch": 1.8031746031746032,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.00011675947142055241,
+      "loss": 0.8362,
+      "step": 568
+    },
+    {
+      "epoch": 1.8063492063492064,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.0001165116461182858,
+      "loss": 0.5821,
+      "step": 569
+    },
+    {
+      "epoch": 1.8095238095238095,
+      "grad_norm": 1.9765625,
+      "learning_rate": 0.00011626371651948838,
+      "loss": 0.8609,
+      "step": 570
+    },
+    {
+      "epoch": 1.8126984126984127,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.0001160156841902182,
+      "loss": 0.6396,
+      "step": 571
+    },
+    {
+      "epoch": 1.8158730158730159,
+      "grad_norm": 1.75,
+      "learning_rate": 0.00011576755069718228,
+      "loss": 0.7322,
+      "step": 572
+    },
+    {
+      "epoch": 1.819047619047619,
+      "grad_norm": 2.484375,
+      "learning_rate": 0.00011551931760772661,
+      "loss": 0.7561,
+      "step": 573
+    },
+    {
+      "epoch": 1.8222222222222222,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00011527098648982632,
+      "loss": 0.7126,
+      "step": 574
+    },
+    {
+      "epoch": 1.8253968253968254,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00011502255891207572,
+      "loss": 0.5396,
+      "step": 575
+    },
+    {
+      "epoch": 1.8285714285714287,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.00011477403644367839,
+      "loss": 0.8063,
+      "step": 576
+    },
+    {
+      "epoch": 1.831746031746032,
+      "grad_norm": 1.90625,
+      "learning_rate": 0.00011452542065443727,
+      "loss": 0.6789,
+      "step": 577
+    },
+    {
+      "epoch": 1.834920634920635,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00011427671311474488,
+      "loss": 0.6365,
+      "step": 578
+    },
+    {
+      "epoch": 1.8380952380952382,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00011402791539557321,
+      "loss": 0.5512,
+      "step": 579
+    },
+    {
+      "epoch": 1.8412698412698414,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.0001137790290684638,
+      "loss": 0.6179,
+      "step": 580
+    },
+    {
+      "epoch": 1.8444444444444446,
+      "grad_norm": 2.84375,
+      "learning_rate": 0.00011353005570551802,
+      "loss": 0.8093,
+      "step": 581
+    },
+    {
+      "epoch": 1.8476190476190477,
+      "grad_norm": 1.7578125,
+      "learning_rate": 0.00011328099687938695,
+      "loss": 0.8066,
+      "step": 582
+    },
+    {
+      "epoch": 1.8507936507936509,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00011303185416326148,
+      "loss": 0.6374,
+      "step": 583
+    },
+    {
+      "epoch": 1.853968253968254,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00011278262913086237,
+      "loss": 0.7085,
+      "step": 584
+    },
+    {
+      "epoch": 1.8571428571428572,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.00011253332335643043,
+      "loss": 0.6773,
+      "step": 585
+    },
+    {
+      "epoch": 1.8603174603174604,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00011228393841471643,
+      "loss": 0.5361,
+      "step": 586
+    },
+    {
+      "epoch": 1.8634920634920635,
+      "grad_norm": 1.7265625,
+      "learning_rate": 0.00011203447588097116,
+      "loss": 0.5985,
+      "step": 587
+    },
+    {
+      "epoch": 1.8666666666666667,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.0001117849373309356,
+      "loss": 0.6972,
+      "step": 588
+    },
+    {
+      "epoch": 1.8698412698412699,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.00011153532434083083,
+      "loss": 0.8232,
+      "step": 589
+    },
+    {
+      "epoch": 1.873015873015873,
+      "grad_norm": 2.0,
+      "learning_rate": 0.00011128563848734816,
+      "loss": 0.5272,
+      "step": 590
+    },
+    {
+      "epoch": 1.8761904761904762,
+      "grad_norm": 1.9140625,
+      "learning_rate": 0.00011103588134763918,
+      "loss": 0.6779,
+      "step": 591
+    },
+    {
+      "epoch": 1.8793650793650793,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00011078605449930568,
+      "loss": 0.6414,
+      "step": 592
+    },
+    {
+      "epoch": 1.8825396825396825,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00011053615952038984,
+      "loss": 0.6568,
+      "step": 593
+    },
+    {
+      "epoch": 1.8857142857142857,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00011028619798936417,
+      "loss": 0.5842,
+      "step": 594
+    },
+    {
+      "epoch": 1.8888888888888888,
+      "grad_norm": 1.7109375,
+      "learning_rate": 0.00011003617148512149,
+      "loss": 1.0061,
+      "step": 595
+    },
+    {
+      "epoch": 1.892063492063492,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.00010978608158696517,
+      "loss": 0.6884,
+      "step": 596
+    },
+    {
+      "epoch": 1.8952380952380952,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.00010953592987459886,
+      "loss": 0.7298,
+      "step": 597
+    },
+    {
+      "epoch": 1.8984126984126983,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.0001092857179281168,
+      "loss": 0.7001,
+      "step": 598
+    },
+    {
+      "epoch": 1.9015873015873015,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.00010903544732799355,
+      "loss": 0.6302,
+      "step": 599
+    },
+    {
+      "epoch": 1.9047619047619047,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00010878511965507434,
+      "loss": 0.5222,
+      "step": 600
+    },
+    {
+      "epoch": 1.9079365079365078,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00010853473649056472,
+      "loss": 0.6731,
+      "step": 601
+    },
+    {
+      "epoch": 1.911111111111111,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.00010828429941602084,
+      "loss": 0.6014,
+      "step": 602
+    },
+    {
+      "epoch": 1.9142857142857141,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00010803381001333943,
+      "loss": 0.5049,
+      "step": 603
+    },
+    {
+      "epoch": 1.9174603174603173,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00010778326986474765,
+      "loss": 0.5385,
+      "step": 604
+    },
+    {
+      "epoch": 1.9206349206349205,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.00010753268055279329,
+      "loss": 0.7527,
+      "step": 605
+    },
+    {
+      "epoch": 1.9238095238095239,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00010728204366033459,
+      "loss": 0.6401,
+      "step": 606
+    },
+    {
+      "epoch": 1.926984126984127,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00010703136077053039,
+      "loss": 0.6815,
+      "step": 607
+    },
+    {
+      "epoch": 1.9301587301587302,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00010678063346683009,
+      "loss": 0.7056,
+      "step": 608
+    },
+    {
+      "epoch": 1.9333333333333333,
+      "grad_norm": 2.796875,
+      "learning_rate": 0.00010652986333296357,
+      "loss": 0.7179,
+      "step": 609
+    },
+    {
+      "epoch": 1.9365079365079365,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00010627905195293135,
+      "loss": 0.6339,
+      "step": 610
+    },
+    {
+      "epoch": 1.9396825396825397,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00010602820091099437,
+      "loss": 0.5683,
+      "step": 611
+    },
+    {
+      "epoch": 1.9428571428571428,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00010577731179166414,
+      "loss": 0.605,
+      "step": 612
+    },
+    {
+      "epoch": 1.946031746031746,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.00010552638617969272,
+      "loss": 0.6433,
+      "step": 613
+    },
+    {
+      "epoch": 1.9492063492063492,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00010527542566006266,
+      "loss": 0.6109,
+      "step": 614
+    },
+    {
+      "epoch": 1.9523809523809523,
+      "grad_norm": 2.234375,
+      "learning_rate": 0.00010502443181797697,
+      "loss": 0.7565,
+      "step": 615
+    },
+    {
+      "epoch": 1.9555555555555557,
+      "grad_norm": 1.9609375,
+      "learning_rate": 0.00010477340623884917,
+      "loss": 0.7454,
+      "step": 616
+    },
+    {
+      "epoch": 1.9587301587301589,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00010452235050829326,
+      "loss": 0.52,
+      "step": 617
+    },
+    {
+      "epoch": 1.961904761904762,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.00010427126621211372,
+      "loss": 0.9651,
+      "step": 618
+    },
+    {
+      "epoch": 1.9650793650793652,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.00010402015493629533,
+      "loss": 0.7549,
+      "step": 619
+    },
+    {
+      "epoch": 1.9682539682539684,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00010376901826699348,
+      "loss": 0.5406,
+      "step": 620
+    },
+    {
+      "epoch": 1.9714285714285715,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00010351785779052378,
+      "loss": 0.5363,
+      "step": 621
+    },
+    {
+      "epoch": 1.9746031746031747,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.00010326667509335239,
+      "loss": 0.6404,
+      "step": 622
+    },
+    {
+      "epoch": 1.9777777777777779,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00010301547176208568,
+      "loss": 0.6006,
+      "step": 623
+    },
+    {
+      "epoch": 1.980952380952381,
+      "grad_norm": 1.8359375,
+      "learning_rate": 0.00010276424938346044,
+      "loss": 0.6102,
+      "step": 624
+    },
+    {
+      "epoch": 1.9841269841269842,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.00010251300954433376,
+      "loss": 0.6636,
+      "step": 625
+    },
+    {
+      "epoch": 1.9873015873015873,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.00010226175383167298,
+      "loss": 0.6489,
+      "step": 626
+    },
+    {
+      "epoch": 1.9904761904761905,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00010201048383254577,
+      "loss": 0.5294,
+      "step": 627
+    },
+    {
+      "epoch": 1.9936507936507937,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00010175920113410998,
+      "loss": 0.5294,
+      "step": 628
+    },
+    {
+      "epoch": 1.9968253968253968,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00010150790732360372,
+      "loss": 0.6688,
+      "step": 629
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.8515625,
+      "learning_rate": 0.00010125660398833528,
+      "loss": 0.8424,
+      "step": 630
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1260,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 315,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.0399754156756173e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}