diff --git "a/checkpoint-1260/trainer_state.json" "b/checkpoint-1260/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1260/trainer_state.json"
@@ -0,0 +1,8853 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 1260,
+  "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
+    },
+    {
+      "epoch": 2.003174603174603,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00010100529271567307,
+      "loss": 0.2694,
+      "step": 631
+    },
+    {
+      "epoch": 2.0063492063492063,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.00010075397509303572,
+      "loss": 0.3088,
+      "step": 632
+    },
+    {
+      "epoch": 2.0095238095238095,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.0001005026527078819,
+      "loss": 0.4431,
+      "step": 633
+    },
+    {
+      "epoch": 2.0126984126984127,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00010025132714770041,
+      "loss": 0.4355,
+      "step": 634
+    },
+    {
+      "epoch": 2.015873015873016,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.0001,
+      "loss": 0.3432,
+      "step": 635
+    },
+    {
+      "epoch": 2.019047619047619,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.974867285229963e-05,
+      "loss": 0.4214,
+      "step": 636
+    },
+    {
+      "epoch": 2.022222222222222,
+      "grad_norm": 1.515625,
+      "learning_rate": 9.94973472921181e-05,
+      "loss": 0.3939,
+      "step": 637
+    },
+    {
+      "epoch": 2.0253968253968253,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.924602490696431e-05,
+      "loss": 0.3744,
+      "step": 638
+    },
+    {
+      "epoch": 2.0285714285714285,
+      "grad_norm": 0.92578125,
+      "learning_rate": 9.899470728432695e-05,
+      "loss": 0.3339,
+      "step": 639
+    },
+    {
+      "epoch": 2.0317460317460316,
+      "grad_norm": 0.97265625,
+      "learning_rate": 9.874339601166473e-05,
+      "loss": 0.384,
+      "step": 640
+    },
+    {
+      "epoch": 2.034920634920635,
+      "grad_norm": 1.34375,
+      "learning_rate": 9.849209267639629e-05,
+      "loss": 0.3892,
+      "step": 641
+    },
+    {
+      "epoch": 2.038095238095238,
+      "grad_norm": 0.71484375,
+      "learning_rate": 9.824079886589003e-05,
+      "loss": 0.2687,
+      "step": 642
+    },
+    {
+      "epoch": 2.041269841269841,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.798951616745428e-05,
+      "loss": 0.4066,
+      "step": 643
+    },
+    {
+      "epoch": 2.0444444444444443,
+      "grad_norm": 1.28125,
+      "learning_rate": 9.773824616832705e-05,
+      "loss": 0.3515,
+      "step": 644
+    },
+    {
+      "epoch": 2.0476190476190474,
+      "grad_norm": 0.95703125,
+      "learning_rate": 9.748699045566626e-05,
+      "loss": 0.2874,
+      "step": 645
+    },
+    {
+      "epoch": 2.0507936507936506,
+      "grad_norm": 0.97265625,
+      "learning_rate": 9.723575061653957e-05,
+      "loss": 0.3316,
+      "step": 646
+    },
+    {
+      "epoch": 2.0539682539682538,
+      "grad_norm": 1.0,
+      "learning_rate": 9.698452823791431e-05,
+      "loss": 0.3451,
+      "step": 647
+    },
+    {
+      "epoch": 2.057142857142857,
+      "grad_norm": 1.234375,
+      "learning_rate": 9.673332490664764e-05,
+      "loss": 0.3872,
+      "step": 648
+    },
+    {
+      "epoch": 2.06031746031746,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.648214220947624e-05,
+      "loss": 0.3857,
+      "step": 649
+    },
+    {
+      "epoch": 2.0634920634920633,
+      "grad_norm": 0.91015625,
+      "learning_rate": 9.623098173300654e-05,
+      "loss": 0.3292,
+      "step": 650
+    },
+    {
+      "epoch": 2.066666666666667,
+      "grad_norm": 0.91796875,
+      "learning_rate": 9.597984506370467e-05,
+      "loss": 0.291,
+      "step": 651
+    },
+    {
+      "epoch": 2.06984126984127,
+      "grad_norm": 1.78125,
+      "learning_rate": 9.572873378788633e-05,
+      "loss": 0.3926,
+      "step": 652
+    },
+    {
+      "epoch": 2.073015873015873,
+      "grad_norm": 1.3984375,
+      "learning_rate": 9.547764949170675e-05,
+      "loss": 0.4275,
+      "step": 653
+    },
+    {
+      "epoch": 2.0761904761904764,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.522659376115086e-05,
+      "loss": 0.3434,
+      "step": 654
+    },
+    {
+      "epoch": 2.0793650793650795,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.497556818202306e-05,
+      "loss": 0.4208,
+      "step": 655
+    },
+    {
+      "epoch": 2.0825396825396827,
+      "grad_norm": 0.96484375,
+      "learning_rate": 9.472457433993735e-05,
+      "loss": 0.3783,
+      "step": 656
+    },
+    {
+      "epoch": 2.085714285714286,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.447361382030731e-05,
+      "loss": 0.4837,
+      "step": 657
+    },
+    {
+      "epoch": 2.088888888888889,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.422268820833588e-05,
+      "loss": 0.3464,
+      "step": 658
+    },
+    {
+      "epoch": 2.092063492063492,
+      "grad_norm": 1.453125,
+      "learning_rate": 9.397179908900566e-05,
+      "loss": 0.5424,
+      "step": 659
+    },
+    {
+      "epoch": 2.0952380952380953,
+      "grad_norm": 0.98046875,
+      "learning_rate": 9.372094804706867e-05,
+      "loss": 0.3258,
+      "step": 660
+    },
+    {
+      "epoch": 2.0984126984126985,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.347013666703641e-05,
+      "loss": 0.3608,
+      "step": 661
+    },
+    {
+      "epoch": 2.1015873015873017,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.321936653316995e-05,
+      "loss": 0.3818,
+      "step": 662
+    },
+    {
+      "epoch": 2.104761904761905,
+      "grad_norm": 1.2265625,
+      "learning_rate": 9.296863922946963e-05,
+      "loss": 0.4647,
+      "step": 663
+    },
+    {
+      "epoch": 2.107936507936508,
+      "grad_norm": 1.203125,
+      "learning_rate": 9.271795633966543e-05,
+      "loss": 0.3625,
+      "step": 664
+    },
+    {
+      "epoch": 2.111111111111111,
+      "grad_norm": 0.859375,
+      "learning_rate": 9.246731944720675e-05,
+      "loss": 0.289,
+      "step": 665
+    },
+    {
+      "epoch": 2.1142857142857143,
+      "grad_norm": 1.2734375,
+      "learning_rate": 9.221673013525235e-05,
+      "loss": 0.4508,
+      "step": 666
+    },
+    {
+      "epoch": 2.1174603174603175,
+      "grad_norm": 1.578125,
+      "learning_rate": 9.196618998666059e-05,
+      "loss": 0.5279,
+      "step": 667
+    },
+    {
+      "epoch": 2.1206349206349207,
+      "grad_norm": 1.1953125,
+      "learning_rate": 9.171570058397917e-05,
+      "loss": 0.368,
+      "step": 668
+    },
+    {
+      "epoch": 2.123809523809524,
+      "grad_norm": 1.3359375,
+      "learning_rate": 9.146526350943532e-05,
+      "loss": 0.4549,
+      "step": 669
+    },
+    {
+      "epoch": 2.126984126984127,
+      "grad_norm": 1.3984375,
+      "learning_rate": 9.121488034492569e-05,
+      "loss": 0.4058,
+      "step": 670
+    },
+    {
+      "epoch": 2.13015873015873,
+      "grad_norm": 1.328125,
+      "learning_rate": 9.096455267200643e-05,
+      "loss": 0.445,
+      "step": 671
+    },
+    {
+      "epoch": 2.1333333333333333,
+      "grad_norm": 1.359375,
+      "learning_rate": 9.071428207188324e-05,
+      "loss": 0.4949,
+      "step": 672
+    },
+    {
+      "epoch": 2.1365079365079365,
+      "grad_norm": 1.328125,
+      "learning_rate": 9.046407012540115e-05,
+      "loss": 0.4749,
+      "step": 673
+    },
+    {
+      "epoch": 2.1396825396825396,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.021391841303484e-05,
+      "loss": 0.3438,
+      "step": 674
+    },
+    {
+      "epoch": 2.142857142857143,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.99638285148785e-05,
+      "loss": 0.3742,
+      "step": 675
+    },
+    {
+      "epoch": 2.146031746031746,
+      "grad_norm": 1.3359375,
+      "learning_rate": 8.971380201063587e-05,
+      "loss": 0.4277,
+      "step": 676
+    },
+    {
+      "epoch": 2.149206349206349,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.946384047961018e-05,
+      "loss": 0.4721,
+      "step": 677
+    },
+    {
+      "epoch": 2.1523809523809523,
+      "grad_norm": 1.390625,
+      "learning_rate": 8.921394550069434e-05,
+      "loss": 0.4714,
+      "step": 678
+    },
+    {
+      "epoch": 2.1555555555555554,
+      "grad_norm": 1.1015625,
+      "learning_rate": 8.896411865236084e-05,
+      "loss": 0.349,
+      "step": 679
+    },
+    {
+      "epoch": 2.1587301587301586,
+      "grad_norm": 1.2890625,
+      "learning_rate": 8.871436151265184e-05,
+      "loss": 0.4615,
+      "step": 680
+    },
+    {
+      "epoch": 2.1619047619047618,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.846467565916917e-05,
+      "loss": 0.4284,
+      "step": 681
+    },
+    {
+      "epoch": 2.165079365079365,
+      "grad_norm": 1.078125,
+      "learning_rate": 8.821506266906444e-05,
+      "loss": 0.4101,
+      "step": 682
+    },
+    {
+      "epoch": 2.168253968253968,
+      "grad_norm": 0.734375,
+      "learning_rate": 8.796552411902886e-05,
+      "loss": 0.2756,
+      "step": 683
+    },
+    {
+      "epoch": 2.1714285714285713,
+      "grad_norm": 1.7109375,
+      "learning_rate": 8.77160615852836e-05,
+      "loss": 0.3601,
+      "step": 684
+    },
+    {
+      "epoch": 2.1746031746031744,
+      "grad_norm": 1.3828125,
+      "learning_rate": 8.746667664356956e-05,
+      "loss": 0.4789,
+      "step": 685
+    },
+    {
+      "epoch": 2.1777777777777776,
+      "grad_norm": 0.890625,
+      "learning_rate": 8.721737086913766e-05,
+      "loss": 0.2943,
+      "step": 686
+    },
+    {
+      "epoch": 2.1809523809523808,
+      "grad_norm": 1.296875,
+      "learning_rate": 8.696814583673856e-05,
+      "loss": 0.5221,
+      "step": 687
+    },
+    {
+      "epoch": 2.1841269841269844,
+      "grad_norm": 2.453125,
+      "learning_rate": 8.671900312061307e-05,
+      "loss": 0.4562,
+      "step": 688
+    },
+    {
+      "epoch": 2.1873015873015875,
+      "grad_norm": 1.1328125,
+      "learning_rate": 8.646994429448199e-05,
+      "loss": 0.391,
+      "step": 689
+    },
+    {
+      "epoch": 2.1904761904761907,
+      "grad_norm": 1.34375,
+      "learning_rate": 8.62209709315362e-05,
+      "loss": 0.4538,
+      "step": 690
+    },
+    {
+      "epoch": 2.193650793650794,
+      "grad_norm": 1.203125,
+      "learning_rate": 8.597208460442681e-05,
+      "loss": 0.4248,
+      "step": 691
+    },
+    {
+      "epoch": 2.196825396825397,
+      "grad_norm": 1.0390625,
+      "learning_rate": 8.572328688525513e-05,
+      "loss": 0.3982,
+      "step": 692
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.125,
+      "learning_rate": 8.547457934556274e-05,
+      "loss": 0.4328,
+      "step": 693
+    },
+    {
+      "epoch": 2.2031746031746033,
+      "grad_norm": 1.21875,
+      "learning_rate": 8.522596355632164e-05,
+      "loss": 0.3847,
+      "step": 694
+    },
+    {
+      "epoch": 2.2063492063492065,
+      "grad_norm": 0.953125,
+      "learning_rate": 8.497744108792429e-05,
+      "loss": 0.3613,
+      "step": 695
+    },
+    {
+      "epoch": 2.2095238095238097,
+      "grad_norm": 1.0078125,
+      "learning_rate": 8.47290135101737e-05,
+      "loss": 0.392,
+      "step": 696
+    },
+    {
+      "epoch": 2.212698412698413,
+      "grad_norm": 2.796875,
+      "learning_rate": 8.44806823922734e-05,
+      "loss": 0.3758,
+      "step": 697
+    },
+    {
+      "epoch": 2.215873015873016,
+      "grad_norm": 1.3046875,
+      "learning_rate": 8.423244930281775e-05,
+      "loss": 0.3447,
+      "step": 698
+    },
+    {
+      "epoch": 2.219047619047619,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.398431580978181e-05,
+      "loss": 0.2935,
+      "step": 699
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.80078125,
+      "learning_rate": 8.373628348051165e-05,
+      "loss": 0.2817,
+      "step": 700
+    },
+    {
+      "epoch": 2.2253968253968255,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.348835388171421e-05,
+      "loss": 0.3881,
+      "step": 701
+    },
+    {
+      "epoch": 2.2285714285714286,
+      "grad_norm": 1.15625,
+      "learning_rate": 8.32405285794476e-05,
+      "loss": 0.3386,
+      "step": 702
+    },
+    {
+      "epoch": 2.231746031746032,
+      "grad_norm": 1.4609375,
+      "learning_rate": 8.299280913911111e-05,
+      "loss": 0.4531,
+      "step": 703
+    },
+    {
+      "epoch": 2.234920634920635,
+      "grad_norm": 0.92578125,
+      "learning_rate": 8.274519712543534e-05,
+      "loss": 0.2924,
+      "step": 704
+    },
+    {
+      "epoch": 2.238095238095238,
+      "grad_norm": 0.875,
+      "learning_rate": 8.249769410247239e-05,
+      "loss": 0.3409,
+      "step": 705
+    },
+    {
+      "epoch": 2.2412698412698413,
+      "grad_norm": 1.046875,
+      "learning_rate": 8.225030163358589e-05,
+      "loss": 0.3245,
+      "step": 706
+    },
+    {
+      "epoch": 2.2444444444444445,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.200302128144104e-05,
+      "loss": 0.3558,
+      "step": 707
+    },
+    {
+      "epoch": 2.2476190476190476,
+      "grad_norm": 1.390625,
+      "learning_rate": 8.175585460799499e-05,
+      "loss": 0.3143,
+      "step": 708
+    },
+    {
+      "epoch": 2.250793650793651,
+      "grad_norm": 1.0078125,
+      "learning_rate": 8.150880317448676e-05,
+      "loss": 0.3951,
+      "step": 709
+    },
+    {
+      "epoch": 2.253968253968254,
+      "grad_norm": 0.91015625,
+      "learning_rate": 8.126186854142752e-05,
+      "loss": 0.3655,
+      "step": 710
+    },
+    {
+      "epoch": 2.257142857142857,
+      "grad_norm": 0.984375,
+      "learning_rate": 8.101505226859063e-05,
+      "loss": 0.2894,
+      "step": 711
+    },
+    {
+      "epoch": 2.2603174603174603,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.076835591500173e-05,
+      "loss": 0.3238,
+      "step": 712
+    },
+    {
+      "epoch": 2.2634920634920634,
+      "grad_norm": 1.234375,
+      "learning_rate": 8.052178103892914e-05,
+      "loss": 0.3676,
+      "step": 713
+    },
+    {
+      "epoch": 2.2666666666666666,
+      "grad_norm": 1.328125,
+      "learning_rate": 8.027532919787371e-05,
+      "loss": 0.3691,
+      "step": 714
+    },
+    {
+      "epoch": 2.2698412698412698,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.002900194855932e-05,
+      "loss": 0.3479,
+      "step": 715
+    },
+    {
+      "epoch": 2.273015873015873,
+      "grad_norm": 0.91796875,
+      "learning_rate": 7.978280084692267e-05,
+      "loss": 0.353,
+      "step": 716
+    },
+    {
+      "epoch": 2.276190476190476,
+      "grad_norm": 1.453125,
+      "learning_rate": 7.953672744810375e-05,
+      "loss": 0.4352,
+      "step": 717
+    },
+    {
+      "epoch": 2.2793650793650793,
+      "grad_norm": 0.8671875,
+      "learning_rate": 7.929078330643589e-05,
+      "loss": 0.3503,
+      "step": 718
+    },
+    {
+      "epoch": 2.2825396825396824,
+      "grad_norm": 0.92578125,
+      "learning_rate": 7.904496997543595e-05,
+      "loss": 0.3429,
+      "step": 719
+    },
+    {
+      "epoch": 2.2857142857142856,
+      "grad_norm": 1.15625,
+      "learning_rate": 7.879928900779456e-05,
+      "loss": 0.4511,
+      "step": 720
+    },
+    {
+      "epoch": 2.2888888888888888,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.855374195536618e-05,
+      "loss": 0.3826,
+      "step": 721
+    },
+    {
+      "epoch": 2.292063492063492,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.830833036915948e-05,
+      "loss": 0.3938,
+      "step": 722
+    },
+    {
+      "epoch": 2.295238095238095,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.80630557993274e-05,
+      "loss": 0.3785,
+      "step": 723
+    },
+    {
+      "epoch": 2.2984126984126982,
+      "grad_norm": 1.1484375,
+      "learning_rate": 7.781791979515748e-05,
+      "loss": 0.3847,
+      "step": 724
+    },
+    {
+      "epoch": 2.3015873015873014,
+      "grad_norm": 1.0546875,
+      "learning_rate": 7.75729239050619e-05,
+      "loss": 0.3841,
+      "step": 725
+    },
+    {
+      "epoch": 2.3047619047619046,
+      "grad_norm": 0.90234375,
+      "learning_rate": 7.732806967656785e-05,
+      "loss": 0.3225,
+      "step": 726
+    },
+    {
+      "epoch": 2.3079365079365077,
+      "grad_norm": 1.390625,
+      "learning_rate": 7.708335865630774e-05,
+      "loss": 0.3714,
+      "step": 727
+    },
+    {
+      "epoch": 2.311111111111111,
+      "grad_norm": 0.85546875,
+      "learning_rate": 7.683879239000936e-05,
+      "loss": 0.3355,
+      "step": 728
+    },
+    {
+      "epoch": 2.314285714285714,
+      "grad_norm": 0.91796875,
+      "learning_rate": 7.659437242248617e-05,
+      "loss": 0.2994,
+      "step": 729
+    },
+    {
+      "epoch": 2.317460317460317,
+      "grad_norm": 0.859375,
+      "learning_rate": 7.635010029762756e-05,
+      "loss": 0.3591,
+      "step": 730
+    },
+    {
+      "epoch": 2.320634920634921,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.610597755838895e-05,
+      "loss": 0.3489,
+      "step": 731
+    },
+    {
+      "epoch": 2.323809523809524,
+      "grad_norm": 1.2578125,
+      "learning_rate": 7.586200574678231e-05,
+      "loss": 0.4117,
+      "step": 732
+    },
+    {
+      "epoch": 2.326984126984127,
+      "grad_norm": 0.83203125,
+      "learning_rate": 7.561818640386616e-05,
+      "loss": 0.3369,
+      "step": 733
+    },
+    {
+      "epoch": 2.3301587301587303,
+      "grad_norm": 1.1953125,
+      "learning_rate": 7.5374521069736e-05,
+      "loss": 0.418,
+      "step": 734
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 1.40625,
+      "learning_rate": 7.513101128351454e-05,
+      "loss": 0.3984,
+      "step": 735
+    },
+    {
+      "epoch": 2.3365079365079366,
+      "grad_norm": 0.99609375,
+      "learning_rate": 7.488765858334188e-05,
+      "loss": 0.368,
+      "step": 736
+    },
+    {
+      "epoch": 2.33968253968254,
+      "grad_norm": 0.87890625,
+      "learning_rate": 7.464446450636596e-05,
+      "loss": 0.2867,
+      "step": 737
+    },
+    {
+      "epoch": 2.342857142857143,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.440143058873274e-05,
+      "loss": 0.3273,
+      "step": 738
+    },
+    {
+      "epoch": 2.346031746031746,
+      "grad_norm": 0.82421875,
+      "learning_rate": 7.415855836557653e-05,
+      "loss": 0.2462,
+      "step": 739
+    },
+    {
+      "epoch": 2.3492063492063493,
+      "grad_norm": 1.015625,
+      "learning_rate": 7.391584937101033e-05,
+      "loss": 0.3512,
+      "step": 740
+    },
+    {
+      "epoch": 2.3523809523809525,
+      "grad_norm": 1.46875,
+      "learning_rate": 7.367330513811598e-05,
+      "loss": 0.5001,
+      "step": 741
+    },
+    {
+      "epoch": 2.3555555555555556,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.343092719893474e-05,
+      "loss": 0.3594,
+      "step": 742
+    },
+    {
+      "epoch": 2.358730158730159,
+      "grad_norm": 0.89453125,
+      "learning_rate": 7.318871708445738e-05,
+      "loss": 0.3201,
+      "step": 743
+    },
+    {
+      "epoch": 2.361904761904762,
+      "grad_norm": 1.640625,
+      "learning_rate": 7.294667632461463e-05,
+      "loss": 0.501,
+      "step": 744
+    },
+    {
+      "epoch": 2.365079365079365,
+      "grad_norm": 1.2109375,
+      "learning_rate": 7.270480644826749e-05,
+      "loss": 0.3816,
+      "step": 745
+    },
+    {
+      "epoch": 2.3682539682539683,
+      "grad_norm": 1.2421875,
+      "learning_rate": 7.246310898319752e-05,
+      "loss": 0.4339,
+      "step": 746
+    },
+    {
+      "epoch": 2.3714285714285714,
+      "grad_norm": 1.140625,
+      "learning_rate": 7.222158545609727e-05,
+      "loss": 0.4128,
+      "step": 747
+    },
+    {
+      "epoch": 2.3746031746031746,
+      "grad_norm": 1.1171875,
+      "learning_rate": 7.198023739256066e-05,
+      "loss": 0.3913,
+      "step": 748
+    },
+    {
+      "epoch": 2.3777777777777778,
+      "grad_norm": 0.875,
+      "learning_rate": 7.173906631707317e-05,
+      "loss": 0.3008,
+      "step": 749
+    },
+    {
+      "epoch": 2.380952380952381,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.149807375300239e-05,
+      "loss": 0.5586,
+      "step": 750
+    },
+    {
+      "epoch": 2.384126984126984,
+      "grad_norm": 0.76953125,
+      "learning_rate": 7.125726122258837e-05,
+      "loss": 0.3376,
+      "step": 751
+    },
+    {
+      "epoch": 2.3873015873015873,
+      "grad_norm": 1.1640625,
+      "learning_rate": 7.10166302469339e-05,
+      "loss": 0.3717,
+      "step": 752
+    },
+    {
+      "epoch": 2.3904761904761904,
+      "grad_norm": 0.89453125,
+      "learning_rate": 7.077618234599501e-05,
+      "loss": 0.3263,
+      "step": 753
+    },
+    {
+      "epoch": 2.3936507936507936,
+      "grad_norm": 0.97265625,
+      "learning_rate": 7.053591903857138e-05,
+      "loss": 0.2982,
+      "step": 754
+    },
+    {
+      "epoch": 2.3968253968253967,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.029584184229653e-05,
+      "loss": 0.3273,
+      "step": 755
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.005595227362858e-05,
+      "loss": 0.3509,
+      "step": 756
+    },
+    {
+      "epoch": 2.403174603174603,
+      "grad_norm": 0.8671875,
+      "learning_rate": 6.98162518478404e-05,
+      "loss": 0.3078,
+      "step": 757
+    },
+    {
+      "epoch": 2.4063492063492062,
+      "grad_norm": 0.80859375,
+      "learning_rate": 6.957674207901018e-05,
+      "loss": 0.2682,
+      "step": 758
+    },
+    {
+      "epoch": 2.4095238095238094,
+      "grad_norm": 1.2109375,
+      "learning_rate": 6.933742448001183e-05,
+      "loss": 0.4161,
+      "step": 759
+    },
+    {
+      "epoch": 2.4126984126984126,
+      "grad_norm": 1.296875,
+      "learning_rate": 6.909830056250527e-05,
+      "loss": 0.3541,
+      "step": 760
+    },
+    {
+      "epoch": 2.4158730158730157,
+      "grad_norm": 0.9453125,
+      "learning_rate": 6.885937183692717e-05,
+      "loss": 0.3131,
+      "step": 761
+    },
+    {
+      "epoch": 2.419047619047619,
+      "grad_norm": 0.97265625,
+      "learning_rate": 6.862063981248125e-05,
+      "loss": 0.3959,
+      "step": 762
+    },
+    {
+      "epoch": 2.422222222222222,
+      "grad_norm": 1.140625,
+      "learning_rate": 6.838210599712869e-05,
+      "loss": 0.4169,
+      "step": 763
+    },
+    {
+      "epoch": 2.425396825396825,
+      "grad_norm": 0.91015625,
+      "learning_rate": 6.814377189757876e-05,
+      "loss": 0.3018,
+      "step": 764
+    },
+    {
+      "epoch": 2.4285714285714284,
+      "grad_norm": 0.8671875,
+      "learning_rate": 6.790563901927907e-05,
+      "loss": 0.3005,
+      "step": 765
+    },
+    {
+      "epoch": 2.431746031746032,
+      "grad_norm": 0.7890625,
+      "learning_rate": 6.766770886640637e-05,
+      "loss": 0.2802,
+      "step": 766
+    },
+    {
+      "epoch": 2.434920634920635,
+      "grad_norm": 0.9140625,
+      "learning_rate": 6.742998294185678e-05,
+      "loss": 0.3592,
+      "step": 767
+    },
+    {
+      "epoch": 2.4380952380952383,
+      "grad_norm": 1.125,
+      "learning_rate": 6.71924627472365e-05,
+      "loss": 0.3235,
+      "step": 768
+    },
+    {
+      "epoch": 2.4412698412698415,
+      "grad_norm": 0.77734375,
+      "learning_rate": 6.695514978285215e-05,
+      "loss": 0.2693,
+      "step": 769
+    },
+    {
+      "epoch": 2.4444444444444446,
+      "grad_norm": 0.875,
+      "learning_rate": 6.671804554770135e-05,
+      "loss": 0.2598,
+      "step": 770
+    },
+    {
+      "epoch": 2.447619047619048,
+      "grad_norm": 0.92578125,
+      "learning_rate": 6.648115153946333e-05,
+      "loss": 0.3215,
+      "step": 771
+    },
+    {
+      "epoch": 2.450793650793651,
+      "grad_norm": 1.109375,
+      "learning_rate": 6.624446925448944e-05,
+      "loss": 0.4423,
+      "step": 772
+    },
+    {
+      "epoch": 2.453968253968254,
+      "grad_norm": 0.82421875,
+      "learning_rate": 6.600800018779356e-05,
+      "loss": 0.3098,
+      "step": 773
+    },
+    {
+      "epoch": 2.4571428571428573,
+      "grad_norm": 0.79296875,
+      "learning_rate": 6.577174583304289e-05,
+      "loss": 0.2384,
+      "step": 774
+    },
+    {
+      "epoch": 2.4603174603174605,
+      "grad_norm": 0.85546875,
+      "learning_rate": 6.55357076825483e-05,
+      "loss": 0.3022,
+      "step": 775
+    },
+    {
+      "epoch": 2.4634920634920636,
+      "grad_norm": 1.2578125,
+      "learning_rate": 6.529988722725507e-05,
+      "loss": 0.3149,
+      "step": 776
+    },
+    {
+      "epoch": 2.466666666666667,
+      "grad_norm": 1.109375,
+      "learning_rate": 6.506428595673331e-05,
+      "loss": 0.2943,
+      "step": 777
+    },
+    {
+      "epoch": 2.46984126984127,
+      "grad_norm": 1.3828125,
+      "learning_rate": 6.482890535916876e-05,
+      "loss": 0.4057,
+      "step": 778
+    },
+    {
+      "epoch": 2.473015873015873,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.459374692135312e-05,
+      "loss": 0.321,
+      "step": 779
+    },
+    {
+      "epoch": 2.4761904761904763,
+      "grad_norm": 0.8046875,
+      "learning_rate": 6.435881212867493e-05,
+      "loss": 0.2725,
+      "step": 780
+    },
+    {
+      "epoch": 2.4793650793650794,
+      "grad_norm": 0.7421875,
+      "learning_rate": 6.412410246511004e-05,
+      "loss": 0.3071,
+      "step": 781
+    },
+    {
+      "epoch": 2.4825396825396826,
+      "grad_norm": 1.0859375,
+      "learning_rate": 6.388961941321225e-05,
+      "loss": 0.368,
+      "step": 782
+    },
+    {
+      "epoch": 2.4857142857142858,
+      "grad_norm": 0.875,
+      "learning_rate": 6.365536445410397e-05,
+      "loss": 0.3139,
+      "step": 783
+    },
+    {
+      "epoch": 2.488888888888889,
+      "grad_norm": 0.99609375,
+      "learning_rate": 6.342133906746676e-05,
+      "loss": 0.3526,
+      "step": 784
+    },
+    {
+      "epoch": 2.492063492063492,
+      "grad_norm": 1.0859375,
+      "learning_rate": 6.318754473153221e-05,
+      "loss": 0.3559,
+      "step": 785
+    },
+    {
+      "epoch": 2.4952380952380953,
+      "grad_norm": 0.9375,
+      "learning_rate": 6.295398292307242e-05,
+      "loss": 0.3453,
+      "step": 786
+    },
+    {
+      "epoch": 2.4984126984126984,
+      "grad_norm": 0.89453125,
+      "learning_rate": 6.272065511739067e-05,
+      "loss": 0.3346,
+      "step": 787
+    },
+    {
+      "epoch": 2.5015873015873016,
+      "grad_norm": 1.1640625,
+      "learning_rate": 6.248756278831228e-05,
+      "loss": 0.2742,
+      "step": 788
+    },
+    {
+      "epoch": 2.5047619047619047,
+      "grad_norm": 0.76953125,
+      "learning_rate": 6.225470740817496e-05,
+      "loss": 0.32,
+      "step": 789
+    },
+    {
+      "epoch": 2.507936507936508,
+      "grad_norm": 0.984375,
+      "learning_rate": 6.20220904478199e-05,
+      "loss": 0.2951,
+      "step": 790
+    },
+    {
+      "epoch": 2.511111111111111,
+      "grad_norm": 0.8515625,
+      "learning_rate": 6.178971337658225e-05,
+      "loss": 0.337,
+      "step": 791
+    },
+    {
+      "epoch": 2.5142857142857142,
+      "grad_norm": 0.85546875,
+      "learning_rate": 6.155757766228192e-05,
+      "loss": 0.3086,
+      "step": 792
+    },
+    {
+      "epoch": 2.5174603174603174,
+      "grad_norm": 0.6015625,
+      "learning_rate": 6.132568477121426e-05,
+      "loss": 0.2602,
+      "step": 793
+    },
+    {
+      "epoch": 2.5206349206349206,
+      "grad_norm": 1.171875,
+      "learning_rate": 6.109403616814069e-05,
+      "loss": 0.3609,
+      "step": 794
+    },
+    {
+      "epoch": 2.5238095238095237,
+      "grad_norm": 0.79296875,
+      "learning_rate": 6.086263331627976e-05,
+      "loss": 0.3326,
+      "step": 795
+    },
+    {
+      "epoch": 2.526984126984127,
+      "grad_norm": 1.0703125,
+      "learning_rate": 6.063147767729764e-05,
+      "loss": 0.2818,
+      "step": 796
+    },
+    {
+      "epoch": 2.53015873015873,
+      "grad_norm": 2.1875,
+      "learning_rate": 6.040057071129901e-05,
+      "loss": 0.4045,
+      "step": 797
+    },
+    {
+      "epoch": 2.533333333333333,
+      "grad_norm": 0.84375,
+      "learning_rate": 6.016991387681774e-05,
+      "loss": 0.3041,
+      "step": 798
+    },
+    {
+      "epoch": 2.5365079365079364,
+      "grad_norm": 1.296875,
+      "learning_rate": 5.993950863080773e-05,
+      "loss": 0.368,
+      "step": 799
+    },
+    {
+      "epoch": 2.5396825396825395,
+      "grad_norm": 0.70703125,
+      "learning_rate": 5.9709356428633746e-05,
+      "loss": 0.2315,
+      "step": 800
+    },
+    {
+      "epoch": 2.5428571428571427,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5.94794587240622e-05,
+      "loss": 0.3353,
+      "step": 801
+    },
+    {
+      "epoch": 2.546031746031746,
+      "grad_norm": 0.84765625,
+      "learning_rate": 5.9249816969251914e-05,
+      "loss": 0.3296,
+      "step": 802
+    },
+    {
+      "epoch": 2.549206349206349,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.90204326147451e-05,
+      "loss": 0.3961,
+      "step": 803
+    },
+    {
+      "epoch": 2.552380952380952,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.879130710945791e-05,
+      "loss": 0.4509,
+      "step": 804
+    },
+    {
+      "epoch": 2.5555555555555554,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.856244190067159e-05,
+      "loss": 0.3449,
+      "step": 805
+    },
+    {
+      "epoch": 2.5587301587301585,
+      "grad_norm": 1.40625,
+      "learning_rate": 5.8333838434023203e-05,
+      "loss": 0.4789,
+      "step": 806
+    },
+    {
+      "epoch": 2.5619047619047617,
+      "grad_norm": 0.9609375,
+      "learning_rate": 5.810549815349653e-05,
+      "loss": 0.3823,
+      "step": 807
+    },
+    {
+      "epoch": 2.565079365079365,
+      "grad_norm": 0.8046875,
+      "learning_rate": 5.787742250141279e-05,
+      "loss": 0.2819,
+      "step": 808
+    },
+    {
+      "epoch": 2.568253968253968,
+      "grad_norm": 0.70703125,
+      "learning_rate": 5.7649612918421816e-05,
+      "loss": 0.2503,
+      "step": 809
+    },
+    {
+      "epoch": 2.571428571428571,
+      "grad_norm": 0.94921875,
+      "learning_rate": 5.7422070843492734e-05,
+      "loss": 0.3181,
+      "step": 810
+    },
+    {
+      "epoch": 2.5746031746031743,
+      "grad_norm": 0.8828125,
+      "learning_rate": 5.719479771390494e-05,
+      "loss": 0.3024,
+      "step": 811
+    },
+    {
+      "epoch": 2.5777777777777775,
+      "grad_norm": 1.4140625,
+      "learning_rate": 5.696779496523913e-05,
+      "loss": 0.4931,
+      "step": 812
+    },
+    {
+      "epoch": 2.580952380952381,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.674106403136792e-05,
+      "loss": 0.4034,
+      "step": 813
+    },
+    {
+      "epoch": 2.5841269841269843,
+      "grad_norm": 0.796875,
+      "learning_rate": 5.651460634444716e-05,
+      "loss": 0.2961,
+      "step": 814
+    },
+    {
+      "epoch": 2.5873015873015874,
+      "grad_norm": 0.6796875,
+      "learning_rate": 5.6288423334906735e-05,
+      "loss": 0.2475,
+      "step": 815
+    },
+    {
+      "epoch": 2.5904761904761906,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.606251643144136e-05,
+      "loss": 0.3227,
+      "step": 816
+    },
+    {
+      "epoch": 2.5936507936507938,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.5836887061001875e-05,
+      "loss": 0.3589,
+      "step": 817
+    },
+    {
+      "epoch": 2.596825396825397,
+      "grad_norm": 0.8984375,
+      "learning_rate": 5.561153664878603e-05,
+      "loss": 0.3451,
+      "step": 818
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.4921875,
+      "learning_rate": 5.5386466618229436e-05,
+      "loss": 0.4239,
+      "step": 819
+    },
+    {
+      "epoch": 2.6031746031746033,
+      "grad_norm": 0.67578125,
+      "learning_rate": 5.5161678390996796e-05,
+      "loss": 0.2481,
+      "step": 820
+    },
+    {
+      "epoch": 2.6063492063492064,
+      "grad_norm": 1.21875,
+      "learning_rate": 5.493717338697267e-05,
+      "loss": 0.4634,
+      "step": 821
+    },
+    {
+      "epoch": 2.6095238095238096,
+      "grad_norm": 0.99609375,
+      "learning_rate": 5.471295302425276e-05,
+      "loss": 0.3226,
+      "step": 822
+    },
+    {
+      "epoch": 2.6126984126984127,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.4489018719134654e-05,
+      "loss": 0.3926,
+      "step": 823
+    },
+    {
+      "epoch": 2.615873015873016,
+      "grad_norm": 0.875,
+      "learning_rate": 5.4265371886109163e-05,
+      "loss": 0.3566,
+      "step": 824
+    },
+    {
+      "epoch": 2.619047619047619,
+      "grad_norm": 0.8046875,
+      "learning_rate": 5.404201393785122e-05,
+      "loss": 0.2529,
+      "step": 825
+    },
+    {
+      "epoch": 2.6222222222222222,
+      "grad_norm": 0.9921875,
+      "learning_rate": 5.381894628521105e-05,
+      "loss": 0.4329,
+      "step": 826
+    },
+    {
+      "epoch": 2.6253968253968254,
+      "grad_norm": 0.89453125,
+      "learning_rate": 5.359617033720519e-05,
+      "loss": 0.3394,
+      "step": 827
+    },
+    {
+      "epoch": 2.6285714285714286,
+      "grad_norm": 1.5234375,
+      "learning_rate": 5.337368750100755e-05,
+      "loss": 0.4235,
+      "step": 828
+    },
+    {
+      "epoch": 2.6317460317460317,
+      "grad_norm": 0.70703125,
+      "learning_rate": 5.315149918194067e-05,
+      "loss": 0.2788,
+      "step": 829
+    },
+    {
+      "epoch": 2.634920634920635,
+      "grad_norm": 1.3359375,
+      "learning_rate": 5.292960678346675e-05,
+      "loss": 0.4028,
+      "step": 830
+    },
+    {
+      "epoch": 2.638095238095238,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.270801170717874e-05,
+      "loss": 0.4042,
+      "step": 831
+    },
+    {
+      "epoch": 2.641269841269841,
+      "grad_norm": 0.77734375,
+      "learning_rate": 5.2486715352791636e-05,
+      "loss": 0.298,
+      "step": 832
+    },
+    {
+      "epoch": 2.6444444444444444,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.2265719118133404e-05,
+      "loss": 0.2849,
+      "step": 833
+    },
+    {
+      "epoch": 2.6476190476190475,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5.204502439913641e-05,
+      "loss": 0.2876,
+      "step": 834
+    },
+    {
+      "epoch": 2.6507936507936507,
+      "grad_norm": 0.7578125,
+      "learning_rate": 5.182463258982846e-05,
+      "loss": 0.2602,
+      "step": 835
+    },
+    {
+      "epoch": 2.653968253968254,
+      "grad_norm": 0.890625,
+      "learning_rate": 5.160454508232397e-05,
+      "loss": 0.3465,
+      "step": 836
+    },
+    {
+      "epoch": 2.657142857142857,
+      "grad_norm": 1.0,
+      "learning_rate": 5.138476326681534e-05,
+      "loss": 0.3359,
+      "step": 837
+    },
+    {
+      "epoch": 2.66031746031746,
+      "grad_norm": 0.8046875,
+      "learning_rate": 5.116528853156384e-05,
+      "loss": 0.2723,
+      "step": 838
+    },
+    {
+      "epoch": 2.6634920634920634,
+      "grad_norm": 0.88671875,
+      "learning_rate": 5.094612226289128e-05,
+      "loss": 0.3337,
+      "step": 839
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.7578125,
+      "learning_rate": 5.072726584517086e-05,
+      "loss": 0.2558,
+      "step": 840
+    },
+    {
+      "epoch": 2.6698412698412697,
+      "grad_norm": 0.7890625,
+      "learning_rate": 5.0508720660818684e-05,
+      "loss": 0.2919,
+      "step": 841
+    },
+    {
+      "epoch": 2.6730158730158733,
+      "grad_norm": 0.7265625,
+      "learning_rate": 5.029048809028496e-05,
+      "loss": 0.2784,
+      "step": 842
+    },
+    {
+      "epoch": 2.6761904761904765,
+      "grad_norm": 0.90625,
+      "learning_rate": 5.007256951204512e-05,
+      "loss": 0.2693,
+      "step": 843
+    },
+    {
+      "epoch": 2.6793650793650796,
+      "grad_norm": 0.91015625,
+      "learning_rate": 4.98549663025914e-05,
+      "loss": 0.2652,
+      "step": 844
+    },
+    {
+      "epoch": 2.682539682539683,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.9637679836423924e-05,
+      "loss": 0.3137,
+      "step": 845
+    },
+    {
+      "epoch": 2.685714285714286,
+      "grad_norm": 0.9140625,
+      "learning_rate": 4.942071148604215e-05,
+      "loss": 0.3335,
+      "step": 846
+    },
+    {
+      "epoch": 2.688888888888889,
+      "grad_norm": 0.87890625,
+      "learning_rate": 4.920406262193602e-05,
+      "loss": 0.2466,
+      "step": 847
+    },
+    {
+      "epoch": 2.6920634920634923,
+      "grad_norm": 0.72265625,
+      "learning_rate": 4.898773461257754e-05,
+      "loss": 0.274,
+      "step": 848
+    },
+    {
+      "epoch": 2.6952380952380954,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.8771728824412e-05,
+      "loss": 0.3877,
+      "step": 849
+    },
+    {
+      "epoch": 2.6984126984126986,
+      "grad_norm": 0.890625,
+      "learning_rate": 4.8556046621849346e-05,
+      "loss": 0.3699,
+      "step": 850
+    },
+    {
+      "epoch": 2.7015873015873018,
+      "grad_norm": 1.078125,
+      "learning_rate": 4.8340689367255643e-05,
+      "loss": 0.3785,
+      "step": 851
+    },
+    {
+      "epoch": 2.704761904761905,
+      "grad_norm": 1.3359375,
+      "learning_rate": 4.812565842094426e-05,
+      "loss": 0.3714,
+      "step": 852
+    },
+    {
+      "epoch": 2.707936507936508,
+      "grad_norm": 0.80078125,
+      "learning_rate": 4.791095514116758e-05,
+      "loss": 0.3141,
+      "step": 853
+    },
+    {
+      "epoch": 2.7111111111111112,
+      "grad_norm": 0.7734375,
+      "learning_rate": 4.769658088410819e-05,
+      "loss": 0.2937,
+      "step": 854
+    },
+    {
+      "epoch": 2.7142857142857144,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.748253700387042e-05,
+      "loss": 0.3064,
+      "step": 855
+    },
+    {
+      "epoch": 2.7174603174603176,
+      "grad_norm": 0.8671875,
+      "learning_rate": 4.726882485247177e-05,
+      "loss": 0.256,
+      "step": 856
+    },
+    {
+      "epoch": 2.7206349206349207,
+      "grad_norm": 0.96484375,
+      "learning_rate": 4.705544577983429e-05,
+      "loss": 0.3584,
+      "step": 857
+    },
+    {
+      "epoch": 2.723809523809524,
+      "grad_norm": 0.625,
+      "learning_rate": 4.684240113377619e-05,
+      "loss": 0.2067,
+      "step": 858
+    },
+    {
+      "epoch": 2.726984126984127,
+      "grad_norm": 0.80859375,
+      "learning_rate": 4.6629692260003245e-05,
+      "loss": 0.2946,
+      "step": 859
+    },
+    {
+      "epoch": 2.7301587301587302,
+      "grad_norm": 0.5859375,
+      "learning_rate": 4.6417320502100316e-05,
+      "loss": 0.194,
+      "step": 860
+    },
+    {
+      "epoch": 2.7333333333333334,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.620528720152288e-05,
+      "loss": 0.3077,
+      "step": 861
+    },
+    {
+      "epoch": 2.7365079365079366,
+      "grad_norm": 0.8359375,
+      "learning_rate": 4.5993593697588424e-05,
+      "loss": 0.2563,
+      "step": 862
+    },
+    {
+      "epoch": 2.7396825396825397,
+      "grad_norm": 0.75390625,
+      "learning_rate": 4.578224132746821e-05,
+      "loss": 0.2242,
+      "step": 863
+    },
+    {
+      "epoch": 2.742857142857143,
+      "grad_norm": 0.7421875,
+      "learning_rate": 4.557123142617873e-05,
+      "loss": 0.3038,
+      "step": 864
+    },
+    {
+      "epoch": 2.746031746031746,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.5360565326573104e-05,
+      "loss": 0.3418,
+      "step": 865
+    },
+    {
+      "epoch": 2.749206349206349,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.515024435933299e-05,
+      "loss": 0.3777,
+      "step": 866
+    },
+    {
+      "epoch": 2.7523809523809524,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.494026985295987e-05,
+      "loss": 0.3456,
+      "step": 867
+    },
+    {
+      "epoch": 2.7555555555555555,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.473064313376687e-05,
+      "loss": 0.3167,
+      "step": 868
+    },
+    {
+      "epoch": 2.7587301587301587,
+      "grad_norm": 0.8984375,
+      "learning_rate": 4.4521365525870276e-05,
+      "loss": 0.3128,
+      "step": 869
+    },
+    {
+      "epoch": 2.761904761904762,
+      "grad_norm": 0.71484375,
+      "learning_rate": 4.431243835118124e-05,
+      "loss": 0.2748,
+      "step": 870
+    },
+    {
+      "epoch": 2.765079365079365,
+      "grad_norm": 0.84765625,
+      "learning_rate": 4.4103862929397265e-05,
+      "loss": 0.2864,
+      "step": 871
+    },
+    {
+      "epoch": 2.768253968253968,
+      "grad_norm": 0.8671875,
+      "learning_rate": 4.389564057799414e-05,
+      "loss": 0.3589,
+      "step": 872
+    },
+    {
+      "epoch": 2.7714285714285714,
+      "grad_norm": 0.59765625,
+      "learning_rate": 4.368777261221737e-05,
+      "loss": 0.2018,
+      "step": 873
+    },
+    {
+      "epoch": 2.7746031746031745,
+      "grad_norm": 0.8203125,
+      "learning_rate": 4.348026034507402e-05,
+      "loss": 0.3309,
+      "step": 874
+    },
+    {
+      "epoch": 2.7777777777777777,
+      "grad_norm": 0.92578125,
+      "learning_rate": 4.327310508732437e-05,
+      "loss": 0.3023,
+      "step": 875
+    },
+    {
+      "epoch": 2.780952380952381,
+      "grad_norm": 0.88671875,
+      "learning_rate": 4.306630814747358e-05,
+      "loss": 0.2683,
+      "step": 876
+    },
+    {
+      "epoch": 2.784126984126984,
+      "grad_norm": 0.59375,
+      "learning_rate": 4.28598708317635e-05,
+      "loss": 0.2109,
+      "step": 877
+    },
+    {
+      "epoch": 2.787301587301587,
+      "grad_norm": 0.828125,
+      "learning_rate": 4.265379444416445e-05,
+      "loss": 0.2887,
+      "step": 878
+    },
+    {
+      "epoch": 2.7904761904761903,
+      "grad_norm": 0.7890625,
+      "learning_rate": 4.244808028636689e-05,
+      "loss": 0.2735,
+      "step": 879
+    },
+    {
+      "epoch": 2.7936507936507935,
+      "grad_norm": 0.87890625,
+      "learning_rate": 4.224272965777326e-05,
+      "loss": 0.2468,
+      "step": 880
+    },
+    {
+      "epoch": 2.7968253968253967,
+      "grad_norm": 0.828125,
+      "learning_rate": 4.203774385548969e-05,
+      "loss": 0.2968,
+      "step": 881
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.81640625,
+      "learning_rate": 4.183312417431793e-05,
+      "loss": 0.3254,
+      "step": 882
+    },
+    {
+      "epoch": 2.803174603174603,
+      "grad_norm": 1.4140625,
+      "learning_rate": 4.16288719067471e-05,
+      "loss": 0.4478,
+      "step": 883
+    },
+    {
+      "epoch": 2.806349206349206,
+      "grad_norm": 0.671875,
+      "learning_rate": 4.1424988342945545e-05,
+      "loss": 0.2171,
+      "step": 884
+    },
+    {
+      "epoch": 2.8095238095238093,
+      "grad_norm": 0.75390625,
+      "learning_rate": 4.12214747707527e-05,
+      "loss": 0.2867,
+      "step": 885
+    },
+    {
+      "epoch": 2.8126984126984125,
+      "grad_norm": 0.89453125,
+      "learning_rate": 4.1018332475670793e-05,
+      "loss": 0.2554,
+      "step": 886
+    },
+    {
+      "epoch": 2.8158730158730156,
+      "grad_norm": 0.8671875,
+      "learning_rate": 4.081556274085704e-05,
+      "loss": 0.271,
+      "step": 887
+    },
+    {
+      "epoch": 2.819047619047619,
+      "grad_norm": 0.8828125,
+      "learning_rate": 4.0613166847115306e-05,
+      "loss": 0.3127,
+      "step": 888
+    },
+    {
+      "epoch": 2.822222222222222,
+      "grad_norm": 1.4140625,
+      "learning_rate": 4.041114607288799e-05,
+      "loss": 0.3051,
+      "step": 889
+    },
+    {
+      "epoch": 2.825396825396825,
+      "grad_norm": 0.65234375,
+      "learning_rate": 4.020950169424815e-05,
+      "loss": 0.265,
+      "step": 890
+    },
+    {
+      "epoch": 2.8285714285714287,
+      "grad_norm": 0.78515625,
+      "learning_rate": 4.0008234984891226e-05,
+      "loss": 0.2543,
+      "step": 891
+    },
+    {
+      "epoch": 2.831746031746032,
+      "grad_norm": 0.6640625,
+      "learning_rate": 3.9807347216127175e-05,
+      "loss": 0.2326,
+      "step": 892
+    },
+    {
+      "epoch": 2.834920634920635,
+      "grad_norm": 0.9609375,
+      "learning_rate": 3.960683965687232e-05,
+      "loss": 0.3814,
+      "step": 893
+    },
+    {
+      "epoch": 2.8380952380952382,
+      "grad_norm": 0.828125,
+      "learning_rate": 3.940671357364137e-05,
+      "loss": 0.3098,
+      "step": 894
+    },
+    {
+      "epoch": 2.8412698412698414,
+      "grad_norm": 1.578125,
+      "learning_rate": 3.920697023053949e-05,
+      "loss": 0.4053,
+      "step": 895
+    },
+    {
+      "epoch": 2.8444444444444446,
+      "grad_norm": 0.671875,
+      "learning_rate": 3.900761088925411e-05,
+      "loss": 0.2907,
+      "step": 896
+    },
+    {
+      "epoch": 2.8476190476190477,
+      "grad_norm": 0.71484375,
+      "learning_rate": 3.8808636809047204e-05,
+      "loss": 0.2572,
+      "step": 897
+    },
+    {
+      "epoch": 2.850793650793651,
+      "grad_norm": 0.81640625,
+      "learning_rate": 3.8610049246747235e-05,
+      "loss": 0.3257,
+      "step": 898
+    },
+    {
+      "epoch": 2.853968253968254,
+      "grad_norm": 0.7890625,
+      "learning_rate": 3.841184945674114e-05,
+      "loss": 0.2688,
+      "step": 899
+    },
+    {
+      "epoch": 2.857142857142857,
+      "grad_norm": 0.83984375,
+      "learning_rate": 3.821403869096658e-05,
+      "loss": 0.3234,
+      "step": 900
+    },
+    {
+      "epoch": 2.8603174603174604,
+      "grad_norm": 0.6328125,
+      "learning_rate": 3.8016618198903774e-05,
+      "loss": 0.2387,
+      "step": 901
+    },
+    {
+      "epoch": 2.8634920634920635,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.7819589227567886e-05,
+      "loss": 0.2825,
+      "step": 902
+    },
+    {
+      "epoch": 2.8666666666666667,
+      "grad_norm": 0.84765625,
+      "learning_rate": 3.7622953021500976e-05,
+      "loss": 0.2912,
+      "step": 903
+    },
+    {
+      "epoch": 2.86984126984127,
+      "grad_norm": 0.859375,
+      "learning_rate": 3.742671082276423e-05,
+      "loss": 0.2718,
+      "step": 904
+    },
+    {
+      "epoch": 2.873015873015873,
+      "grad_norm": 0.61328125,
+      "learning_rate": 3.7230863870929964e-05,
+      "loss": 0.2302,
+      "step": 905
+    },
+    {
+      "epoch": 2.876190476190476,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.7035413403073994e-05,
+      "loss": 0.2803,
+      "step": 906
+    },
+    {
+      "epoch": 2.8793650793650793,
+      "grad_norm": 0.546875,
+      "learning_rate": 3.6840360653767714e-05,
+      "loss": 0.208,
+      "step": 907
+    },
+    {
+      "epoch": 2.8825396825396825,
+      "grad_norm": 0.8203125,
+      "learning_rate": 3.66457068550703e-05,
+      "loss": 0.2348,
+      "step": 908
+    },
+    {
+      "epoch": 2.8857142857142857,
+      "grad_norm": 0.64453125,
+      "learning_rate": 3.645145323652094e-05,
+      "loss": 0.2462,
+      "step": 909
+    },
+    {
+      "epoch": 2.888888888888889,
+      "grad_norm": 0.7734375,
+      "learning_rate": 3.6257601025131026e-05,
+      "loss": 0.2098,
+      "step": 910
+    },
+    {
+      "epoch": 2.892063492063492,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.6064151445376516e-05,
+      "loss": 0.4087,
+      "step": 911
+    },
+    {
+      "epoch": 2.895238095238095,
+      "grad_norm": 0.80078125,
+      "learning_rate": 3.587110571919011e-05,
+      "loss": 0.2782,
+      "step": 912
+    },
+    {
+      "epoch": 2.8984126984126983,
+      "grad_norm": 0.69921875,
+      "learning_rate": 3.5678465065953494e-05,
+      "loss": 0.2174,
+      "step": 913
+    },
+    {
+      "epoch": 2.9015873015873015,
+      "grad_norm": 0.7265625,
+      "learning_rate": 3.5486230702489764e-05,
+      "loss": 0.2424,
+      "step": 914
+    },
+    {
+      "epoch": 2.9047619047619047,
+      "grad_norm": 1.4609375,
+      "learning_rate": 3.52944038430556e-05,
+      "loss": 0.3647,
+      "step": 915
+    },
+    {
+      "epoch": 2.907936507936508,
+      "grad_norm": 0.87109375,
+      "learning_rate": 3.510298569933371e-05,
+      "loss": 0.3122,
+      "step": 916
+    },
+    {
+      "epoch": 2.911111111111111,
+      "grad_norm": 1.1953125,
+      "learning_rate": 3.491197748042515e-05,
+      "loss": 0.3585,
+      "step": 917
+    },
+    {
+      "epoch": 2.914285714285714,
+      "grad_norm": 0.86328125,
+      "learning_rate": 3.472138039284163e-05,
+      "loss": 0.2642,
+      "step": 918
+    },
+    {
+      "epoch": 2.9174603174603173,
+      "grad_norm": 1.2265625,
+      "learning_rate": 3.453119564049797e-05,
+      "loss": 0.3574,
+      "step": 919
+    },
+    {
+      "epoch": 2.9206349206349205,
+      "grad_norm": 0.69140625,
+      "learning_rate": 3.4341424424704375e-05,
+      "loss": 0.2424,
+      "step": 920
+    },
+    {
+      "epoch": 2.923809523809524,
+      "grad_norm": 0.68359375,
+      "learning_rate": 3.415206794415901e-05,
+      "loss": 0.249,
+      "step": 921
+    },
+    {
+      "epoch": 2.9269841269841272,
+      "grad_norm": 0.671875,
+      "learning_rate": 3.396312739494032e-05,
+      "loss": 0.278,
+      "step": 922
+    },
+    {
+      "epoch": 2.9301587301587304,
+      "grad_norm": 0.6171875,
+      "learning_rate": 3.377460397049951e-05,
+      "loss": 0.2035,
+      "step": 923
+    },
+    {
+      "epoch": 2.9333333333333336,
+      "grad_norm": 0.9609375,
+      "learning_rate": 3.3586498861653e-05,
+      "loss": 0.3572,
+      "step": 924
+    },
+    {
+      "epoch": 2.9365079365079367,
+      "grad_norm": 1.0703125,
+      "learning_rate": 3.339881325657484e-05,
+      "loss": 0.3226,
+      "step": 925
+    },
+    {
+      "epoch": 2.93968253968254,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.3211548340789364e-05,
+      "loss": 0.2921,
+      "step": 926
+    },
+    {
+      "epoch": 2.942857142857143,
+      "grad_norm": 0.94921875,
+      "learning_rate": 3.302470529716354e-05,
+      "loss": 0.303,
+      "step": 927
+    },
+    {
+      "epoch": 2.9460317460317462,
+      "grad_norm": 1.03125,
+      "learning_rate": 3.2838285305899585e-05,
+      "loss": 0.2829,
+      "step": 928
+    },
+    {
+      "epoch": 2.9492063492063494,
+      "grad_norm": 0.92578125,
+      "learning_rate": 3.2652289544527506e-05,
+      "loss": 0.2285,
+      "step": 929
+    },
+    {
+      "epoch": 2.9523809523809526,
+      "grad_norm": 1.2265625,
+      "learning_rate": 3.246671918789755e-05,
+      "loss": 0.2935,
+      "step": 930
+    },
+    {
+      "epoch": 2.9555555555555557,
+      "grad_norm": 1.078125,
+      "learning_rate": 3.2281575408173005e-05,
+      "loss": 0.2978,
+      "step": 931
+    },
+    {
+      "epoch": 2.958730158730159,
+      "grad_norm": 0.7578125,
+      "learning_rate": 3.2096859374822585e-05,
+      "loss": 0.2696,
+      "step": 932
+    },
+    {
+      "epoch": 2.961904761904762,
+      "grad_norm": 0.70703125,
+      "learning_rate": 3.1912572254613184e-05,
+      "loss": 0.2502,
+      "step": 933
+    },
+    {
+      "epoch": 2.965079365079365,
+      "grad_norm": 1.515625,
+      "learning_rate": 3.172871521160236e-05,
+      "loss": 0.3108,
+      "step": 934
+    },
+    {
+      "epoch": 2.9682539682539684,
+      "grad_norm": 0.83203125,
+      "learning_rate": 3.154528940713113e-05,
+      "loss": 0.319,
+      "step": 935
+    },
+    {
+      "epoch": 2.9714285714285715,
+      "grad_norm": 0.66015625,
+      "learning_rate": 3.136229599981661e-05,
+      "loss": 0.2346,
+      "step": 936
+    },
+    {
+      "epoch": 2.9746031746031747,
+      "grad_norm": 0.80078125,
+      "learning_rate": 3.117973614554456e-05,
+      "loss": 0.2647,
+      "step": 937
+    },
+    {
+      "epoch": 2.977777777777778,
+      "grad_norm": 0.81640625,
+      "learning_rate": 3.099761099746229e-05,
+      "loss": 0.2988,
+      "step": 938
+    },
+    {
+      "epoch": 2.980952380952381,
+      "grad_norm": 1.0546875,
+      "learning_rate": 3.0815921705971165e-05,
+      "loss": 0.3171,
+      "step": 939
+    },
+    {
+      "epoch": 2.984126984126984,
+      "grad_norm": 0.5859375,
+      "learning_rate": 3.063466941871952e-05,
+      "loss": 0.2036,
+      "step": 940
+    },
+    {
+      "epoch": 2.9873015873015873,
+      "grad_norm": 0.78125,
+      "learning_rate": 3.0453855280595324e-05,
+      "loss": 0.2565,
+      "step": 941
+    },
+    {
+      "epoch": 2.9904761904761905,
+      "grad_norm": 0.7421875,
+      "learning_rate": 3.0273480433718927e-05,
+      "loss": 0.2354,
+      "step": 942
+    },
+    {
+      "epoch": 2.9936507936507937,
+      "grad_norm": 0.78515625,
+      "learning_rate": 3.0093546017435938e-05,
+      "loss": 0.2725,
+      "step": 943
+    },
+    {
+      "epoch": 2.996825396825397,
+      "grad_norm": 0.8125,
+      "learning_rate": 2.991405316830985e-05,
+      "loss": 0.2601,
+      "step": 944
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.70703125,
+      "learning_rate": 2.9735003020115092e-05,
+      "loss": 0.2092,
+      "step": 945
+    },
+    {
+      "epoch": 3.003174603174603,
+      "grad_norm": 0.455078125,
+      "learning_rate": 2.9556396703829704e-05,
+      "loss": 0.1002,
+      "step": 946
+    },
+    {
+      "epoch": 3.0063492063492063,
+      "grad_norm": 0.453125,
+      "learning_rate": 2.937823534762827e-05,
+      "loss": 0.1044,
+      "step": 947
+    },
+    {
+      "epoch": 3.0095238095238095,
+      "grad_norm": 0.59375,
+      "learning_rate": 2.920052007687475e-05,
+      "loss": 0.1164,
+      "step": 948
+    },
+    {
+      "epoch": 3.0126984126984127,
+      "grad_norm": 0.388671875,
+      "learning_rate": 2.9023252014115332e-05,
+      "loss": 0.0978,
+      "step": 949
+    },
+    {
+      "epoch": 3.015873015873016,
+      "grad_norm": 0.80859375,
+      "learning_rate": 2.8846432279071467e-05,
+      "loss": 0.1222,
+      "step": 950
+    },
+    {
+      "epoch": 3.019047619047619,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.8670061988632714e-05,
+      "loss": 0.0999,
+      "step": 951
+    },
+    {
+      "epoch": 3.022222222222222,
+      "grad_norm": 0.5625,
+      "learning_rate": 2.8494142256849677e-05,
+      "loss": 0.1184,
+      "step": 952
+    },
+    {
+      "epoch": 3.0253968253968253,
+      "grad_norm": 0.44140625,
+      "learning_rate": 2.831867419492703e-05,
+      "loss": 0.103,
+      "step": 953
+    },
+    {
+      "epoch": 3.0285714285714285,
+      "grad_norm": 0.474609375,
+      "learning_rate": 2.814365891121634e-05,
+      "loss": 0.0872,
+      "step": 954
+    },
+    {
+      "epoch": 3.0317460317460316,
+      "grad_norm": 0.41015625,
+      "learning_rate": 2.7969097511209308e-05,
+      "loss": 0.1049,
+      "step": 955
+    },
+    {
+      "epoch": 3.034920634920635,
+      "grad_norm": 0.74609375,
+      "learning_rate": 2.77949910975306e-05,
+      "loss": 0.1563,
+      "step": 956
+    },
+    {
+      "epoch": 3.038095238095238,
+      "grad_norm": 0.671875,
+      "learning_rate": 2.762134076993094e-05,
+      "loss": 0.1169,
+      "step": 957
+    },
+    {
+      "epoch": 3.041269841269841,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.7448147625280217e-05,
+      "loss": 0.1103,
+      "step": 958
+    },
+    {
+      "epoch": 3.0444444444444443,
+      "grad_norm": 0.47265625,
+      "learning_rate": 2.7275412757560416e-05,
+      "loss": 0.1004,
+      "step": 959
+    },
+    {
+      "epoch": 3.0476190476190474,
+      "grad_norm": 0.482421875,
+      "learning_rate": 2.7103137257858868e-05,
+      "loss": 0.1002,
+      "step": 960
+    },
+    {
+      "epoch": 3.0507936507936506,
+      "grad_norm": 0.6875,
+      "learning_rate": 2.693132221436122e-05,
+      "loss": 0.1526,
+      "step": 961
+    },
+    {
+      "epoch": 3.0539682539682538,
+      "grad_norm": 0.48046875,
+      "learning_rate": 2.6759968712344697e-05,
+      "loss": 0.0948,
+      "step": 962
+    },
+    {
+      "epoch": 3.057142857142857,
+      "grad_norm": 0.60546875,
+      "learning_rate": 2.6589077834171172e-05,
+      "loss": 0.12,
+      "step": 963
+    },
+    {
+      "epoch": 3.06031746031746,
+      "grad_norm": 0.439453125,
+      "learning_rate": 2.6418650659280253e-05,
+      "loss": 0.1093,
+      "step": 964
+    },
+    {
+      "epoch": 3.0634920634920633,
+      "grad_norm": 0.396484375,
+      "learning_rate": 2.624868826418262e-05,
+      "loss": 0.098,
+      "step": 965
+    },
+    {
+      "epoch": 3.066666666666667,
+      "grad_norm": 0.546875,
+      "learning_rate": 2.6079191722453155e-05,
+      "loss": 0.1244,
+      "step": 966
+    },
+    {
+      "epoch": 3.06984126984127,
+      "grad_norm": 0.69140625,
+      "learning_rate": 2.5910162104724146e-05,
+      "loss": 0.1795,
+      "step": 967
+    },
+    {
+      "epoch": 3.073015873015873,
+      "grad_norm": 0.51171875,
+      "learning_rate": 2.5741600478678472e-05,
+      "loss": 0.1243,
+      "step": 968
+    },
+    {
+      "epoch": 3.0761904761904764,
+      "grad_norm": 0.455078125,
+      "learning_rate": 2.5573507909043014e-05,
+      "loss": 0.0941,
+      "step": 969
+    },
+    {
+      "epoch": 3.0793650793650795,
+      "grad_norm": 0.474609375,
+      "learning_rate": 2.540588545758179e-05,
+      "loss": 0.0931,
+      "step": 970
+    },
+    {
+      "epoch": 3.0825396825396827,
+      "grad_norm": 0.78125,
+      "learning_rate": 2.5238734183089308e-05,
+      "loss": 0.1253,
+      "step": 971
+    },
+    {
+      "epoch": 3.085714285714286,
+      "grad_norm": 0.56640625,
+      "learning_rate": 2.5072055141383877e-05,
+      "loss": 0.1126,
+      "step": 972
+    },
+    {
+      "epoch": 3.088888888888889,
+      "grad_norm": 0.49609375,
+      "learning_rate": 2.4905849385300883e-05,
+      "loss": 0.102,
+      "step": 973
+    },
+    {
+      "epoch": 3.092063492063492,
+      "grad_norm": 0.515625,
+      "learning_rate": 2.4740117964686217e-05,
+      "loss": 0.1169,
+      "step": 974
+    },
+    {
+      "epoch": 3.0952380952380953,
+      "grad_norm": 0.5234375,
+      "learning_rate": 2.4574861926389615e-05,
+      "loss": 0.1015,
+      "step": 975
+    },
+    {
+      "epoch": 3.0984126984126985,
+      "grad_norm": 0.4609375,
+      "learning_rate": 2.4410082314258032e-05,
+      "loss": 0.1101,
+      "step": 976
+    },
+    {
+      "epoch": 3.1015873015873017,
+      "grad_norm": 0.625,
+      "learning_rate": 2.4245780169129084e-05,
+      "loss": 0.1352,
+      "step": 977
+    },
+    {
+      "epoch": 3.104761904761905,
+      "grad_norm": 0.439453125,
+      "learning_rate": 2.4081956528824357e-05,
+      "loss": 0.0936,
+      "step": 978
+    },
+    {
+      "epoch": 3.107936507936508,
+      "grad_norm": 0.4375,
+      "learning_rate": 2.3918612428143016e-05,
+      "loss": 0.0993,
+      "step": 979
+    },
+    {
+      "epoch": 3.111111111111111,
+      "grad_norm": 0.466796875,
+      "learning_rate": 2.37557488988552e-05,
+      "loss": 0.0915,
+      "step": 980
+    },
+    {
+      "epoch": 3.1142857142857143,
+      "grad_norm": 0.76953125,
+      "learning_rate": 2.359336696969546e-05,
+      "loss": 0.1211,
+      "step": 981
+    },
+    {
+      "epoch": 3.1174603174603175,
+      "grad_norm": 0.49609375,
+      "learning_rate": 2.3431467666356355e-05,
+      "loss": 0.0849,
+      "step": 982
+    },
+    {
+      "epoch": 3.1206349206349207,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.32700520114818e-05,
+      "loss": 0.0958,
+      "step": 983
+    },
+    {
+      "epoch": 3.123809523809524,
+      "grad_norm": 0.7265625,
+      "learning_rate": 2.3109121024660872e-05,
+      "loss": 0.1338,
+      "step": 984
+    },
+    {
+      "epoch": 3.126984126984127,
+      "grad_norm": 0.8125,
+      "learning_rate": 2.2948675722421086e-05,
+      "loss": 0.1311,
+      "step": 985
+    },
+    {
+      "epoch": 3.13015873015873,
+      "grad_norm": 1.1796875,
+      "learning_rate": 2.2788717118222202e-05,
+      "loss": 0.1809,
+      "step": 986
+    },
+    {
+      "epoch": 3.1333333333333333,
+      "grad_norm": 0.4296875,
+      "learning_rate": 2.262924622244973e-05,
+      "loss": 0.1059,
+      "step": 987
+    },
+    {
+      "epoch": 3.1365079365079365,
+      "grad_norm": 0.60546875,
+      "learning_rate": 2.247026404240845e-05,
+      "loss": 0.0934,
+      "step": 988
+    },
+    {
+      "epoch": 3.1396825396825396,
+      "grad_norm": 0.58203125,
+      "learning_rate": 2.2311771582316255e-05,
+      "loss": 0.1327,
+      "step": 989
+    },
+    {
+      "epoch": 3.142857142857143,
+      "grad_norm": 0.46875,
+      "learning_rate": 2.2153769843297667e-05,
+      "loss": 0.0972,
+      "step": 990
+    },
+    {
+      "epoch": 3.146031746031746,
+      "grad_norm": 0.60546875,
+      "learning_rate": 2.199625982337753e-05,
+      "loss": 0.1368,
+      "step": 991
+    },
+    {
+      "epoch": 3.149206349206349,
+      "grad_norm": 0.478515625,
+      "learning_rate": 2.1839242517474766e-05,
+      "loss": 0.1167,
+      "step": 992
+    },
+    {
+      "epoch": 3.1523809523809523,
+      "grad_norm": 0.4765625,
+      "learning_rate": 2.168271891739597e-05,
+      "loss": 0.0954,
+      "step": 993
+    },
+    {
+      "epoch": 3.1555555555555554,
+      "grad_norm": 0.3984375,
+      "learning_rate": 2.1526690011829297e-05,
+      "loss": 0.0778,
+      "step": 994
+    },
+    {
+      "epoch": 3.1587301587301586,
+      "grad_norm": 0.453125,
+      "learning_rate": 2.137115678633811e-05,
+      "loss": 0.0948,
+      "step": 995
+    },
+    {
+      "epoch": 3.1619047619047618,
+      "grad_norm": 0.62890625,
+      "learning_rate": 2.121612022335482e-05,
+      "loss": 0.1113,
+      "step": 996
+    },
+    {
+      "epoch": 3.165079365079365,
+      "grad_norm": 0.4453125,
+      "learning_rate": 2.106158130217456e-05,
+      "loss": 0.1037,
+      "step": 997
+    },
+    {
+      "epoch": 3.168253968253968,
+      "grad_norm": 0.515625,
+      "learning_rate": 2.0907540998949172e-05,
+      "loss": 0.1126,
+      "step": 998
+    },
+    {
+      "epoch": 3.1714285714285713,
+      "grad_norm": 0.5859375,
+      "learning_rate": 2.0754000286680942e-05,
+      "loss": 0.1152,
+      "step": 999
+    },
+    {
+      "epoch": 3.1746031746031744,
+      "grad_norm": 0.453125,
+      "learning_rate": 2.0600960135216462e-05,
+      "loss": 0.1002,
+      "step": 1000
+    },
+    {
+      "epoch": 3.1777777777777776,
+      "grad_norm": 0.62109375,
+      "learning_rate": 2.044842151124051e-05,
+      "loss": 0.1276,
+      "step": 1001
+    },
+    {
+      "epoch": 3.1809523809523808,
+      "grad_norm": 0.462890625,
+      "learning_rate": 2.02963853782699e-05,
+      "loss": 0.0921,
+      "step": 1002
+    },
+    {
+      "epoch": 3.1841269841269844,
+      "grad_norm": 0.349609375,
+      "learning_rate": 2.014485269664751e-05,
+      "loss": 0.0751,
+      "step": 1003
+    },
+    {
+      "epoch": 3.1873015873015875,
+      "grad_norm": 0.6328125,
+      "learning_rate": 1.99938244235361e-05,
+      "loss": 0.0987,
+      "step": 1004
+    },
+    {
+      "epoch": 3.1904761904761907,
+      "grad_norm": 0.484375,
+      "learning_rate": 1.9843301512912327e-05,
+      "loss": 0.091,
+      "step": 1005
+    },
+    {
+      "epoch": 3.193650793650794,
+      "grad_norm": 0.5,
+      "learning_rate": 1.9693284915560738e-05,
+      "loss": 0.1198,
+      "step": 1006
+    },
+    {
+      "epoch": 3.196825396825397,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.9543775579067636e-05,
+      "loss": 0.1382,
+      "step": 1007
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.93947744478153e-05,
+      "loss": 0.1328,
+      "step": 1008
+    },
+    {
+      "epoch": 3.2031746031746033,
+      "grad_norm": 0.470703125,
+      "learning_rate": 1.9246282462975808e-05,
+      "loss": 0.0925,
+      "step": 1009
+    },
+    {
+      "epoch": 3.2063492063492065,
+      "grad_norm": 0.515625,
+      "learning_rate": 1.9098300562505266e-05,
+      "loss": 0.1051,
+      "step": 1010
+    },
+    {
+      "epoch": 3.2095238095238097,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.8950829681137805e-05,
+      "loss": 0.1156,
+      "step": 1011
+    },
+    {
+      "epoch": 3.212698412698413,
+      "grad_norm": 0.515625,
+      "learning_rate": 1.8803870750379603e-05,
+      "loss": 0.0791,
+      "step": 1012
+    },
+    {
+      "epoch": 3.215873015873016,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.865742469850318e-05,
+      "loss": 0.1502,
+      "step": 1013
+    },
+    {
+      "epoch": 3.219047619047619,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.851149245054138e-05,
+      "loss": 0.0906,
+      "step": 1014
+    },
+    {
+      "epoch": 3.2222222222222223,
+      "grad_norm": 0.482421875,
+      "learning_rate": 1.8366074928281607e-05,
+      "loss": 0.0928,
+      "step": 1015
+    },
+    {
+      "epoch": 3.2253968253968255,
+      "grad_norm": 0.474609375,
+      "learning_rate": 1.8221173050259978e-05,
+      "loss": 0.1086,
+      "step": 1016
+    },
+    {
+      "epoch": 3.2285714285714286,
+      "grad_norm": 0.5390625,
+      "learning_rate": 1.807678773175545e-05,
+      "loss": 0.0807,
+      "step": 1017
+    },
+    {
+      "epoch": 3.231746031746032,
+      "grad_norm": 0.48828125,
+      "learning_rate": 1.7932919884784217e-05,
+      "loss": 0.0955,
+      "step": 1018
+    },
+    {
+      "epoch": 3.234920634920635,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.778957041809377e-05,
+      "loss": 0.2458,
+      "step": 1019
+    },
+    {
+      "epoch": 3.238095238095238,
+      "grad_norm": 0.5,
+      "learning_rate": 1.7646740237157256e-05,
+      "loss": 0.096,
+      "step": 1020
+    },
+    {
+      "epoch": 3.2412698412698413,
+      "grad_norm": 0.546875,
+      "learning_rate": 1.7504430244167747e-05,
+      "loss": 0.1168,
+      "step": 1021
+    },
+    {
+      "epoch": 3.2444444444444445,
+      "grad_norm": 0.78515625,
+      "learning_rate": 1.736264133803246e-05,
+      "loss": 0.1291,
+      "step": 1022
+    },
+    {
+      "epoch": 3.2476190476190476,
+      "grad_norm": 0.5625,
+      "learning_rate": 1.722137441436721e-05,
+      "loss": 0.098,
+      "step": 1023
+    },
+    {
+      "epoch": 3.250793650793651,
+      "grad_norm": 0.58203125,
+      "learning_rate": 1.7080630365490657e-05,
+      "loss": 0.1123,
+      "step": 1024
+    },
+    {
+      "epoch": 3.253968253968254,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.6940410080418723e-05,
+      "loss": 0.1164,
+      "step": 1025
+    },
+    {
+      "epoch": 3.257142857142857,
+      "grad_norm": 0.83984375,
+      "learning_rate": 1.6800714444858944e-05,
+      "loss": 0.1172,
+      "step": 1026
+    },
+    {
+      "epoch": 3.2603174603174603,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.6661544341204848e-05,
+      "loss": 0.1097,
+      "step": 1027
+    },
+    {
+      "epoch": 3.2634920634920634,
+      "grad_norm": 0.5234375,
+      "learning_rate": 1.6522900648530448e-05,
+      "loss": 0.1042,
+      "step": 1028
+    },
+    {
+      "epoch": 3.2666666666666666,
+      "grad_norm": 0.466796875,
+      "learning_rate": 1.6384784242584672e-05,
+      "loss": 0.0955,
+      "step": 1029
+    },
+    {
+      "epoch": 3.2698412698412698,
+      "grad_norm": 0.51953125,
+      "learning_rate": 1.6247195995785837e-05,
+      "loss": 0.1189,
+      "step": 1030
+    },
+    {
+      "epoch": 3.273015873015873,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.611013677721602e-05,
+      "loss": 0.0899,
+      "step": 1031
+    },
+    {
+      "epoch": 3.276190476190476,
+      "grad_norm": 0.55078125,
+      "learning_rate": 1.59736074526158e-05,
+      "loss": 0.1073,
+      "step": 1032
+    },
+    {
+      "epoch": 3.2793650793650793,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.583760888437854e-05,
+      "loss": 0.0891,
+      "step": 1033
+    },
+    {
+      "epoch": 3.2825396825396824,
+      "grad_norm": 0.91796875,
+      "learning_rate": 1.5702141931545143e-05,
+      "loss": 0.1556,
+      "step": 1034
+    },
+    {
+      "epoch": 3.2857142857142856,
+      "grad_norm": 0.5625,
+      "learning_rate": 1.5567207449798515e-05,
+      "loss": 0.0979,
+      "step": 1035
+    },
+    {
+      "epoch": 3.2888888888888888,
+      "grad_norm": 0.58984375,
+      "learning_rate": 1.5432806291458147e-05,
+      "loss": 0.1072,
+      "step": 1036
+    },
+    {
+      "epoch": 3.292063492063492,
+      "grad_norm": 0.6484375,
+      "learning_rate": 1.5298939305474813e-05,
+      "loss": 0.1286,
+      "step": 1037
+    },
+    {
+      "epoch": 3.295238095238095,
+      "grad_norm": 0.4609375,
+      "learning_rate": 1.5165607337425158e-05,
+      "loss": 0.0939,
+      "step": 1038
+    },
+    {
+      "epoch": 3.2984126984126982,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.503281122950636e-05,
+      "loss": 0.1034,
+      "step": 1039
+    },
+    {
+      "epoch": 3.3015873015873014,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.4900551820530828e-05,
+      "loss": 0.089,
+      "step": 1040
+    },
+    {
+      "epoch": 3.3047619047619046,
+      "grad_norm": 0.38671875,
+      "learning_rate": 1.4768829945920836e-05,
+      "loss": 0.0749,
+      "step": 1041
+    },
+    {
+      "epoch": 3.3079365079365077,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.4637646437703335e-05,
+      "loss": 0.1119,
+      "step": 1042
+    },
+    {
+      "epoch": 3.311111111111111,
+      "grad_norm": 0.40625,
+      "learning_rate": 1.4507002124504676e-05,
+      "loss": 0.0845,
+      "step": 1043
+    },
+    {
+      "epoch": 3.314285714285714,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.4376897831545344e-05,
+      "loss": 0.0882,
+      "step": 1044
+    },
+    {
+      "epoch": 3.317460317460317,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.4247334380634792e-05,
+      "loss": 0.1066,
+      "step": 1045
+    },
+    {
+      "epoch": 3.320634920634921,
+      "grad_norm": 0.58203125,
+      "learning_rate": 1.4118312590166128e-05,
+      "loss": 0.1088,
+      "step": 1046
+    },
+    {
+      "epoch": 3.323809523809524,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.3989833275111141e-05,
+      "loss": 0.1014,
+      "step": 1047
+    },
+    {
+      "epoch": 3.326984126984127,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.386189724701501e-05,
+      "loss": 0.0895,
+      "step": 1048
+    },
+    {
+      "epoch": 3.3301587301587303,
+      "grad_norm": 0.77734375,
+      "learning_rate": 1.373450531399122e-05,
+      "loss": 0.1366,
+      "step": 1049
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.6171875,
+      "learning_rate": 1.3607658280716473e-05,
+      "loss": 0.1139,
+      "step": 1050
+    },
+    {
+      "epoch": 3.3365079365079366,
+      "grad_norm": 0.390625,
+      "learning_rate": 1.3481356948425538e-05,
+      "loss": 0.0691,
+      "step": 1051
+    },
+    {
+      "epoch": 3.33968253968254,
+      "grad_norm": 0.478515625,
+      "learning_rate": 1.3355602114906284e-05,
+      "loss": 0.0945,
+      "step": 1052
+    },
+    {
+      "epoch": 3.342857142857143,
+      "grad_norm": 0.546875,
+      "learning_rate": 1.3230394574494597e-05,
+      "loss": 0.0812,
+      "step": 1053
+    },
+    {
+      "epoch": 3.346031746031746,
+      "grad_norm": 0.57421875,
+      "learning_rate": 1.3105735118069374e-05,
+      "loss": 0.1045,
+      "step": 1054
+    },
+    {
+      "epoch": 3.3492063492063493,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.2981624533047432e-05,
+      "loss": 0.1459,
+      "step": 1055
+    },
+    {
+      "epoch": 3.3523809523809525,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.2858063603378723e-05,
+      "loss": 0.0883,
+      "step": 1056
+    },
+    {
+      "epoch": 3.3555555555555556,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.2735053109541239e-05,
+      "loss": 0.1342,
+      "step": 1057
+    },
+    {
+      "epoch": 3.358730158730159,
+      "grad_norm": 0.455078125,
+      "learning_rate": 1.261259382853608e-05,
+      "loss": 0.0764,
+      "step": 1058
+    },
+    {
+      "epoch": 3.361904761904762,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.2490686533882656e-05,
+      "loss": 0.0958,
+      "step": 1059
+    },
+    {
+      "epoch": 3.365079365079365,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.2369331995613665e-05,
+      "loss": 0.1046,
+      "step": 1060
+    },
+    {
+      "epoch": 3.3682539682539683,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.2248530980270324e-05,
+      "loss": 0.1131,
+      "step": 1061
+    },
+    {
+      "epoch": 3.3714285714285714,
+      "grad_norm": 0.56640625,
+      "learning_rate": 1.2128284250897525e-05,
+      "loss": 0.1046,
+      "step": 1062
+    },
+    {
+      "epoch": 3.3746031746031746,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.2008592567038956e-05,
+      "loss": 0.0859,
+      "step": 1063
+    },
+    {
+      "epoch": 3.3777777777777778,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.188945668473236e-05,
+      "loss": 0.1033,
+      "step": 1064
+    },
+    {
+      "epoch": 3.380952380952381,
+      "grad_norm": 0.61328125,
+      "learning_rate": 1.1770877356504683e-05,
+      "loss": 0.1091,
+      "step": 1065
+    },
+    {
+      "epoch": 3.384126984126984,
+      "grad_norm": 0.86328125,
+      "learning_rate": 1.1652855331367429e-05,
+      "loss": 0.1432,
+      "step": 1066
+    },
+    {
+      "epoch": 3.3873015873015873,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.1535391354811842e-05,
+      "loss": 0.1289,
+      "step": 1067
+    },
+    {
+      "epoch": 3.3904761904761904,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.1418486168804255e-05,
+      "loss": 0.1194,
+      "step": 1068
+    },
+    {
+      "epoch": 3.3936507936507936,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.1302140511781378e-05,
+      "loss": 0.1478,
+      "step": 1069
+    },
+    {
+      "epoch": 3.3968253968253967,
+      "grad_norm": 0.48046875,
+      "learning_rate": 1.1186355118645554e-05,
+      "loss": 0.1031,
+      "step": 1070
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.64453125,
+      "learning_rate": 1.1071130720760293e-05,
+      "loss": 0.1179,
+      "step": 1071
+    },
+    {
+      "epoch": 3.403174603174603,
+      "grad_norm": 0.451171875,
+      "learning_rate": 1.0956468045945512e-05,
+      "loss": 0.0934,
+      "step": 1072
+    },
+    {
+      "epoch": 3.4063492063492062,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0842367818472988e-05,
+      "loss": 0.0911,
+      "step": 1073
+    },
+    {
+      "epoch": 3.4095238095238094,
+      "grad_norm": 0.52734375,
+      "learning_rate": 1.0728830759061781e-05,
+      "loss": 0.11,
+      "step": 1074
+    },
+    {
+      "epoch": 3.4126984126984126,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0615857584873623e-05,
+      "loss": 0.1004,
+      "step": 1075
+    },
+    {
+      "epoch": 3.4158730158730157,
+      "grad_norm": 0.42578125,
+      "learning_rate": 1.0503449009508515e-05,
+      "loss": 0.0838,
+      "step": 1076
+    },
+    {
+      "epoch": 3.419047619047619,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.0391605743000099e-05,
+      "loss": 0.1376,
+      "step": 1077
+    },
+    {
+      "epoch": 3.422222222222222,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.0280328491811263e-05,
+      "loss": 0.1389,
+      "step": 1078
+    },
+    {
+      "epoch": 3.425396825396825,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.0169617958829558e-05,
+      "loss": 0.1028,
+      "step": 1079
+    },
+    {
+      "epoch": 3.4285714285714284,
+      "grad_norm": 0.5,
+      "learning_rate": 1.0059474843362892e-05,
+      "loss": 0.0926,
+      "step": 1080
+    },
+    {
+      "epoch": 3.431746031746032,
+      "grad_norm": 0.55078125,
+      "learning_rate": 9.949899841135079e-06,
+      "loss": 0.1149,
+      "step": 1081
+    },
+    {
+      "epoch": 3.434920634920635,
+      "grad_norm": 0.61328125,
+      "learning_rate": 9.840893644281346e-06,
+      "loss": 0.0804,
+      "step": 1082
+    },
+    {
+      "epoch": 3.4380952380952383,
+      "grad_norm": 0.62109375,
+      "learning_rate": 9.73245694134408e-06,
+      "loss": 0.1117,
+      "step": 1083
+    },
+    {
+      "epoch": 3.4412698412698415,
+      "grad_norm": 0.65234375,
+      "learning_rate": 9.624590417268476e-06,
+      "loss": 0.1255,
+      "step": 1084
+    },
+    {
+      "epoch": 3.4444444444444446,
+      "grad_norm": 0.486328125,
+      "learning_rate": 9.517294753398064e-06,
+      "loss": 0.0976,
+      "step": 1085
+    },
+    {
+      "epoch": 3.447619047619048,
+      "grad_norm": 0.408203125,
+      "learning_rate": 9.410570627470593e-06,
+      "loss": 0.0889,
+      "step": 1086
+    },
+    {
+      "epoch": 3.450793650793651,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.30441871361366e-06,
+      "loss": 0.0875,
+      "step": 1087
+    },
+    {
+      "epoch": 3.453968253968254,
+      "grad_norm": 0.404296875,
+      "learning_rate": 9.198839682340432e-06,
+      "loss": 0.0921,
+      "step": 1088
+    },
+    {
+      "epoch": 3.4571428571428573,
+      "grad_norm": 0.66796875,
+      "learning_rate": 9.09383420054547e-06,
+      "loss": 0.1246,
+      "step": 1089
+    },
+    {
+      "epoch": 3.4603174603174605,
+      "grad_norm": 0.6171875,
+      "learning_rate": 8.989402931500434e-06,
+      "loss": 0.1117,
+      "step": 1090
+    },
+    {
+      "epoch": 3.4634920634920636,
+      "grad_norm": 0.4609375,
+      "learning_rate": 8.885546534850009e-06,
+      "loss": 0.0878,
+      "step": 1091
+    },
+    {
+      "epoch": 3.466666666666667,
+      "grad_norm": 0.546875,
+      "learning_rate": 8.782265666607614e-06,
+      "loss": 0.0864,
+      "step": 1092
+    },
+    {
+      "epoch": 3.46984126984127,
+      "grad_norm": 0.486328125,
+      "learning_rate": 8.679560979151391e-06,
+      "loss": 0.097,
+      "step": 1093
+    },
+    {
+      "epoch": 3.473015873015873,
+      "grad_norm": 0.5234375,
+      "learning_rate": 8.577433121219902e-06,
+      "loss": 0.1139,
+      "step": 1094
+    },
+    {
+      "epoch": 3.4761904761904763,
+      "grad_norm": 0.48828125,
+      "learning_rate": 8.475882737908248e-06,
+      "loss": 0.0828,
+      "step": 1095
+    },
+    {
+      "epoch": 3.4793650793650794,
+      "grad_norm": 0.51171875,
+      "learning_rate": 8.37491047066382e-06,
+      "loss": 0.0871,
+      "step": 1096
+    },
+    {
+      "epoch": 3.4825396825396826,
+      "grad_norm": 0.66015625,
+      "learning_rate": 8.274516957282351e-06,
+      "loss": 0.0928,
+      "step": 1097
+    },
+    {
+      "epoch": 3.4857142857142858,
+      "grad_norm": 0.3828125,
+      "learning_rate": 8.174702831903836e-06,
+      "loss": 0.0756,
+      "step": 1098
+    },
+    {
+      "epoch": 3.488888888888889,
+      "grad_norm": 0.59375,
+      "learning_rate": 8.075468725008517e-06,
+      "loss": 0.0955,
+      "step": 1099
+    },
+    {
+      "epoch": 3.492063492063492,
+      "grad_norm": 0.490234375,
+      "learning_rate": 7.976815263412963e-06,
+      "loss": 0.089,
+      "step": 1100
+    },
+    {
+      "epoch": 3.4952380952380953,
+      "grad_norm": 0.55859375,
+      "learning_rate": 7.878743070266047e-06,
+      "loss": 0.0946,
+      "step": 1101
+    },
+    {
+      "epoch": 3.4984126984126984,
+      "grad_norm": 1.2109375,
+      "learning_rate": 7.781252765045077e-06,
+      "loss": 0.164,
+      "step": 1102
+    },
+    {
+      "epoch": 3.5015873015873016,
+      "grad_norm": 0.546875,
+      "learning_rate": 7.684344963551749e-06,
+      "loss": 0.1231,
+      "step": 1103
+    },
+    {
+      "epoch": 3.5047619047619047,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.588020277908426e-06,
+      "loss": 0.0732,
+      "step": 1104
+    },
+    {
+      "epoch": 3.507936507936508,
+      "grad_norm": 0.482421875,
+      "learning_rate": 7.492279316554207e-06,
+      "loss": 0.0861,
+      "step": 1105
+    },
+    {
+      "epoch": 3.511111111111111,
+      "grad_norm": 0.46875,
+      "learning_rate": 7.397122684240998e-06,
+      "loss": 0.0811,
+      "step": 1106
+    },
+    {
+      "epoch": 3.5142857142857142,
+      "grad_norm": 0.578125,
+      "learning_rate": 7.30255098202981e-06,
+      "loss": 0.1079,
+      "step": 1107
+    },
+    {
+      "epoch": 3.5174603174603174,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.208564807286944e-06,
+      "loss": 0.0938,
+      "step": 1108
+    },
+    {
+      "epoch": 3.5206349206349206,
+      "grad_norm": 0.5546875,
+      "learning_rate": 7.1151647536801255e-06,
+      "loss": 0.0923,
+      "step": 1109
+    },
+    {
+      "epoch": 3.5238095238095237,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.022351411174866e-06,
+      "loss": 0.0806,
+      "step": 1110
+    },
+    {
+      "epoch": 3.526984126984127,
+      "grad_norm": 0.63671875,
+      "learning_rate": 6.930125366030682e-06,
+      "loss": 0.1119,
+      "step": 1111
+    },
+    {
+      "epoch": 3.53015873015873,
+      "grad_norm": 0.5625,
+      "learning_rate": 6.838487200797405e-06,
+      "loss": 0.1161,
+      "step": 1112
+    },
+    {
+      "epoch": 3.533333333333333,
+      "grad_norm": 0.494140625,
+      "learning_rate": 6.747437494311504e-06,
+      "loss": 0.0769,
+      "step": 1113
+    },
+    {
+      "epoch": 3.5365079365079364,
+      "grad_norm": 0.65234375,
+      "learning_rate": 6.656976821692374e-06,
+      "loss": 0.1054,
+      "step": 1114
+    },
+    {
+      "epoch": 3.5396825396825395,
+      "grad_norm": 0.625,
+      "learning_rate": 6.5671057543387985e-06,
+      "loss": 0.1121,
+      "step": 1115
+    },
+    {
+      "epoch": 3.5428571428571427,
+      "grad_norm": 0.423828125,
+      "learning_rate": 6.477824859925297e-06,
+      "loss": 0.09,
+      "step": 1116
+    },
+    {
+      "epoch": 3.546031746031746,
+      "grad_norm": 0.482421875,
+      "learning_rate": 6.389134702398514e-06,
+      "loss": 0.0952,
+      "step": 1117
+    },
+    {
+      "epoch": 3.549206349206349,
+      "grad_norm": 0.435546875,
+      "learning_rate": 6.301035841973702e-06,
+      "loss": 0.0961,
+      "step": 1118
+    },
+    {
+      "epoch": 3.552380952380952,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.213528835131122e-06,
+      "loss": 0.0917,
+      "step": 1119
+    },
+    {
+      "epoch": 3.5555555555555554,
+      "grad_norm": 0.5078125,
+      "learning_rate": 6.126614234612593e-06,
+      "loss": 0.1067,
+      "step": 1120
+    },
+    {
+      "epoch": 3.5587301587301585,
+      "grad_norm": 0.58984375,
+      "learning_rate": 6.040292589417973e-06,
+      "loss": 0.1022,
+      "step": 1121
+    },
+    {
+      "epoch": 3.5619047619047617,
+      "grad_norm": 0.5234375,
+      "learning_rate": 5.954564444801702e-06,
+      "loss": 0.116,
+      "step": 1122
+    },
+    {
+      "epoch": 3.565079365079365,
+      "grad_norm": 0.51171875,
+      "learning_rate": 5.869430342269311e-06,
+      "loss": 0.0947,
+      "step": 1123
+    },
+    {
+      "epoch": 3.568253968253968,
+      "grad_norm": 0.384765625,
+      "learning_rate": 5.784890819574073e-06,
+      "loss": 0.0832,
+      "step": 1124
+    },
+    {
+      "epoch": 3.571428571428571,
+      "grad_norm": 0.67578125,
+      "learning_rate": 5.700946410713548e-06,
+      "loss": 0.1236,
+      "step": 1125
+    },
+    {
+      "epoch": 3.5746031746031743,
+      "grad_norm": 0.6015625,
+      "learning_rate": 5.6175976459262645e-06,
+      "loss": 0.109,
+      "step": 1126
+    },
+    {
+      "epoch": 3.5777777777777775,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.53484505168832e-06,
+      "loss": 0.15,
+      "step": 1127
+    },
+    {
+      "epoch": 3.580952380952381,
+      "grad_norm": 0.53515625,
+      "learning_rate": 5.452689150710055e-06,
+      "loss": 0.1036,
+      "step": 1128
+    },
+    {
+      "epoch": 3.5841269841269843,
+      "grad_norm": 0.5234375,
+      "learning_rate": 5.371130461932817e-06,
+      "loss": 0.0791,
+      "step": 1129
+    },
+    {
+      "epoch": 3.5873015873015874,
+      "grad_norm": 0.53515625,
+      "learning_rate": 5.290169500525577e-06,
+      "loss": 0.1205,
+      "step": 1130
+    },
+    {
+      "epoch": 3.5904761904761906,
+      "grad_norm": 0.52734375,
+      "learning_rate": 5.2098067778817986e-06,
+      "loss": 0.0927,
+      "step": 1131
+    },
+    {
+      "epoch": 3.5936507936507938,
+      "grad_norm": 0.66796875,
+      "learning_rate": 5.130042801616109e-06,
+      "loss": 0.1226,
+      "step": 1132
+    },
+    {
+      "epoch": 3.596825396825397,
+      "grad_norm": 0.6640625,
+      "learning_rate": 5.0508780755611165e-06,
+      "loss": 0.1115,
+      "step": 1133
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.50390625,
+      "learning_rate": 4.972313099764269e-06,
+      "loss": 0.1004,
+      "step": 1134
+    },
+    {
+      "epoch": 3.6031746031746033,
+      "grad_norm": 0.6640625,
+      "learning_rate": 4.8943483704846475e-06,
+      "loss": 0.1095,
+      "step": 1135
+    },
+    {
+      "epoch": 3.6063492063492064,
+      "grad_norm": 0.54296875,
+      "learning_rate": 4.816984380189859e-06,
+      "loss": 0.1076,
+      "step": 1136
+    },
+    {
+      "epoch": 3.6095238095238096,
+      "grad_norm": 0.7109375,
+      "learning_rate": 4.740221617552931e-06,
+      "loss": 0.1556,
+      "step": 1137
+    },
+    {
+      "epoch": 3.6126984126984127,
+      "grad_norm": 0.51953125,
+      "learning_rate": 4.664060567449169e-06,
+      "loss": 0.1089,
+      "step": 1138
+    },
+    {
+      "epoch": 3.615873015873016,
+      "grad_norm": 0.54296875,
+      "learning_rate": 4.5885017109531525e-06,
+      "loss": 0.0954,
+      "step": 1139
+    },
+    {
+      "epoch": 3.619047619047619,
+      "grad_norm": 0.390625,
+      "learning_rate": 4.513545525335705e-06,
+      "loss": 0.083,
+      "step": 1140
+    },
+    {
+      "epoch": 3.6222222222222222,
+      "grad_norm": 0.5234375,
+      "learning_rate": 4.439192484060817e-06,
+      "loss": 0.1072,
+      "step": 1141
+    },
+    {
+      "epoch": 3.6253968253968254,
+      "grad_norm": 0.796875,
+      "learning_rate": 4.3654430567827364e-06,
+      "loss": 0.1622,
+      "step": 1142
+    },
+    {
+      "epoch": 3.6285714285714286,
+      "grad_norm": 0.546875,
+      "learning_rate": 4.292297709342885e-06,
+      "loss": 0.0908,
+      "step": 1143
+    },
+    {
+      "epoch": 3.6317460317460317,
+      "grad_norm": 0.4765625,
+      "learning_rate": 4.2197569037670585e-06,
+      "loss": 0.1081,
+      "step": 1144
+    },
+    {
+      "epoch": 3.634920634920635,
+      "grad_norm": 0.55859375,
+      "learning_rate": 4.147821098262405e-06,
+      "loss": 0.0842,
+      "step": 1145
+    },
+    {
+      "epoch": 3.638095238095238,
+      "grad_norm": 0.6640625,
+      "learning_rate": 4.076490747214567e-06,
+      "loss": 0.1561,
+      "step": 1146
+    },
+    {
+      "epoch": 3.641269841269841,
+      "grad_norm": 0.44921875,
+      "learning_rate": 4.005766301184832e-06,
+      "loss": 0.0969,
+      "step": 1147
+    },
+    {
+      "epoch": 3.6444444444444444,
+      "grad_norm": 0.69921875,
+      "learning_rate": 3.935648206907194e-06,
+      "loss": 0.1109,
+      "step": 1148
+    },
+    {
+      "epoch": 3.6476190476190475,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.866136907285666e-06,
+      "loss": 0.1565,
+      "step": 1149
+    },
+    {
+      "epoch": 3.6507936507936507,
+      "grad_norm": 0.84765625,
+      "learning_rate": 3.797232841391407e-06,
+      "loss": 0.1445,
+      "step": 1150
+    },
+    {
+      "epoch": 3.653968253968254,
+      "grad_norm": 0.73828125,
+      "learning_rate": 3.7289364444599406e-06,
+      "loss": 0.1324,
+      "step": 1151
+    },
+    {
+      "epoch": 3.657142857142857,
+      "grad_norm": 0.486328125,
+      "learning_rate": 3.6612481478884186e-06,
+      "loss": 0.0973,
+      "step": 1152
+    },
+    {
+      "epoch": 3.66031746031746,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.5941683792329315e-06,
+      "loss": 0.1278,
+      "step": 1153
+    },
+    {
+      "epoch": 3.6634920634920634,
+      "grad_norm": 0.455078125,
+      "learning_rate": 3.527697562205723e-06,
+      "loss": 0.0862,
+      "step": 1154
+    },
+    {
+      "epoch": 3.6666666666666665,
+      "grad_norm": 0.796875,
+      "learning_rate": 3.461836116672612e-06,
+      "loss": 0.1487,
+      "step": 1155
+    },
+    {
+      "epoch": 3.6698412698412697,
+      "grad_norm": 0.4609375,
+      "learning_rate": 3.396584458650309e-06,
+      "loss": 0.1099,
+      "step": 1156
+    },
+    {
+      "epoch": 3.6730158730158733,
+      "grad_norm": 0.41796875,
+      "learning_rate": 3.3319430003037167e-06,
+      "loss": 0.0788,
+      "step": 1157
+    },
+    {
+      "epoch": 3.6761904761904765,
+      "grad_norm": 0.6640625,
+      "learning_rate": 3.267912149943431e-06,
+      "loss": 0.1309,
+      "step": 1158
+    },
+    {
+      "epoch": 3.6793650793650796,
+      "grad_norm": 0.6171875,
+      "learning_rate": 3.2044923120230995e-06,
+      "loss": 0.1,
+      "step": 1159
+    },
+    {
+      "epoch": 3.682539682539683,
+      "grad_norm": 0.78515625,
+      "learning_rate": 3.1416838871368924e-06,
+      "loss": 0.1553,
+      "step": 1160
+    },
+    {
+      "epoch": 3.685714285714286,
+      "grad_norm": 0.447265625,
+      "learning_rate": 3.0794872720169567e-06,
+      "loss": 0.0843,
+      "step": 1161
+    },
+    {
+      "epoch": 3.688888888888889,
+      "grad_norm": 1.0703125,
+      "learning_rate": 3.017902859530919e-06,
+      "loss": 0.1582,
+      "step": 1162
+    },
+    {
+      "epoch": 3.6920634920634923,
+      "grad_norm": 0.578125,
+      "learning_rate": 2.956931038679389e-06,
+      "loss": 0.091,
+      "step": 1163
+    },
+    {
+      "epoch": 3.6952380952380954,
+      "grad_norm": 0.453125,
+      "learning_rate": 2.8965721945935275e-06,
+      "loss": 0.0916,
+      "step": 1164
+    },
+    {
+      "epoch": 3.6984126984126986,
+      "grad_norm": 0.9375,
+      "learning_rate": 2.836826708532603e-06,
+      "loss": 0.15,
+      "step": 1165
+    },
+    {
+      "epoch": 3.7015873015873018,
+      "grad_norm": 0.84765625,
+      "learning_rate": 2.7776949578815935e-06,
+      "loss": 0.1631,
+      "step": 1166
+    },
+    {
+      "epoch": 3.704761904761905,
+      "grad_norm": 0.75390625,
+      "learning_rate": 2.7191773161487356e-06,
+      "loss": 0.1098,
+      "step": 1167
+    },
+    {
+      "epoch": 3.707936507936508,
+      "grad_norm": 0.5078125,
+      "learning_rate": 2.6612741529632777e-06,
+      "loss": 0.0951,
+      "step": 1168
+    },
+    {
+      "epoch": 3.7111111111111112,
+      "grad_norm": 0.67578125,
+      "learning_rate": 2.603985834073075e-06,
+      "loss": 0.1293,
+      "step": 1169
+    },
+    {
+      "epoch": 3.7142857142857144,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.5473127213422763e-06,
+      "loss": 0.0924,
+      "step": 1170
+    },
+    {
+      "epoch": 3.7174603174603176,
+      "grad_norm": 2.234375,
+      "learning_rate": 2.4912551727490852e-06,
+      "loss": 0.211,
+      "step": 1171
+    },
+    {
+      "epoch": 3.7206349206349207,
+      "grad_norm": 0.578125,
+      "learning_rate": 2.435813542383403e-06,
+      "loss": 0.1102,
+      "step": 1172
+    },
+    {
+      "epoch": 3.723809523809524,
+      "grad_norm": 0.58984375,
+      "learning_rate": 2.380988180444732e-06,
+      "loss": 0.1255,
+      "step": 1173
+    },
+    {
+      "epoch": 3.726984126984127,
+      "grad_norm": 4.625,
+      "learning_rate": 2.326779433239845e-06,
+      "loss": 0.3375,
+      "step": 1174
+    },
+    {
+      "epoch": 3.7301587301587302,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.273187643180652e-06,
+      "loss": 0.1155,
+      "step": 1175
+    },
+    {
+      "epoch": 3.7333333333333334,
+      "grad_norm": 0.7734375,
+      "learning_rate": 2.2202131487820242e-06,
+      "loss": 0.1265,
+      "step": 1176
+    },
+    {
+      "epoch": 3.7365079365079366,
+      "grad_norm": 0.65625,
+      "learning_rate": 2.167856284659675e-06,
+      "loss": 0.0954,
+      "step": 1177
+    },
+    {
+      "epoch": 3.7396825396825397,
+      "grad_norm": 0.578125,
+      "learning_rate": 2.116117381527982e-06,
+      "loss": 0.1043,
+      "step": 1178
+    },
+    {
+      "epoch": 3.742857142857143,
+      "grad_norm": 0.53515625,
+      "learning_rate": 2.064996766198002e-06,
+      "loss": 0.1156,
+      "step": 1179
+    },
+    {
+      "epoch": 3.746031746031746,
+      "grad_norm": 0.55078125,
+      "learning_rate": 2.014494761575314e-06,
+      "loss": 0.1268,
+      "step": 1180
+    },
+    {
+      "epoch": 3.749206349206349,
+      "grad_norm": 0.427734375,
+      "learning_rate": 1.9646116866580354e-06,
+      "loss": 0.0927,
+      "step": 1181
+    },
+    {
+      "epoch": 3.7523809523809524,
+      "grad_norm": 0.87109375,
+      "learning_rate": 1.9153478565347773e-06,
+      "loss": 0.153,
+      "step": 1182
+    },
+    {
+      "epoch": 3.7555555555555555,
+      "grad_norm": 0.5546875,
+      "learning_rate": 1.8667035823826562e-06,
+      "loss": 0.0855,
+      "step": 1183
+    },
+    {
+      "epoch": 3.7587301587301587,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.8186791714653427e-06,
+      "loss": 0.1177,
+      "step": 1184
+    },
+    {
+      "epoch": 3.761904761904762,
+      "grad_norm": 0.6015625,
+      "learning_rate": 1.771274927131139e-06,
+      "loss": 0.0914,
+      "step": 1185
+    },
+    {
+      "epoch": 3.765079365079365,
+      "grad_norm": 0.51953125,
+      "learning_rate": 1.7244911488109915e-06,
+      "loss": 0.089,
+      "step": 1186
+    },
+    {
+      "epoch": 3.768253968253968,
+      "grad_norm": 0.4921875,
+      "learning_rate": 1.6783281320166822e-06,
+      "loss": 0.0867,
+      "step": 1187
+    },
+    {
+      "epoch": 3.7714285714285714,
+      "grad_norm": 0.58984375,
+      "learning_rate": 1.6327861683389179e-06,
+      "loss": 0.1109,
+      "step": 1188
+    },
+    {
+      "epoch": 3.7746031746031745,
+      "grad_norm": 0.5546875,
+      "learning_rate": 1.5878655454454882e-06,
+      "loss": 0.1109,
+      "step": 1189
+    },
+    {
+      "epoch": 3.7777777777777777,
+      "grad_norm": 0.578125,
+      "learning_rate": 1.543566547079467e-06,
+      "loss": 0.0904,
+      "step": 1190
+    },
+    {
+      "epoch": 3.780952380952381,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.4998894530573903e-06,
+      "loss": 0.0911,
+      "step": 1191
+    },
+    {
+      "epoch": 3.784126984126984,
+      "grad_norm": 0.56640625,
+      "learning_rate": 1.4568345392675375e-06,
+      "loss": 0.0876,
+      "step": 1192
+    },
+    {
+      "epoch": 3.787301587301587,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.414402077668131e-06,
+      "loss": 0.1129,
+      "step": 1193
+    },
+    {
+      "epoch": 3.7904761904761903,
+      "grad_norm": 0.546875,
+      "learning_rate": 1.3725923362856607e-06,
+      "loss": 0.1097,
+      "step": 1194
+    },
+    {
+      "epoch": 3.7936507936507935,
+      "grad_norm": 0.455078125,
+      "learning_rate": 1.3314055792131964e-06,
+      "loss": 0.1017,
+      "step": 1195
+    },
+    {
+      "epoch": 3.7968253968253967,
+      "grad_norm": 0.6328125,
+      "learning_rate": 1.2908420666086551e-06,
+      "loss": 0.1053,
+      "step": 1196
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.2509020546932372e-06,
+      "loss": 0.0853,
+      "step": 1197
+    },
+    {
+      "epoch": 3.803174603174603,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.211585795749781e-06,
+      "loss": 0.0893,
+      "step": 1198
+    },
+    {
+      "epoch": 3.806349206349206,
+      "grad_norm": 0.416015625,
+      "learning_rate": 1.1728935381211226e-06,
+      "loss": 0.0846,
+      "step": 1199
+    },
+    {
+      "epoch": 3.8095238095238093,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.134825526208605e-06,
+      "loss": 0.0998,
+      "step": 1200
+    },
+    {
+      "epoch": 3.8126984126984125,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.0973820004704705e-06,
+      "loss": 0.0959,
+      "step": 1201
+    },
+    {
+      "epoch": 3.8158730158730156,
+      "grad_norm": 1.0,
+      "learning_rate": 1.0605631974203834e-06,
+      "loss": 0.1924,
+      "step": 1202
+    },
+    {
+      "epoch": 3.819047619047619,
+      "grad_norm": 0.416015625,
+      "learning_rate": 1.0243693496259088e-06,
+      "loss": 0.0904,
+      "step": 1203
+    },
+    {
+      "epoch": 3.822222222222222,
+      "grad_norm": 0.921875,
+      "learning_rate": 9.888006857070697e-07,
+      "loss": 0.1601,
+      "step": 1204
+    },
+    {
+      "epoch": 3.825396825396825,
+      "grad_norm": 0.427734375,
+      "learning_rate": 9.538574303348813e-07,
+      "loss": 0.0844,
+      "step": 1205
+    },
+    {
+      "epoch": 3.8285714285714287,
+      "grad_norm": 0.51953125,
+      "learning_rate": 9.195398042299297e-07,
+      "loss": 0.0921,
+      "step": 1206
+    },
+    {
+      "epoch": 3.831746031746032,
+      "grad_norm": 0.55859375,
+      "learning_rate": 8.858480241609845e-07,
+      "loss": 0.1068,
+      "step": 1207
+    },
+    {
+      "epoch": 3.834920634920635,
+      "grad_norm": 0.419921875,
+      "learning_rate": 8.527823029436443e-07,
+      "loss": 0.0909,
+      "step": 1208
+    },
+    {
+      "epoch": 3.8380952380952382,
+      "grad_norm": 0.546875,
+      "learning_rate": 8.20342849438982e-07,
+      "loss": 0.1072,
+      "step": 1209
+    },
+    {
+      "epoch": 3.8412698412698414,
+      "grad_norm": 0.46875,
+      "learning_rate": 7.885298685522235e-07,
+      "loss": 0.0916,
+      "step": 1210
+    },
+    {
+      "epoch": 3.8444444444444446,
+      "grad_norm": 0.427734375,
+      "learning_rate": 7.573435612314273e-07,
+      "loss": 0.0908,
+      "step": 1211
+    },
+    {
+      "epoch": 3.8476190476190477,
+      "grad_norm": 0.48046875,
+      "learning_rate": 7.267841244662621e-07,
+      "loss": 0.0906,
+      "step": 1212
+    },
+    {
+      "epoch": 3.850793650793651,
+      "grad_norm": 0.466796875,
+      "learning_rate": 6.968517512867534e-07,
+      "loss": 0.0954,
+      "step": 1213
+    },
+    {
+      "epoch": 3.853968253968254,
+      "grad_norm": 0.48046875,
+      "learning_rate": 6.675466307620281e-07,
+      "loss": 0.096,
+      "step": 1214
+    },
+    {
+      "epoch": 3.857142857142857,
+      "grad_norm": 0.60546875,
+      "learning_rate": 6.388689479991605e-07,
+      "loss": 0.1053,
+      "step": 1215
+    },
+    {
+      "epoch": 3.8603174603174604,
+      "grad_norm": 0.734375,
+      "learning_rate": 6.108188841419726e-07,
+      "loss": 0.095,
+      "step": 1216
+    },
+    {
+      "epoch": 3.8634920634920635,
+      "grad_norm": 0.78515625,
+      "learning_rate": 5.833966163699245e-07,
+      "loss": 0.147,
+      "step": 1217
+    },
+    {
+      "epoch": 3.8666666666666667,
+      "grad_norm": 0.7109375,
+      "learning_rate": 5.566023178969482e-07,
+      "loss": 0.1261,
+      "step": 1218
+    },
+    {
+      "epoch": 3.86984126984127,
+      "grad_norm": 0.68359375,
+      "learning_rate": 5.304361579704043e-07,
+      "loss": 0.1324,
+      "step": 1219
+    },
+    {
+      "epoch": 3.873015873015873,
+      "grad_norm": 0.56640625,
+      "learning_rate": 5.048983018699827e-07,
+      "loss": 0.1025,
+      "step": 1220
+    },
+    {
+      "epoch": 3.876190476190476,
+      "grad_norm": 0.408203125,
+      "learning_rate": 4.79988910906659e-07,
+      "loss": 0.0725,
+      "step": 1221
+    },
+    {
+      "epoch": 3.8793650793650793,
+      "grad_norm": 0.470703125,
+      "learning_rate": 4.557081424216847e-07,
+      "loss": 0.0841,
+      "step": 1222
+    },
+    {
+      "epoch": 3.8825396825396825,
+      "grad_norm": 0.5,
+      "learning_rate": 4.320561497855868e-07,
+      "loss": 0.1126,
+      "step": 1223
+    },
+    {
+      "epoch": 3.8857142857142857,
+      "grad_norm": 0.498046875,
+      "learning_rate": 4.090330823972033e-07,
+      "loss": 0.0919,
+      "step": 1224
+    },
+    {
+      "epoch": 3.888888888888889,
+      "grad_norm": 0.8984375,
+      "learning_rate": 3.866390856827495e-07,
+      "loss": 0.1602,
+      "step": 1225
+    },
+    {
+      "epoch": 3.892063492063492,
+      "grad_norm": 0.41796875,
+      "learning_rate": 3.648743010948863e-07,
+      "loss": 0.0733,
+      "step": 1226
+    },
+    {
+      "epoch": 3.895238095238095,
+      "grad_norm": 0.78125,
+      "learning_rate": 3.4373886611179797e-07,
+      "loss": 0.1612,
+      "step": 1227
+    },
+    {
+      "epoch": 3.8984126984126983,
+      "grad_norm": 0.412109375,
+      "learning_rate": 3.2323291423639324e-07,
+      "loss": 0.0883,
+      "step": 1228
+    },
+    {
+      "epoch": 3.9015873015873015,
+      "grad_norm": 0.79296875,
+      "learning_rate": 3.03356574995417e-07,
+      "loss": 0.1231,
+      "step": 1229
+    },
+    {
+      "epoch": 3.9047619047619047,
+      "grad_norm": 0.4921875,
+      "learning_rate": 2.841099739386066e-07,
+      "loss": 0.0962,
+      "step": 1230
+    },
+    {
+      "epoch": 3.907936507936508,
+      "grad_norm": 0.8984375,
+      "learning_rate": 2.654932326379811e-07,
+      "loss": 0.1264,
+      "step": 1231
+    },
+    {
+      "epoch": 3.911111111111111,
+      "grad_norm": 0.9140625,
+      "learning_rate": 2.4750646868697545e-07,
+      "loss": 0.1314,
+      "step": 1232
+    },
+    {
+      "epoch": 3.914285714285714,
+      "grad_norm": 0.443359375,
+      "learning_rate": 2.3014979569978556e-07,
+      "loss": 0.0788,
+      "step": 1233
+    },
+    {
+      "epoch": 3.9174603174603173,
+      "grad_norm": 0.490234375,
+      "learning_rate": 2.134233233105798e-07,
+      "loss": 0.0878,
+      "step": 1234
+    },
+    {
+      "epoch": 3.9206349206349205,
+      "grad_norm": 0.75,
+      "learning_rate": 1.973271571728441e-07,
+      "loss": 0.1184,
+      "step": 1235
+    },
+    {
+      "epoch": 3.923809523809524,
+      "grad_norm": 0.5546875,
+      "learning_rate": 1.81861398958727e-07,
+      "loss": 0.0998,
+      "step": 1236
+    },
+    {
+      "epoch": 3.9269841269841272,
+      "grad_norm": 0.5859375,
+      "learning_rate": 1.6702614635834002e-07,
+      "loss": 0.1158,
+      "step": 1237
+    },
+    {
+      "epoch": 3.9301587301587304,
+      "grad_norm": 0.5625,
+      "learning_rate": 1.5282149307922487e-07,
+      "loss": 0.1122,
+      "step": 1238
+    },
+    {
+      "epoch": 3.9333333333333336,
+      "grad_norm": 0.46484375,
+      "learning_rate": 1.3924752884568737e-07,
+      "loss": 0.0953,
+      "step": 1239
+    },
+    {
+      "epoch": 3.9365079365079367,
+      "grad_norm": 0.498046875,
+      "learning_rate": 1.2630433939825327e-07,
+      "loss": 0.1024,
+      "step": 1240
+    },
+    {
+      "epoch": 3.93968253968254,
+      "grad_norm": 0.455078125,
+      "learning_rate": 1.1399200649314657e-07,
+      "loss": 0.0845,
+      "step": 1241
+    },
+    {
+      "epoch": 3.942857142857143,
+      "grad_norm": 0.64453125,
+      "learning_rate": 1.0231060790174552e-07,
+      "loss": 0.1132,
+      "step": 1242
+    },
+    {
+      "epoch": 3.9460317460317462,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.126021741012736e-08,
+      "loss": 0.0773,
+      "step": 1243
+    },
+    {
+      "epoch": 3.9492063492063494,
+      "grad_norm": 0.375,
+      "learning_rate": 8.084090481855766e-08,
+      "loss": 0.0729,
+      "step": 1244
+    },
+    {
+      "epoch": 3.9523809523809526,
+      "grad_norm": 0.9921875,
+      "learning_rate": 7.105273594107953e-08,
+      "loss": 0.1897,
+      "step": 1245
+    },
+    {
+      "epoch": 3.9555555555555557,
+      "grad_norm": 0.51171875,
+      "learning_rate": 6.189577260508061e-08,
+      "loss": 0.1033,
+      "step": 1246
+    },
+    {
+      "epoch": 3.958730158730159,
+      "grad_norm": 0.60546875,
+      "learning_rate": 5.337007265091565e-08,
+      "loss": 0.1342,
+      "step": 1247
+    },
+    {
+      "epoch": 3.961904761904762,
+      "grad_norm": 0.6171875,
+      "learning_rate": 4.547568993155116e-08,
+      "loss": 0.0954,
+      "step": 1248
+    },
+    {
+      "epoch": 3.965079365079365,
+      "grad_norm": 0.53125,
+      "learning_rate": 3.82126743121991e-08,
+      "loss": 0.0992,
+      "step": 1249
+    },
+    {
+      "epoch": 3.9682539682539684,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.1581071670006015e-08,
+      "loss": 0.0747,
+      "step": 1250
+    },
+    {
+      "epoch": 3.9714285714285715,
+      "grad_norm": 0.75,
+      "learning_rate": 2.5580923893775423e-08,
+      "loss": 0.1148,
+      "step": 1251
+    },
+    {
+      "epoch": 3.9746031746031747,
+      "grad_norm": 0.5078125,
+      "learning_rate": 2.0212268883701422e-08,
+      "loss": 0.1182,
+      "step": 1252
+    },
+    {
+      "epoch": 3.977777777777778,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.5475140551146627e-08,
+      "loss": 0.1222,
+      "step": 1253
+    },
+    {
+      "epoch": 3.980952380952381,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.1369568818364596e-08,
+      "loss": 0.1154,
+      "step": 1254
+    },
+    {
+      "epoch": 3.984126984126984,
+      "grad_norm": 0.5703125,
+      "learning_rate": 7.895579618388827e-09,
+      "loss": 0.1106,
+      "step": 1255
+    },
+    {
+      "epoch": 3.9873015873015873,
+      "grad_norm": 0.419921875,
+      "learning_rate": 5.053194894821811e-09,
+      "loss": 0.0889,
+      "step": 1256
+    },
+    {
+      "epoch": 3.9904761904761905,
+      "grad_norm": 0.466796875,
+      "learning_rate": 2.8424326017018055e-09,
+      "loss": 0.0983,
+      "step": 1257
+    },
+    {
+      "epoch": 3.9936507936507937,
+      "grad_norm": 0.478515625,
+      "learning_rate": 1.2633067034251157e-09,
+      "loss": 0.0913,
+      "step": 1258
+    },
+    {
+      "epoch": 3.996825396825397,
+      "grad_norm": 0.95703125,
+      "learning_rate": 3.15827174590666e-10,
+      "loss": 0.1436,
+      "step": 1259
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0,
+      "loss": 0.101,
+      "step": 1260
+    }
+  ],
+  "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": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.068394448144302e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}