diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17530 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 2500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 198.46498107910156,
+      "learning_rate": 4e-05,
+      "loss": 13.9937,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 109.62651824951172,
+      "learning_rate": 8e-05,
+      "loss": 15.67,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 134.88104248046875,
+      "learning_rate": 0.00012,
+      "loss": 14.5824,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 152.51585388183594,
+      "learning_rate": 0.00016,
+      "loss": 10.3129,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 160.01882934570312,
+      "learning_rate": 0.0002,
+      "loss": 7.9769,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 91.6689682006836,
+      "learning_rate": 0.00024,
+      "loss": 5.9942,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 466.60211181640625,
+      "learning_rate": 0.00028000000000000003,
+      "loss": 7.6427,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 252.0084991455078,
+      "learning_rate": 0.00032,
+      "loss": 3.8709,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 50.1002311706543,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 2.3483,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 21.06104278564453,
+      "learning_rate": 0.0004,
+      "loss": 2.1463,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 22.925750732421875,
+      "learning_rate": 0.00044,
+      "loss": 1.5907,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 25.341272354125977,
+      "learning_rate": 0.00048,
+      "loss": 1.2098,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 25.70902442932129,
+      "learning_rate": 0.0005200000000000001,
+      "loss": 1.1089,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 20.201156616210938,
+      "learning_rate": 0.0005600000000000001,
+      "loss": 1.2093,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 17.095195770263672,
+      "learning_rate": 0.0006,
+      "loss": 1.1944,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 123.61097717285156,
+      "learning_rate": 0.00064,
+      "loss": 1.0811,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 98.7990493774414,
+      "learning_rate": 0.00068,
+      "loss": 1.4094,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 21.133268356323242,
+      "learning_rate": 0.0007199999999999999,
+      "loss": 1.1378,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 26.2519474029541,
+      "learning_rate": 0.00076,
+      "loss": 0.9975,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 38.6169548034668,
+      "learning_rate": 0.0008,
+      "loss": 1.2893,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 23.271162033081055,
+      "learning_rate": 0.00084,
+      "loss": 1.647,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.758883953094482,
+      "learning_rate": 0.00088,
+      "loss": 1.075,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.218874931335449,
+      "learning_rate": 0.00092,
+      "loss": 1.453,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.329564094543457,
+      "learning_rate": 0.00096,
+      "loss": 1.2021,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.698463439941406,
+      "learning_rate": 0.001,
+      "loss": 1.2338,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.821223735809326,
+      "learning_rate": 0.00099999959720016,
+      "loss": 0.784,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.575422763824463,
+      "learning_rate": 0.0009999983888012895,
+      "loss": 1.1515,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.32179069519043,
+      "learning_rate": 0.0009999963748053354,
+      "loss": 1.3096,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 13.78110408782959,
+      "learning_rate": 0.0009999935552155422,
+      "loss": 0.9014,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.031979560852051,
+      "learning_rate": 0.0009999899300364532,
+      "loss": 0.7677,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.106215953826904,
+      "learning_rate": 0.0009999854992739094,
+      "loss": 1.2908,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.8942975997924805,
+      "learning_rate": 0.000999980262935049,
+      "loss": 1.372,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.614978790283203,
+      "learning_rate": 0.0009999742210283098,
+      "loss": 1.4209,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.2029447555542,
+      "learning_rate": 0.000999967373563426,
+      "loss": 1.4461,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 20.205286026000977,
+      "learning_rate": 0.0009999597205514296,
+      "loss": 1.5293,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.320289611816406,
+      "learning_rate": 0.0009999512620046521,
+      "loss": 1.2611,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.132470607757568,
+      "learning_rate": 0.0009999419979367214,
+      "loss": 1.0019,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.694629192352295,
+      "learning_rate": 0.000999931928362564,
+      "loss": 0.9374,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.455288887023926,
+      "learning_rate": 0.0009999210532984039,
+      "loss": 0.9952,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.4660868644714355,
+      "learning_rate": 0.000999909372761763,
+      "loss": 0.9217,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 5.032826900482178,
+      "learning_rate": 0.0009998968867714609,
+      "loss": 0.9008,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.1872518062591553,
+      "learning_rate": 0.0009998835953476148,
+      "loss": 0.9994,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3714139461517334,
+      "learning_rate": 0.0009998694985116404,
+      "loss": 0.8307,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.928600788116455,
+      "learning_rate": 0.0009998545962862503,
+      "loss": 0.6185,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.19402551651001,
+      "learning_rate": 0.0009998388886954545,
+      "loss": 1.4364,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.967038869857788,
+      "learning_rate": 0.0009998223757645616,
+      "loss": 0.8799,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.072585582733154,
+      "learning_rate": 0.000999805057520177,
+      "loss": 0.7038,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.064746618270874,
+      "learning_rate": 0.000999786933990204,
+      "loss": 0.807,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3330047130584717,
+      "learning_rate": 0.0009997680052038433,
+      "loss": 0.9374,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.375366687774658,
+      "learning_rate": 0.0009997482711915926,
+      "loss": 1.0914,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 9.891619682312012,
+      "learning_rate": 0.0009997277319852475,
+      "loss": 1.422,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.3420145511627197,
+      "learning_rate": 0.0009997063876179006,
+      "loss": 0.8199,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.8788793087005615,
+      "learning_rate": 0.0009996842381239423,
+      "loss": 0.6645,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.8013806343078613,
+      "learning_rate": 0.0009996612835390594,
+      "loss": 1.1793,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.254058599472046,
+      "learning_rate": 0.0009996375239002368,
+      "loss": 0.8033,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.536509037017822,
+      "learning_rate": 0.0009996129592457557,
+      "loss": 0.788,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.6948862075805664,
+      "learning_rate": 0.0009995875896151945,
+      "loss": 0.7904,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2026126384735107,
+      "learning_rate": 0.0009995614150494292,
+      "loss": 0.8706,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.0636916160583496,
+      "learning_rate": 0.000999534435590632,
+      "loss": 0.7061,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.495323181152344,
+      "learning_rate": 0.0009995066512822719,
+      "loss": 0.5935,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 13.370635986328125,
+      "learning_rate": 0.0009994780621691156,
+      "loss": 0.8862,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 16.39713478088379,
+      "learning_rate": 0.0009994486682972253,
+      "loss": 1.3776,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.6794373989105225,
+      "learning_rate": 0.0009994184697139604,
+      "loss": 0.6426,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.0153560638427734,
+      "learning_rate": 0.0009993874664679773,
+      "loss": 0.9902,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 47.08489990234375,
+      "learning_rate": 0.000999355658609228,
+      "loss": 0.6822,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.915006399154663,
+      "learning_rate": 0.0009993230461889616,
+      "loss": 0.6011,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.178487777709961,
+      "learning_rate": 0.0009992896292597229,
+      "loss": 0.8264,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.772949695587158,
+      "learning_rate": 0.0009992554078753534,
+      "loss": 0.9408,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 8.095405578613281,
+      "learning_rate": 0.0009992203820909905,
+      "loss": 0.8567,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 7.8492207527160645,
+      "learning_rate": 0.0009991845519630679,
+      "loss": 0.8441,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.9621729850769043,
+      "learning_rate": 0.0009991479175493149,
+      "loss": 0.7266,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.1748576164245605,
+      "learning_rate": 0.000999110478908757,
+      "loss": 0.656,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.264678478240967,
+      "learning_rate": 0.000999072236101715,
+      "loss": 0.6701,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.9336034059524536,
+      "learning_rate": 0.0009990331891898058,
+      "loss": 0.7416,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.59828782081604,
+      "learning_rate": 0.0009989933382359422,
+      "loss": 0.7546,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 4.191848278045654,
+      "learning_rate": 0.0009989526833043317,
+      "loss": 0.6213,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.572232723236084,
+      "learning_rate": 0.0009989112244604772,
+      "loss": 0.6589,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.814861536026001,
+      "learning_rate": 0.0009988689617711777,
+      "loss": 0.6497,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7330659627914429,
+      "learning_rate": 0.0009988258953045263,
+      "loss": 0.7771,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.7193400859832764,
+      "learning_rate": 0.0009987820251299122,
+      "loss": 0.8773,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3554413318634033,
+      "learning_rate": 0.0009987373513180185,
+      "loss": 0.5742,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4223092794418335,
+      "learning_rate": 0.0009986918739408241,
+      "loss": 0.5818,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5654200315475464,
+      "learning_rate": 0.0009986455930716016,
+      "loss": 0.4539,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.5639121532440186,
+      "learning_rate": 0.0009985985087849192,
+      "loss": 0.676,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.9025020599365234,
+      "learning_rate": 0.0009985506211566387,
+      "loss": 0.7457,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.361051321029663,
+      "learning_rate": 0.0009985019302639168,
+      "loss": 0.7766,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.174298048019409,
+      "learning_rate": 0.0009984524361852042,
+      "loss": 0.7099,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.548457145690918,
+      "learning_rate": 0.0009984021390002458,
+      "loss": 0.6322,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3816312551498413,
+      "learning_rate": 0.0009983510387900803,
+      "loss": 0.3994,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.053450584411621,
+      "learning_rate": 0.0009982991356370402,
+      "loss": 0.8004,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.168210506439209,
+      "learning_rate": 0.0009982464296247522,
+      "loss": 0.5039,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.220595598220825,
+      "learning_rate": 0.0009981929208381357,
+      "loss": 0.7232,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2745871543884277,
+      "learning_rate": 0.0009981386093634045,
+      "loss": 0.5518,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4245800971984863,
+      "learning_rate": 0.000998083495288065,
+      "loss": 0.5812,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1687276363372803,
+      "learning_rate": 0.000998027578700917,
+      "loss": 0.5929,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3089672327041626,
+      "learning_rate": 0.000997970859692053,
+      "loss": 0.3611,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5346531867980957,
+      "learning_rate": 0.000997913338352859,
+      "loss": 0.427,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.028717517852783,
+      "learning_rate": 0.0009978550147760133,
+      "loss": 0.8449,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4955990314483643,
+      "learning_rate": 0.0009977958890554867,
+      "loss": 0.6179,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3464155197143555,
+      "learning_rate": 0.0009977359612865424,
+      "loss": 0.7478,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4723050594329834,
+      "learning_rate": 0.000997675231565736,
+      "loss": 0.6001,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.222213625907898,
+      "learning_rate": 0.0009976136999909156,
+      "loss": 0.5286,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6771528720855713,
+      "learning_rate": 0.0009975513666612204,
+      "loss": 0.7665,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2002547979354858,
+      "learning_rate": 0.0009974882316770822,
+      "loss": 0.7081,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1412129402160645,
+      "learning_rate": 0.0009974242951402235,
+      "loss": 0.6101,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0976482629776,
+      "learning_rate": 0.0009973595571536593,
+      "loss": 0.5244,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.140594959259033,
+      "learning_rate": 0.0009972940178216952,
+      "loss": 0.6231,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3024015426635742,
+      "learning_rate": 0.000997227677249928,
+      "loss": 0.7558,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.0104684829711914,
+      "learning_rate": 0.000997160535545246,
+      "loss": 0.4296,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 17.40903091430664,
+      "learning_rate": 0.0009970925928158272,
+      "loss": 0.702,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.216529130935669,
+      "learning_rate": 0.0009970238491711417,
+      "loss": 0.5934,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9890199899673462,
+      "learning_rate": 0.0009969543047219486,
+      "loss": 0.6204,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.5709807872772217,
+      "learning_rate": 0.0009968839595802983,
+      "loss": 0.7424,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1202000379562378,
+      "learning_rate": 0.0009968128138595305,
+      "loss": 0.6059,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1486153602600098,
+      "learning_rate": 0.0009967408676742752,
+      "loss": 0.4717,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2808343172073364,
+      "learning_rate": 0.000996668121140452,
+      "loss": 0.6058,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 7.744408130645752,
+      "learning_rate": 0.0009965945743752704,
+      "loss": 0.4634,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.918769359588623,
+      "learning_rate": 0.0009965202274972286,
+      "loss": 0.6454,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.6794342994689941,
+      "learning_rate": 0.0009964450806261144,
+      "loss": 0.7203,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.4869005680084229,
+      "learning_rate": 0.0009963691338830043,
+      "loss": 0.7561,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.8969289064407349,
+      "learning_rate": 0.0009962923873902637,
+      "loss": 0.4109,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0436840057373047,
+      "learning_rate": 0.0009962148412715462,
+      "loss": 0.4185,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.3000636100769043,
+      "learning_rate": 0.0009961364956517947,
+      "loss": 0.9118,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3613947629928589,
+      "learning_rate": 0.000996057350657239,
+      "loss": 0.6379,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3274770975112915,
+      "learning_rate": 0.0009959774064153978,
+      "loss": 0.7568,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.161910057067871,
+      "learning_rate": 0.000995896663055077,
+      "loss": 0.6747,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1567063331604004,
+      "learning_rate": 0.0009958151207063704,
+      "loss": 0.405,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.83669376373291,
+      "learning_rate": 0.0009957327795006587,
+      "loss": 0.6755,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1880184412002563,
+      "learning_rate": 0.0009956496395706106,
+      "loss": 0.7028,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.983546793460846,
+      "learning_rate": 0.0009955657010501807,
+      "loss": 0.4806,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2585489749908447,
+      "learning_rate": 0.0009954809640746105,
+      "loss": 0.3911,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.7685624361038208,
+      "learning_rate": 0.0009953954287804285,
+      "loss": 0.8569,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.4744935035705566,
+      "learning_rate": 0.000995309095305449,
+      "loss": 0.8051,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3931320905685425,
+      "learning_rate": 0.0009952219637887724,
+      "loss": 0.6439,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.5957046747207642,
+      "learning_rate": 0.0009951340343707852,
+      "loss": 0.9287,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.9483304023742676,
+      "learning_rate": 0.0009950453071931589,
+      "loss": 0.5833,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.8367655277252197,
+      "learning_rate": 0.0009949557823988506,
+      "loss": 0.7067,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1211509704589844,
+      "learning_rate": 0.000994865460132103,
+      "loss": 0.5949,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0638023614883423,
+      "learning_rate": 0.0009947743405384428,
+      "loss": 0.6229,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.4141712188720703,
+      "learning_rate": 0.0009946824237646824,
+      "loss": 0.5557,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9069592952728271,
+      "learning_rate": 0.0009945897099589173,
+      "loss": 0.4445,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1866590976715088,
+      "learning_rate": 0.0009944961992705288,
+      "loss": 0.5154,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.225667953491211,
+      "learning_rate": 0.0009944018918501805,
+      "loss": 0.5838,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9377739429473877,
+      "learning_rate": 0.000994306787849821,
+      "loss": 0.4541,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8814502358436584,
+      "learning_rate": 0.0009942108874226813,
+      "loss": 0.5176,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2999041080474854,
+      "learning_rate": 0.0009941141907232765,
+      "loss": 0.8467,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.313524603843689,
+      "learning_rate": 0.000994016697907404,
+      "loss": 0.6287,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.911753237247467,
+      "learning_rate": 0.0009939184091321445,
+      "loss": 0.5717,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1880959272384644,
+      "learning_rate": 0.0009938193245558607,
+      "loss": 0.6609,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0077195167541504,
+      "learning_rate": 0.0009937194443381972,
+      "loss": 0.4428,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.041214108467102,
+      "learning_rate": 0.0009936187686400814,
+      "loss": 0.4007,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9832210540771484,
+      "learning_rate": 0.0009935172976237219,
+      "loss": 0.5841,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.368139386177063,
+      "learning_rate": 0.0009934150314526084,
+      "loss": 0.6979,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9196540713310242,
+      "learning_rate": 0.0009933119702915125,
+      "loss": 0.427,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9226189255714417,
+      "learning_rate": 0.000993208114306486,
+      "loss": 0.4802,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.027590036392212,
+      "learning_rate": 0.0009931034636648617,
+      "loss": 0.6933,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.157545328140259,
+      "learning_rate": 0.0009929980185352525,
+      "loss": 0.8988,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2173960208892822,
+      "learning_rate": 0.0009928917790875517,
+      "loss": 0.5212,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.3881466388702393,
+      "learning_rate": 0.0009927847454929322,
+      "loss": 0.6278,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.3345468044281006,
+      "learning_rate": 0.0009926769179238466,
+      "loss": 0.5489,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2484301328659058,
+      "learning_rate": 0.0009925682965540262,
+      "loss": 0.6404,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.478821039199829,
+      "learning_rate": 0.0009924588815584821,
+      "loss": 0.5506,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2489778995513916,
+      "learning_rate": 0.0009923486731135034,
+      "loss": 0.7229,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1233000755310059,
+      "learning_rate": 0.000992237671396658,
+      "loss": 0.5407,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9513972997665405,
+      "learning_rate": 0.0009921258765867918,
+      "loss": 0.5216,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3649282455444336,
+      "learning_rate": 0.0009920132888640285,
+      "loss": 0.7023,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0220986604690552,
+      "learning_rate": 0.0009918999084097694,
+      "loss": 0.3746,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2041553258895874,
+      "learning_rate": 0.000991785735406693,
+      "loss": 0.4382,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0214067697525024,
+      "learning_rate": 0.0009916707700387546,
+      "loss": 0.4755,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.347096562385559,
+      "learning_rate": 0.0009915550124911866,
+      "loss": 0.4684,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8509796261787415,
+      "learning_rate": 0.0009914384629504972,
+      "loss": 0.4132,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2879726886749268,
+      "learning_rate": 0.0009913211216044714,
+      "loss": 0.831,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.55633544921875,
+      "learning_rate": 0.000991202988642169,
+      "loss": 0.6302,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1257086992263794,
+      "learning_rate": 0.0009910840642539261,
+      "loss": 0.4989,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1537691354751587,
+      "learning_rate": 0.0009909643486313534,
+      "loss": 0.4553,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0226002931594849,
+      "learning_rate": 0.0009908438419673367,
+      "loss": 0.2587,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.638160467147827,
+      "learning_rate": 0.0009907225444560361,
+      "loss": 0.5438,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1253334283828735,
+      "learning_rate": 0.0009906004562928864,
+      "loss": 0.5459,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2141050100326538,
+      "learning_rate": 0.0009904775776745957,
+      "loss": 0.5722,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9450463056564331,
+      "learning_rate": 0.0009903539087991462,
+      "loss": 0.4112,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1000924110412598,
+      "learning_rate": 0.0009902294498657929,
+      "loss": 0.6422,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0010532140731812,
+      "learning_rate": 0.000990104201075064,
+      "loss": 0.4105,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.442594528198242,
+      "learning_rate": 0.0009899781626287604,
+      "loss": 1.3386,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8499632477760315,
+      "learning_rate": 0.0009898513347299548,
+      "loss": 0.3433,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4625005722045898,
+      "learning_rate": 0.0009897237175829927,
+      "loss": 0.5328,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.2213704586029053,
+      "learning_rate": 0.0009895953113934904,
+      "loss": 0.3893,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9088588356971741,
+      "learning_rate": 0.000989466116368336,
+      "loss": 0.4363,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8356744050979614,
+      "learning_rate": 0.0009893361327156886,
+      "loss": 0.3792,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0905886888504028,
+      "learning_rate": 0.0009892053606449776,
+      "loss": 0.7355,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7066889405250549,
+      "learning_rate": 0.0009890738003669028,
+      "loss": 0.3642,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8947740793228149,
+      "learning_rate": 0.0009889414520934344,
+      "loss": 0.4379,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2436774969100952,
+      "learning_rate": 0.0009888083160378112,
+      "loss": 0.936,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7771862745285034,
+      "learning_rate": 0.0009886743924145426,
+      "loss": 0.4057,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7420421242713928,
+      "learning_rate": 0.000988539681439406,
+      "loss": 0.3971,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1913323402404785,
+      "learning_rate": 0.0009884041833294476,
+      "loss": 0.8226,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1120308637619019,
+      "learning_rate": 0.0009882678983029818,
+      "loss": 0.5881,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8594409823417664,
+      "learning_rate": 0.0009881308265795913,
+      "loss": 0.4841,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0268663167953491,
+      "learning_rate": 0.0009879929683801255,
+      "loss": 0.5372,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.670646071434021,
+      "learning_rate": 0.0009878543239267015,
+      "loss": 0.9652,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.127968430519104,
+      "learning_rate": 0.0009877148934427035,
+      "loss": 0.6472,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7133132815361023,
+      "learning_rate": 0.0009875746771527815,
+      "loss": 0.4247,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.5529571771621704,
+      "learning_rate": 0.0009874336752828522,
+      "loss": 0.4607,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9433883428573608,
+      "learning_rate": 0.0009872918880600974,
+      "loss": 0.4079,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9374203085899353,
+      "learning_rate": 0.0009871493157129648,
+      "loss": 0.4416,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0477722883224487,
+      "learning_rate": 0.0009870059584711668,
+      "loss": 0.6026,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7882351875305176,
+      "learning_rate": 0.0009868618165656804,
+      "loss": 0.4059,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.428831696510315,
+      "learning_rate": 0.0009867168902287472,
+      "loss": 0.5087,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9637430906295776,
+      "learning_rate": 0.000986571179693872,
+      "loss": 0.5065,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0595959424972534,
+      "learning_rate": 0.0009864246851958238,
+      "loss": 0.515,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.024770975112915,
+      "learning_rate": 0.0009862774069706345,
+      "loss": 0.4863,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.939887285232544,
+      "learning_rate": 0.0009861293452555987,
+      "loss": 0.3831,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8128208518028259,
+      "learning_rate": 0.0009859805002892731,
+      "loss": 0.3802,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.2043336629867554,
+      "learning_rate": 0.0009858308723114768,
+      "loss": 0.4981,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.2874799966812134,
+      "learning_rate": 0.0009856804615632903,
+      "loss": 0.4038,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8917880058288574,
+      "learning_rate": 0.000985529268287055,
+      "loss": 0.389,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9789330959320068,
+      "learning_rate": 0.000985377292726374,
+      "loss": 0.4525,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8794920444488525,
+      "learning_rate": 0.0009852245351261097,
+      "loss": 0.3787,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7764577269554138,
+      "learning_rate": 0.0009850709957323854,
+      "loss": 0.333,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1386207342147827,
+      "learning_rate": 0.0009849166747925835,
+      "loss": 0.5108,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1627620458602905,
+      "learning_rate": 0.0009847615725553456,
+      "loss": 0.633,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9973104000091553,
+      "learning_rate": 0.0009846056892705729,
+      "loss": 0.3379,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.2538036108016968,
+      "learning_rate": 0.0009844490251894238,
+      "loss": 0.5177,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9299919605255127,
+      "learning_rate": 0.0009842915805643156,
+      "loss": 0.603,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.3065483570098877,
+      "learning_rate": 0.0009841333556489232,
+      "loss": 0.9071,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0391535758972168,
+      "learning_rate": 0.0009839743506981783,
+      "loss": 0.5029,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.7291414737701416,
+      "learning_rate": 0.0009838145659682694,
+      "loss": 0.5993,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.4736104011535645,
+      "learning_rate": 0.0009836540017166418,
+      "loss": 0.662,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.3694871664047241,
+      "learning_rate": 0.0009834926582019967,
+      "loss": 0.5994,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8153218626976013,
+      "learning_rate": 0.00098333053568429,
+      "loss": 0.4832,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8467690348625183,
+      "learning_rate": 0.0009831676344247342,
+      "loss": 0.416,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.1838133335113525,
+      "learning_rate": 0.0009830039546857954,
+      "loss": 0.5026,
+      "step": 231
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.3025498390197754,
+      "learning_rate": 0.000982839496731194,
+      "loss": 0.7166,
+      "step": 232
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0285696983337402,
+      "learning_rate": 0.0009826742608259047,
+      "loss": 0.4291,
+      "step": 233
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0152981281280518,
+      "learning_rate": 0.0009825082472361558,
+      "loss": 0.4904,
+      "step": 234
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1368327140808105,
+      "learning_rate": 0.000982341456229428,
+      "loss": 0.5851,
+      "step": 235
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.998117983341217,
+      "learning_rate": 0.0009821738880744547,
+      "loss": 0.5286,
+      "step": 236
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0000348091125488,
+      "learning_rate": 0.000982005543041222,
+      "loss": 0.4661,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9506464004516602,
+      "learning_rate": 0.000981836421400967,
+      "loss": 0.4049,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1785210371017456,
+      "learning_rate": 0.0009816665234261787,
+      "loss": 0.6693,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0123759508132935,
+      "learning_rate": 0.0009814958493905962,
+      "loss": 0.4642,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8459728360176086,
+      "learning_rate": 0.0009813243995692097,
+      "loss": 0.5215,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8488337397575378,
+      "learning_rate": 0.000981152174238259,
+      "loss": 0.4818,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.2360423803329468,
+      "learning_rate": 0.0009809791736752334,
+      "loss": 0.4158,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.080568790435791,
+      "learning_rate": 0.000980805398158871,
+      "loss": 0.4346,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8085681796073914,
+      "learning_rate": 0.0009806308479691594,
+      "loss": 0.3514,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.110186219215393,
+      "learning_rate": 0.0009804555233873333,
+      "loss": 0.5312,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0116908550262451,
+      "learning_rate": 0.000980279424695876,
+      "loss": 0.5236,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1067167520523071,
+      "learning_rate": 0.000980102552178517,
+      "loss": 0.5093,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7164646983146667,
+      "learning_rate": 0.0009799249061202336,
+      "loss": 0.2628,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0580260753631592,
+      "learning_rate": 0.0009797464868072487,
+      "loss": 0.3958,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7550976276397705,
+      "learning_rate": 0.0009795672945270317,
+      "loss": 0.3006,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9480244517326355,
+      "learning_rate": 0.000979387329568297,
+      "loss": 0.5276,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0090550184249878,
+      "learning_rate": 0.000979206592221004,
+      "loss": 0.4942,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9291450381278992,
+      "learning_rate": 0.0009790250827763566,
+      "loss": 0.3497,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7596597671508789,
+      "learning_rate": 0.0009788428015268028,
+      "loss": 0.3327,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.089870810508728,
+      "learning_rate": 0.0009786597487660335,
+      "loss": 0.705,
+      "step": 256
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.830388069152832,
+      "learning_rate": 0.000978475924788984,
+      "loss": 0.4211,
+      "step": 257
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6773444414138794,
+      "learning_rate": 0.000978291329891831,
+      "loss": 0.3851,
+      "step": 258
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.1150076389312744,
+      "learning_rate": 0.0009781059643719937,
+      "loss": 0.5231,
+      "step": 259
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.810236394405365,
+      "learning_rate": 0.0009779198285281327,
+      "loss": 0.3863,
+      "step": 260
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5466155409812927,
+      "learning_rate": 0.0009777329226601501,
+      "loss": 0.2657,
+      "step": 261
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7009707093238831,
+      "learning_rate": 0.0009775452470691885,
+      "loss": 0.2697,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1568548679351807,
+      "learning_rate": 0.000977356802057631,
+      "loss": 0.4765,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8914077877998352,
+      "learning_rate": 0.0009771675879290996,
+      "loss": 0.6059,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9748645424842834,
+      "learning_rate": 0.0009769776049884564,
+      "loss": 0.5814,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0154123306274414,
+      "learning_rate": 0.0009767868535418014,
+      "loss": 0.5141,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7690478563308716,
+      "learning_rate": 0.0009765953338964734,
+      "loss": 0.5114,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8405517935752869,
+      "learning_rate": 0.0009764030463610489,
+      "loss": 0.373,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.120673418045044,
+      "learning_rate": 0.0009762099912453413,
+      "loss": 0.5522,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8679952621459961,
+      "learning_rate": 0.0009760161688604008,
+      "loss": 0.3988,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6905337572097778,
+      "learning_rate": 0.0009758215795185139,
+      "loss": 0.3931,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7700511813163757,
+      "learning_rate": 0.0009756262235332029,
+      "loss": 0.4642,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8811553716659546,
+      "learning_rate": 0.0009754301012192253,
+      "loss": 0.4669,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.2538728713989258,
+      "learning_rate": 0.0009752332128925731,
+      "loss": 0.6943,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6508512496948242,
+      "learning_rate": 0.0009750355588704727,
+      "loss": 0.3569,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.923376202583313,
+      "learning_rate": 0.0009748371394713842,
+      "loss": 0.5422,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8921462297439575,
+      "learning_rate": 0.0009746379550150009,
+      "loss": 0.4207,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8827762603759766,
+      "learning_rate": 0.0009744380058222483,
+      "loss": 0.4031,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8882277607917786,
+      "learning_rate": 0.0009742372922152847,
+      "loss": 0.5958,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.792838990688324,
+      "learning_rate": 0.0009740358145174998,
+      "loss": 0.4673,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1534898281097412,
+      "learning_rate": 0.0009738335730535142,
+      "loss": 0.5735,
+      "step": 281
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7786389589309692,
+      "learning_rate": 0.000973630568149179,
+      "loss": 0.3735,
+      "step": 282
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8312042951583862,
+      "learning_rate": 0.000973426800131576,
+      "loss": 0.4239,
+      "step": 283
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8238577246665955,
+      "learning_rate": 0.000973222269329016,
+      "loss": 0.5276,
+      "step": 284
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7258448004722595,
+      "learning_rate": 0.0009730169760710386,
+      "loss": 0.3541,
+      "step": 285
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9199716448783875,
+      "learning_rate": 0.0009728109206884125,
+      "loss": 0.5007,
+      "step": 286
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8485223650932312,
+      "learning_rate": 0.0009726041035131339,
+      "loss": 0.5136,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8818596005439758,
+      "learning_rate": 0.0009723965248784264,
+      "loss": 0.4839,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.1623008251190186,
+      "learning_rate": 0.0009721881851187406,
+      "loss": 0.4095,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 8.145987510681152,
+      "learning_rate": 0.0009719790845697534,
+      "loss": 0.4333,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8713394999504089,
+      "learning_rate": 0.0009717692235683675,
+      "loss": 0.59,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7593088746070862,
+      "learning_rate": 0.0009715586024527108,
+      "loss": 0.4106,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9443697333335876,
+      "learning_rate": 0.0009713472215621359,
+      "loss": 0.6273,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8690384030342102,
+      "learning_rate": 0.0009711350812372197,
+      "loss": 0.3668,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5962545871734619,
+      "learning_rate": 0.0009709221818197624,
+      "loss": 0.2087,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9926095008850098,
+      "learning_rate": 0.0009707085236527873,
+      "loss": 0.4812,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9288908243179321,
+      "learning_rate": 0.0009704941070805405,
+      "loss": 0.4751,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1288870573043823,
+      "learning_rate": 0.0009702789324484897,
+      "loss": 0.5058,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6662505865097046,
+      "learning_rate": 0.0009700630001033244,
+      "loss": 0.3099,
+      "step": 299
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.98651123046875,
+      "learning_rate": 0.0009698463103929542,
+      "loss": 0.6345,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6780020594596863,
+      "learning_rate": 0.0009696288636665098,
+      "loss": 0.2811,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0052082538604736,
+      "learning_rate": 0.0009694106602743411,
+      "loss": 0.5526,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.2784535884857178,
+      "learning_rate": 0.0009691917005680173,
+      "loss": 0.7289,
+      "step": 303
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.3910775184631348,
+      "learning_rate": 0.000968971984900326,
+      "loss": 0.8421,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.083485722541809,
+      "learning_rate": 0.0009687515136252732,
+      "loss": 0.6611,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7115268707275391,
+      "learning_rate": 0.0009685302870980817,
+      "loss": 0.359,
+      "step": 306
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6931566596031189,
+      "learning_rate": 0.000968308305675192,
+      "loss": 0.3099,
+      "step": 307
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1021465063095093,
+      "learning_rate": 0.0009680855697142601,
+      "loss": 0.6551,
+      "step": 308
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8683658242225647,
+      "learning_rate": 0.0009678620795741581,
+      "loss": 0.4638,
+      "step": 309
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8903353214263916,
+      "learning_rate": 0.0009676378356149733,
+      "loss": 0.4573,
+      "step": 310
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7868062853813171,
+      "learning_rate": 0.0009674128381980072,
+      "loss": 0.3408,
+      "step": 311
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9776315689086914,
+      "learning_rate": 0.0009671870876857758,
+      "loss": 0.678,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.906579315662384,
+      "learning_rate": 0.0009669605844420078,
+      "loss": 0.3203,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7466129064559937,
+      "learning_rate": 0.0009667333288316453,
+      "loss": 0.3586,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7382181286811829,
+      "learning_rate": 0.0009665053212208426,
+      "loss": 0.3719,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.095239520072937,
+      "learning_rate": 0.000966276561976965,
+      "loss": 0.5452,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9741299748420715,
+      "learning_rate": 0.0009660470514685895,
+      "loss": 0.3071,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.2990411520004272,
+      "learning_rate": 0.0009658167900655031,
+      "loss": 0.7844,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8142491579055786,
+      "learning_rate": 0.000965585778138703,
+      "loss": 0.4015,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9975658655166626,
+      "learning_rate": 0.0009653540160603955,
+      "loss": 0.3717,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.939786434173584,
+      "learning_rate": 0.0009651215042039955,
+      "loss": 0.4744,
+      "step": 321
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1369673013687134,
+      "learning_rate": 0.0009648882429441257,
+      "loss": 0.4542,
+      "step": 322
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9883975386619568,
+      "learning_rate": 0.0009646542326566167,
+      "loss": 0.2185,
+      "step": 323
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0342304706573486,
+      "learning_rate": 0.0009644194737185058,
+      "loss": 0.651,
+      "step": 324
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.635210394859314,
+      "learning_rate": 0.0009641839665080363,
+      "loss": 0.3247,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3649404048919678,
+      "learning_rate": 0.0009639477114046573,
+      "loss": 0.498,
+      "step": 326
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6436899900436401,
+      "learning_rate": 0.0009637107087890229,
+      "loss": 0.2993,
+      "step": 327
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.089766025543213,
+      "learning_rate": 0.0009634729590429916,
+      "loss": 0.5716,
+      "step": 328
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.47708839178085327,
+      "learning_rate": 0.0009632344625496255,
+      "loss": 0.1999,
+      "step": 329
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5733949542045593,
+      "learning_rate": 0.0009629952196931902,
+      "loss": 0.3446,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7943990230560303,
+      "learning_rate": 0.0009627552308591534,
+      "loss": 0.496,
+      "step": 331
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9379316568374634,
+      "learning_rate": 0.0009625144964341852,
+      "loss": 0.6459,
+      "step": 332
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5505914688110352,
+      "learning_rate": 0.0009622730168061567,
+      "loss": 0.213,
+      "step": 333
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0332014560699463,
+      "learning_rate": 0.0009620307923641394,
+      "loss": 0.5829,
+      "step": 334
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9729798436164856,
+      "learning_rate": 0.0009617878234984055,
+      "loss": 0.3003,
+      "step": 335
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0319576263427734,
+      "learning_rate": 0.0009615441106004263,
+      "loss": 0.5261,
+      "step": 336
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9615458846092224,
+      "learning_rate": 0.0009612996540628718,
+      "loss": 0.4578,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7731225490570068,
+      "learning_rate": 0.0009610544542796101,
+      "loss": 0.4425,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9475051164627075,
+      "learning_rate": 0.0009608085116457069,
+      "loss": 0.3878,
+      "step": 339
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2256535291671753,
+      "learning_rate": 0.0009605618265574251,
+      "loss": 0.5571,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5793543457984924,
+      "learning_rate": 0.0009603143994122233,
+      "loss": 0.2945,
+      "step": 341
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7454169988632202,
+      "learning_rate": 0.0009600662306087561,
+      "loss": 0.3575,
+      "step": 342
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.948676586151123,
+      "learning_rate": 0.0009598173205468728,
+      "loss": 0.3536,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7791875600814819,
+      "learning_rate": 0.0009595676696276172,
+      "loss": 0.4515,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7716416716575623,
+      "learning_rate": 0.0009593172782532268,
+      "loss": 0.4915,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2981531620025635,
+      "learning_rate": 0.0009590661468271318,
+      "loss": 0.6453,
+      "step": 346
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9994156360626221,
+      "learning_rate": 0.000958814275753955,
+      "loss": 0.5649,
+      "step": 347
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7039446234703064,
+      "learning_rate": 0.0009585616654395112,
+      "loss": 0.3333,
+      "step": 348
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9314269423484802,
+      "learning_rate": 0.000958308316290806,
+      "loss": 0.6172,
+      "step": 349
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.97852623462677,
+      "learning_rate": 0.0009580542287160348,
+      "loss": 0.5708,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.345961093902588,
+      "learning_rate": 0.0009577994031245839,
+      "loss": 0.35,
+      "step": 351
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4410769939422607,
+      "learning_rate": 0.0009575438399270279,
+      "loss": 0.4146,
+      "step": 352
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9699118137359619,
+      "learning_rate": 0.0009572875395351301,
+      "loss": 0.6043,
+      "step": 353
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6946624517440796,
+      "learning_rate": 0.0009570305023618417,
+      "loss": 0.3559,
+      "step": 354
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1048985719680786,
+      "learning_rate": 0.0009567727288213005,
+      "loss": 0.4694,
+      "step": 355
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0548882484436035,
+      "learning_rate": 0.0009565142193288312,
+      "loss": 0.3553,
+      "step": 356
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1384878158569336,
+      "learning_rate": 0.0009562549743009443,
+      "loss": 0.3558,
+      "step": 357
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0422585010528564,
+      "learning_rate": 0.0009559949941553349,
+      "loss": 0.4952,
+      "step": 358
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0784997940063477,
+      "learning_rate": 0.0009557342793108832,
+      "loss": 0.4329,
+      "step": 359
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0643510818481445,
+      "learning_rate": 0.0009554728301876525,
+      "loss": 0.4763,
+      "step": 360
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1467616558074951,
+      "learning_rate": 0.0009552106472068898,
+      "loss": 0.6148,
+      "step": 361
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.747022807598114,
+      "learning_rate": 0.0009549477307910237,
+      "loss": 0.2631,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0436701774597168,
+      "learning_rate": 0.0009546840813636653,
+      "loss": 0.5859,
+      "step": 363
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9289217591285706,
+      "learning_rate": 0.0009544196993496062,
+      "loss": 0.3481,
+      "step": 364
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7833895683288574,
+      "learning_rate": 0.0009541545851748186,
+      "loss": 0.415,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8271332383155823,
+      "learning_rate": 0.0009538887392664543,
+      "loss": 0.3895,
+      "step": 366
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9571720361709595,
+      "learning_rate": 0.0009536221620528441,
+      "loss": 0.5283,
+      "step": 367
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8841995000839233,
+      "learning_rate": 0.000953354853963497,
+      "loss": 0.5489,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7545637488365173,
+      "learning_rate": 0.0009530868154290997,
+      "loss": 0.2489,
+      "step": 369
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7772701382637024,
+      "learning_rate": 0.0009528180468815154,
+      "loss": 0.343,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6142757534980774,
+      "learning_rate": 0.0009525485487537842,
+      "loss": 0.3387,
+      "step": 371
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8645892143249512,
+      "learning_rate": 0.0009522783214801211,
+      "loss": 0.3795,
+      "step": 372
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9348011016845703,
+      "learning_rate": 0.0009520073654959162,
+      "loss": 0.262,
+      "step": 373
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6604518890380859,
+      "learning_rate": 0.0009517356812377335,
+      "loss": 0.3821,
+      "step": 374
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8515806794166565,
+      "learning_rate": 0.0009514632691433108,
+      "loss": 0.2919,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9010944366455078,
+      "learning_rate": 0.0009511901296515577,
+      "loss": 0.4409,
+      "step": 376
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.058217167854309,
+      "learning_rate": 0.000950916263202557,
+      "loss": 0.528,
+      "step": 377
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9227722883224487,
+      "learning_rate": 0.0009506416702375617,
+      "loss": 0.6265,
+      "step": 378
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0440716743469238,
+      "learning_rate": 0.0009503663511989962,
+      "loss": 0.5644,
+      "step": 379
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.928469181060791,
+      "learning_rate": 0.0009500903065304539,
+      "loss": 0.4252,
+      "step": 380
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9033165574073792,
+      "learning_rate": 0.0009498135366766982,
+      "loss": 0.5126,
+      "step": 381
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6321996450424194,
+      "learning_rate": 0.0009495360420836603,
+      "loss": 0.2985,
+      "step": 382
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.2747009992599487,
+      "learning_rate": 0.0009492578231984394,
+      "loss": 0.5749,
+      "step": 383
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8401326537132263,
+      "learning_rate": 0.0009489788804693016,
+      "loss": 0.407,
+      "step": 384
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8579046130180359,
+      "learning_rate": 0.0009486992143456792,
+      "loss": 0.4831,
+      "step": 385
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0537229776382446,
+      "learning_rate": 0.00094841882527817,
+      "loss": 0.4648,
+      "step": 386
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0211334228515625,
+      "learning_rate": 0.000948137713718537,
+      "loss": 0.513,
+      "step": 387
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7625662684440613,
+      "learning_rate": 0.0009478558801197064,
+      "loss": 0.3782,
+      "step": 388
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7239591479301453,
+      "learning_rate": 0.0009475733249357689,
+      "loss": 0.5105,
+      "step": 389
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9299130439758301,
+      "learning_rate": 0.0009472900486219768,
+      "loss": 0.5122,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.2116215229034424,
+      "learning_rate": 0.000947006051634745,
+      "loss": 0.4988,
+      "step": 391
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.2965437173843384,
+      "learning_rate": 0.0009467213344316492,
+      "loss": 0.3628,
+      "step": 392
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.783905029296875,
+      "learning_rate": 0.0009464358974714253,
+      "loss": 0.2966,
+      "step": 393
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8939562439918518,
+      "learning_rate": 0.0009461497412139697,
+      "loss": 0.5127,
+      "step": 394
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7586861252784729,
+      "learning_rate": 0.0009458628661203367,
+      "loss": 0.2942,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7497037649154663,
+      "learning_rate": 0.0009455752726527394,
+      "loss": 0.356,
+      "step": 396
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.210220456123352,
+      "learning_rate": 0.0009452869612745483,
+      "loss": 0.6779,
+      "step": 397
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6992549300193787,
+      "learning_rate": 0.0009449979324502904,
+      "loss": 0.4716,
+      "step": 398
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7655842304229736,
+      "learning_rate": 0.0009447081866456489,
+      "loss": 0.2852,
+      "step": 399
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.007626533508301,
+      "learning_rate": 0.0009444177243274617,
+      "loss": 0.5292,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.275146245956421,
+      "learning_rate": 0.000944126545963722,
+      "loss": 0.517,
+      "step": 401
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5170719623565674,
+      "learning_rate": 0.0009438346520235759,
+      "loss": 0.5079,
+      "step": 402
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7846792340278625,
+      "learning_rate": 0.0009435420429773227,
+      "loss": 0.2237,
+      "step": 403
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7035520672798157,
+      "learning_rate": 0.0009432487192964141,
+      "loss": 0.4363,
+      "step": 404
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7420192956924438,
+      "learning_rate": 0.0009429546814534529,
+      "loss": 0.2854,
+      "step": 405
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8806753158569336,
+      "learning_rate": 0.0009426599299221925,
+      "loss": 0.3825,
+      "step": 406
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6074947118759155,
+      "learning_rate": 0.0009423644651775368,
+      "loss": 0.4625,
+      "step": 407
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1053515672683716,
+      "learning_rate": 0.0009420682876955382,
+      "loss": 0.4908,
+      "step": 408
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9846407175064087,
+      "learning_rate": 0.0009417713979533975,
+      "loss": 0.4141,
+      "step": 409
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7464879751205444,
+      "learning_rate": 0.0009414737964294635,
+      "loss": 0.4519,
+      "step": 410
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6927564144134521,
+      "learning_rate": 0.0009411754836032315,
+      "loss": 0.4341,
+      "step": 411
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1267709732055664,
+      "learning_rate": 0.0009408764599553429,
+      "loss": 0.6366,
+      "step": 412
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8825417757034302,
+      "learning_rate": 0.0009405767259675845,
+      "loss": 0.5425,
+      "step": 413
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7009013891220093,
+      "learning_rate": 0.0009402762821228874,
+      "loss": 0.5147,
+      "step": 414
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1513880491256714,
+      "learning_rate": 0.0009399751289053266,
+      "loss": 0.4056,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9277740120887756,
+      "learning_rate": 0.00093967326680012,
+      "loss": 0.5039,
+      "step": 416
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.042993187904358,
+      "learning_rate": 0.0009393706962936275,
+      "loss": 0.4798,
+      "step": 417
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6997524499893188,
+      "learning_rate": 0.0009390674178733507,
+      "loss": 0.2539,
+      "step": 418
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7011858224868774,
+      "learning_rate": 0.0009387634320279315,
+      "loss": 0.4242,
+      "step": 419
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5977184772491455,
+      "learning_rate": 0.0009384587392471515,
+      "loss": 0.4977,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0500438213348389,
+      "learning_rate": 0.0009381533400219318,
+      "loss": 0.6709,
+      "step": 421
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5816748142242432,
+      "learning_rate": 0.0009378472348443314,
+      "loss": 0.279,
+      "step": 422
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.848839521408081,
+      "learning_rate": 0.0009375404242075467,
+      "loss": 0.3981,
+      "step": 423
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9058635234832764,
+      "learning_rate": 0.0009372329086059108,
+      "loss": 0.4599,
+      "step": 424
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6673190593719482,
+      "learning_rate": 0.0009369246885348925,
+      "loss": 0.3555,
+      "step": 425
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6937199831008911,
+      "learning_rate": 0.000936615764491096,
+      "loss": 0.3938,
+      "step": 426
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7188507318496704,
+      "learning_rate": 0.0009363061369722595,
+      "loss": 0.3862,
+      "step": 427
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9323291778564453,
+      "learning_rate": 0.0009359958064772547,
+      "loss": 0.4694,
+      "step": 428
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6754212975502014,
+      "learning_rate": 0.0009356847735060856,
+      "loss": 0.3787,
+      "step": 429
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 33.56079864501953,
+      "learning_rate": 0.0009353730385598887,
+      "loss": 0.505,
+      "step": 430
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8478493094444275,
+      "learning_rate": 0.0009350606021409308,
+      "loss": 0.3732,
+      "step": 431
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7283879518508911,
+      "learning_rate": 0.0009347474647526095,
+      "loss": 0.3683,
+      "step": 432
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5498476624488831,
+      "learning_rate": 0.0009344336268994515,
+      "loss": 0.2378,
+      "step": 433
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.034774661064148,
+      "learning_rate": 0.0009341190890871122,
+      "loss": 0.3822,
+      "step": 434
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.530534029006958,
+      "learning_rate": 0.0009338038518223745,
+      "loss": 0.3638,
+      "step": 435
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1095998287200928,
+      "learning_rate": 0.0009334879156131489,
+      "loss": 0.5286,
+      "step": 436
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9873412251472473,
+      "learning_rate": 0.0009331712809684711,
+      "loss": 0.4645,
+      "step": 437
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.3931673765182495,
+      "learning_rate": 0.000932853948398503,
+      "loss": 0.4341,
+      "step": 438
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.264172077178955,
+      "learning_rate": 0.0009325359184145306,
+      "loss": 0.4385,
+      "step": 439
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8946067094802856,
+      "learning_rate": 0.0009322171915289634,
+      "loss": 0.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.332200527191162,
+      "learning_rate": 0.0009318977682553341,
+      "loss": 0.4598,
+      "step": 441
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5916949510574341,
+      "learning_rate": 0.0009315776491082972,
+      "loss": 0.3265,
+      "step": 442
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0136299133300781,
+      "learning_rate": 0.0009312568346036287,
+      "loss": 0.2513,
+      "step": 443
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6893987655639648,
+      "learning_rate": 0.0009309353252582246,
+      "loss": 0.4062,
+      "step": 444
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6134548783302307,
+      "learning_rate": 0.0009306131215901003,
+      "loss": 0.2513,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0449014902114868,
+      "learning_rate": 0.0009302902241183904,
+      "loss": 0.4366,
+      "step": 446
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0707861185073853,
+      "learning_rate": 0.0009299666333633471,
+      "loss": 0.4392,
+      "step": 447
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8052060008049011,
+      "learning_rate": 0.0009296423498463396,
+      "loss": 0.3718,
+      "step": 448
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6378965973854065,
+      "learning_rate": 0.0009293173740898531,
+      "loss": 0.2438,
+      "step": 449
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6215816736221313,
+      "learning_rate": 0.0009289917066174886,
+      "loss": 0.2705,
+      "step": 450
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8111103773117065,
+      "learning_rate": 0.0009286653479539611,
+      "loss": 0.3831,
+      "step": 451
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7510920166969299,
+      "learning_rate": 0.0009283382986250997,
+      "loss": 0.3326,
+      "step": 452
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.3050328493118286,
+      "learning_rate": 0.0009280105591578458,
+      "loss": 0.6138,
+      "step": 453
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0658737421035767,
+      "learning_rate": 0.0009276821300802534,
+      "loss": 0.752,
+      "step": 454
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7931048274040222,
+      "learning_rate": 0.0009273530119214868,
+      "loss": 0.3653,
+      "step": 455
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5900155901908875,
+      "learning_rate": 0.0009270232052118213,
+      "loss": 0.281,
+      "step": 456
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6616344451904297,
+      "learning_rate": 0.0009266927104826409,
+      "loss": 0.4548,
+      "step": 457
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9000492691993713,
+      "learning_rate": 0.0009263615282664388,
+      "loss": 0.4831,
+      "step": 458
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.582861602306366,
+      "learning_rate": 0.0009260296590968156,
+      "loss": 0.2262,
+      "step": 459
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6026660203933716,
+      "learning_rate": 0.0009256971035084784,
+      "loss": 0.3017,
+      "step": 460
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7343093156814575,
+      "learning_rate": 0.0009253638620372409,
+      "loss": 0.3935,
+      "step": 461
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9290016889572144,
+      "learning_rate": 0.0009250299352200213,
+      "loss": 0.5534,
+      "step": 462
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0864694118499756,
+      "learning_rate": 0.0009246953235948423,
+      "loss": 0.442,
+      "step": 463
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1087675094604492,
+      "learning_rate": 0.0009243600277008301,
+      "loss": 0.575,
+      "step": 464
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9854834079742432,
+      "learning_rate": 0.0009240240480782129,
+      "loss": 0.4485,
+      "step": 465
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0405482053756714,
+      "learning_rate": 0.0009236873852683213,
+      "loss": 0.6087,
+      "step": 466
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8447050452232361,
+      "learning_rate": 0.0009233500398135859,
+      "loss": 0.2756,
+      "step": 467
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5067954063415527,
+      "learning_rate": 0.0009230120122575375,
+      "loss": 0.288,
+      "step": 468
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0583200454711914,
+      "learning_rate": 0.0009226733031448061,
+      "loss": 0.5353,
+      "step": 469
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6912018656730652,
+      "learning_rate": 0.0009223339130211192,
+      "loss": 0.326,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.673800468444824,
+      "learning_rate": 0.0009219938424333023,
+      "loss": 0.6445,
+      "step": 471
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9592889547348022,
+      "learning_rate": 0.0009216530919292767,
+      "loss": 0.5633,
+      "step": 472
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8769776225090027,
+      "learning_rate": 0.0009213116620580596,
+      "loss": 0.6629,
+      "step": 473
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6509552001953125,
+      "learning_rate": 0.0009209695533697624,
+      "loss": 0.2326,
+      "step": 474
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7808894515037537,
+      "learning_rate": 0.0009206267664155906,
+      "loss": 0.4868,
+      "step": 475
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6679558753967285,
+      "learning_rate": 0.0009202833017478421,
+      "loss": 0.3613,
+      "step": 476
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7413969039916992,
+      "learning_rate": 0.0009199391599199072,
+      "loss": 0.3134,
+      "step": 477
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6369844675064087,
+      "learning_rate": 0.0009195943414862666,
+      "loss": 0.3656,
+      "step": 478
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5190037488937378,
+      "learning_rate": 0.000919248847002492,
+      "loss": 0.2726,
+      "step": 479
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5270078182220459,
+      "learning_rate": 0.0009189026770252437,
+      "loss": 0.2478,
+      "step": 480
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6408283710479736,
+      "learning_rate": 0.0009185558321122704,
+      "loss": 0.3945,
+      "step": 481
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8977113366127014,
+      "learning_rate": 0.0009182083128224086,
+      "loss": 0.3491,
+      "step": 482
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7618108987808228,
+      "learning_rate": 0.0009178601197155811,
+      "loss": 0.2124,
+      "step": 483
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0637149810791016,
+      "learning_rate": 0.0009175112533527963,
+      "loss": 0.6958,
+      "step": 484
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.015615463256836,
+      "learning_rate": 0.0009171617142961477,
+      "loss": 0.6763,
+      "step": 485
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8489510416984558,
+      "learning_rate": 0.0009168115031088121,
+      "loss": 0.3546,
+      "step": 486
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.03271484375,
+      "learning_rate": 0.0009164606203550497,
+      "loss": 0.4732,
+      "step": 487
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7214429378509521,
+      "learning_rate": 0.0009161090666002028,
+      "loss": 0.3366,
+      "step": 488
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2109496593475342,
+      "learning_rate": 0.0009157568424106941,
+      "loss": 0.5062,
+      "step": 489
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8485642671585083,
+      "learning_rate": 0.0009154039483540273,
+      "loss": 0.3264,
+      "step": 490
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6159496903419495,
+      "learning_rate": 0.0009150503849987852,
+      "loss": 0.3702,
+      "step": 491
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6375138759613037,
+      "learning_rate": 0.0009146961529146284,
+      "loss": 0.3074,
+      "step": 492
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.4139564037322998,
+      "learning_rate": 0.0009143412526722959,
+      "loss": 0.5054,
+      "step": 493
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0630199909210205,
+      "learning_rate": 0.0009139856848436024,
+      "loss": 0.6669,
+      "step": 494
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.021838665008545,
+      "learning_rate": 0.0009136294500014386,
+      "loss": 0.6217,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5356813073158264,
+      "learning_rate": 0.0009132725487197701,
+      "loss": 0.1972,
+      "step": 496
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.595586359500885,
+      "learning_rate": 0.0009129149815736358,
+      "loss": 0.244,
+      "step": 497
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4425952434539795,
+      "learning_rate": 0.0009125567491391475,
+      "loss": 0.2431,
+      "step": 498
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8622236251831055,
+      "learning_rate": 0.0009121978519934894,
+      "loss": 0.5446,
+      "step": 499
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1301122903823853,
+      "learning_rate": 0.0009118382907149164,
+      "loss": 0.5392,
+      "step": 500
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9431802034378052,
+      "learning_rate": 0.0009114780658827531,
+      "loss": 0.4785,
+      "step": 501
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8420708775520325,
+      "learning_rate": 0.0009111171780773937,
+      "loss": 0.4726,
+      "step": 502
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7980074882507324,
+      "learning_rate": 0.0009107556278803002,
+      "loss": 0.3612,
+      "step": 503
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8379204869270325,
+      "learning_rate": 0.0009103934158740022,
+      "loss": 0.3813,
+      "step": 504
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9188324809074402,
+      "learning_rate": 0.0009100305426420956,
+      "loss": 0.3908,
+      "step": 505
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8034757971763611,
+      "learning_rate": 0.0009096670087692413,
+      "loss": 0.4833,
+      "step": 506
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.3085750341415405,
+      "learning_rate": 0.0009093028148411649,
+      "loss": 0.4636,
+      "step": 507
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0781002044677734,
+      "learning_rate": 0.0009089379614446553,
+      "loss": 0.6314,
+      "step": 508
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5669236779212952,
+      "learning_rate": 0.0009085724491675642,
+      "loss": 0.3209,
+      "step": 509
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8634130954742432,
+      "learning_rate": 0.0009082062785988049,
+      "loss": 0.4951,
+      "step": 510
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.070777177810669,
+      "learning_rate": 0.0009078394503283509,
+      "loss": 0.4716,
+      "step": 511
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0060056447982788,
+      "learning_rate": 0.0009074719649472358,
+      "loss": 0.5135,
+      "step": 512
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.975216269493103,
+      "learning_rate": 0.0009071038230475519,
+      "loss": 0.4754,
+      "step": 513
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.1060266494750977,
+      "learning_rate": 0.000906735025222449,
+      "loss": 0.4641,
+      "step": 514
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7294943928718567,
+      "learning_rate": 0.0009063655720661341,
+      "loss": 0.4142,
+      "step": 515
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8892977237701416,
+      "learning_rate": 0.0009059954641738697,
+      "loss": 0.416,
+      "step": 516
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.9333600401878357,
+      "learning_rate": 0.0009056247021419734,
+      "loss": 0.2511,
+      "step": 517
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8996437788009644,
+      "learning_rate": 0.0009052532865678172,
+      "loss": 0.4687,
+      "step": 518
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7445685267448425,
+      "learning_rate": 0.000904881218049825,
+      "loss": 0.2834,
+      "step": 519
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7955701947212219,
+      "learning_rate": 0.0009045084971874737,
+      "loss": 0.4147,
+      "step": 520
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7194240689277649,
+      "learning_rate": 0.000904135124581291,
+      "loss": 0.3512,
+      "step": 521
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5750693678855896,
+      "learning_rate": 0.0009037611008328543,
+      "loss": 0.2896,
+      "step": 522
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8803348541259766,
+      "learning_rate": 0.0009033864265447906,
+      "loss": 0.3131,
+      "step": 523
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.9199376702308655,
+      "learning_rate": 0.000903011102320775,
+      "loss": 0.5168,
+      "step": 524
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.974337100982666,
+      "learning_rate": 0.0009026351287655293,
+      "loss": 0.4737,
+      "step": 525
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.9163052439689636,
+      "learning_rate": 0.0009022585064848221,
+      "loss": 0.4406,
+      "step": 526
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7733398675918579,
+      "learning_rate": 0.0009018812360854671,
+      "loss": 0.3962,
+      "step": 527
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.0492855310440063,
+      "learning_rate": 0.0009015033181753218,
+      "loss": 0.6456,
+      "step": 528
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.1636652946472168,
+      "learning_rate": 0.0009011247533632876,
+      "loss": 0.5505,
+      "step": 529
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6729533672332764,
+      "learning_rate": 0.0009007455422593077,
+      "loss": 0.4986,
+      "step": 530
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5751211643218994,
+      "learning_rate": 0.0009003656854743667,
+      "loss": 0.2593,
+      "step": 531
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8281634449958801,
+      "learning_rate": 0.00089998518362049,
+      "loss": 0.4845,
+      "step": 532
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8993703126907349,
+      "learning_rate": 0.0008996040373107414,
+      "loss": 0.4441,
+      "step": 533
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.961287796497345,
+      "learning_rate": 0.0008992222471592239,
+      "loss": 0.3793,
+      "step": 534
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8187610507011414,
+      "learning_rate": 0.0008988398137810777,
+      "loss": 0.3344,
+      "step": 535
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.307085633277893,
+      "learning_rate": 0.000898456737792479,
+      "loss": 0.6025,
+      "step": 536
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7470046281814575,
+      "learning_rate": 0.0008980730198106394,
+      "loss": 0.4046,
+      "step": 537
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.2329001426696777,
+      "learning_rate": 0.0008976886604538055,
+      "loss": 0.5285,
+      "step": 538
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.9736171960830688,
+      "learning_rate": 0.0008973036603412566,
+      "loss": 0.3526,
+      "step": 539
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.47753632068634033,
+      "learning_rate": 0.0008969180200933046,
+      "loss": 0.2434,
+      "step": 540
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.708318293094635,
+      "learning_rate": 0.000896531740331293,
+      "loss": 0.3414,
+      "step": 541
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6310819387435913,
+      "learning_rate": 0.0008961448216775954,
+      "loss": 0.4008,
+      "step": 542
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6304932832717896,
+      "learning_rate": 0.0008957572647556148,
+      "loss": 0.3513,
+      "step": 543
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1449286937713623,
+      "learning_rate": 0.0008953690701897827,
+      "loss": 0.3183,
+      "step": 544
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.1888549327850342,
+      "learning_rate": 0.0008949802386055581,
+      "loss": 0.6199,
+      "step": 545
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5465713739395142,
+      "learning_rate": 0.0008945907706294261,
+      "loss": 0.3512,
+      "step": 546
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.1018520593643188,
+      "learning_rate": 0.0008942006668888971,
+      "loss": 0.6905,
+      "step": 547
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8554862141609192,
+      "learning_rate": 0.0008938099280125062,
+      "loss": 0.4642,
+      "step": 548
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.9546778202056885,
+      "learning_rate": 0.0008934185546298116,
+      "loss": 0.3927,
+      "step": 549
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7980965971946716,
+      "learning_rate": 0.0008930265473713938,
+      "loss": 0.3672,
+      "step": 550
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8285545706748962,
+      "learning_rate": 0.0008926339068688546,
+      "loss": 0.4441,
+      "step": 551
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7279841303825378,
+      "learning_rate": 0.0008922406337548161,
+      "loss": 0.4784,
+      "step": 552
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8624037504196167,
+      "learning_rate": 0.0008918467286629199,
+      "loss": 0.2741,
+      "step": 553
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6326537728309631,
+      "learning_rate": 0.0008914521922278255,
+      "loss": 0.3369,
+      "step": 554
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7652431130409241,
+      "learning_rate": 0.0008910570250852097,
+      "loss": 0.3412,
+      "step": 555
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6396991610527039,
+      "learning_rate": 0.0008906612278717657,
+      "loss": 0.2534,
+      "step": 556
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 6.255046367645264,
+      "learning_rate": 0.0008902648012252013,
+      "loss": 0.3538,
+      "step": 557
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8188296556472778,
+      "learning_rate": 0.0008898677457842395,
+      "loss": 0.3985,
+      "step": 558
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0077953338623047,
+      "learning_rate": 0.0008894700621886152,
+      "loss": 0.4399,
+      "step": 559
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7648623585700989,
+      "learning_rate": 0.0008890717510790764,
+      "loss": 0.3273,
+      "step": 560
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.9077563881874084,
+      "learning_rate": 0.0008886728130973814,
+      "loss": 0.3641,
+      "step": 561
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.5442649126052856,
+      "learning_rate": 0.0008882732488862988,
+      "loss": 0.517,
+      "step": 562
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5807144045829773,
+      "learning_rate": 0.0008878730590896064,
+      "loss": 0.2481,
+      "step": 563
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7620820999145508,
+      "learning_rate": 0.0008874722443520898,
+      "loss": 0.4393,
+      "step": 564
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6543344259262085,
+      "learning_rate": 0.0008870708053195413,
+      "loss": 0.334,
+      "step": 565
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.3936197757720947,
+      "learning_rate": 0.0008866687426387592,
+      "loss": 0.5489,
+      "step": 566
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.0267633199691772,
+      "learning_rate": 0.0008862660569575465,
+      "loss": 0.4184,
+      "step": 567
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8449360728263855,
+      "learning_rate": 0.0008858627489247104,
+      "loss": 0.2618,
+      "step": 568
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5409068465232849,
+      "learning_rate": 0.0008854588191900604,
+      "loss": 0.3323,
+      "step": 569
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7773855924606323,
+      "learning_rate": 0.0008850542684044079,
+      "loss": 0.4866,
+      "step": 570
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9105958938598633,
+      "learning_rate": 0.0008846490972195647,
+      "loss": 0.3367,
+      "step": 571
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5734848380088806,
+      "learning_rate": 0.0008842433062883426,
+      "loss": 0.3055,
+      "step": 572
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6108047366142273,
+      "learning_rate": 0.0008838368962645513,
+      "loss": 0.4151,
+      "step": 573
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9906135201454163,
+      "learning_rate": 0.0008834298678029989,
+      "loss": 0.5312,
+      "step": 574
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8111811280250549,
+      "learning_rate": 0.000883022221559489,
+      "loss": 0.2886,
+      "step": 575
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7566137313842773,
+      "learning_rate": 0.0008826139581908211,
+      "loss": 0.3234,
+      "step": 576
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5996237993240356,
+      "learning_rate": 0.000882205078354789,
+      "loss": 0.3402,
+      "step": 577
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.1945191621780396,
+      "learning_rate": 0.0008817955827101794,
+      "loss": 0.3968,
+      "step": 578
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7979175448417664,
+      "learning_rate": 0.0008813854719167713,
+      "loss": 0.4893,
+      "step": 579
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.369984745979309,
+      "learning_rate": 0.0008809747466353355,
+      "loss": 0.6056,
+      "step": 580
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.761488676071167,
+      "learning_rate": 0.0008805634075276318,
+      "loss": 0.3967,
+      "step": 581
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8360129594802856,
+      "learning_rate": 0.0008801514552564096,
+      "loss": 0.4611,
+      "step": 582
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5968575477600098,
+      "learning_rate": 0.0008797388904854064,
+      "loss": 0.3037,
+      "step": 583
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5847429037094116,
+      "learning_rate": 0.000879325713879346,
+      "loss": 0.2928,
+      "step": 584
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7863568067550659,
+      "learning_rate": 0.0008789119261039385,
+      "loss": 0.4884,
+      "step": 585
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.782353937625885,
+      "learning_rate": 0.0008784975278258782,
+      "loss": 0.3364,
+      "step": 586
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8323281407356262,
+      "learning_rate": 0.0008780825197128437,
+      "loss": 0.447,
+      "step": 587
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6234328746795654,
+      "learning_rate": 0.0008776669024334956,
+      "loss": 0.3116,
+      "step": 588
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5174146890640259,
+      "learning_rate": 0.0008772506766574761,
+      "loss": 0.308,
+      "step": 589
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.8928273320198059,
+      "learning_rate": 0.0008768338430554083,
+      "loss": 0.4464,
+      "step": 590
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5359809398651123,
+      "learning_rate": 0.0008764164022988937,
+      "loss": 0.3089,
+      "step": 591
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.861498236656189,
+      "learning_rate": 0.000875998355060513,
+      "loss": 0.7151,
+      "step": 592
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5844029188156128,
+      "learning_rate": 0.0008755797020138234,
+      "loss": 0.2893,
+      "step": 593
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.9362619519233704,
+      "learning_rate": 0.0008751604438333587,
+      "loss": 0.52,
+      "step": 594
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6978433132171631,
+      "learning_rate": 0.0008747405811946271,
+      "loss": 0.3421,
+      "step": 595
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.232534170150757,
+      "learning_rate": 0.0008743201147741112,
+      "loss": 0.3457,
+      "step": 596
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6883057951927185,
+      "learning_rate": 0.000873899045249266,
+      "loss": 0.3763,
+      "step": 597
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7936997413635254,
+      "learning_rate": 0.0008734773732985186,
+      "loss": 0.2822,
+      "step": 598
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.716871976852417,
+      "learning_rate": 0.0008730550996012668,
+      "loss": 0.3876,
+      "step": 599
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5481619834899902,
+      "learning_rate": 0.0008726322248378774,
+      "loss": 0.3264,
+      "step": 600
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4995492398738861,
+      "learning_rate": 0.000872208749689686,
+      "loss": 0.2462,
+      "step": 601
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.8453775644302368,
+      "learning_rate": 0.0008717846748389956,
+      "loss": 0.6325,
+      "step": 602
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.8821859359741211,
+      "learning_rate": 0.0008713600009690752,
+      "loss": 0.5292,
+      "step": 603
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.598285973072052,
+      "learning_rate": 0.0008709347287641592,
+      "loss": 0.3262,
+      "step": 604
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.9492723941802979,
+      "learning_rate": 0.0008705088589094458,
+      "loss": 0.6233,
+      "step": 605
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7275155186653137,
+      "learning_rate": 0.0008700823920910963,
+      "loss": 0.2679,
+      "step": 606
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.8029650449752808,
+      "learning_rate": 0.0008696553289962338,
+      "loss": 0.3213,
+      "step": 607
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7132914662361145,
+      "learning_rate": 0.0008692276703129421,
+      "loss": 0.244,
+      "step": 608
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.9113848209381104,
+      "learning_rate": 0.0008687994167302642,
+      "loss": 0.6058,
+      "step": 609
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.869147777557373,
+      "learning_rate": 0.0008683705689382025,
+      "loss": 0.3599,
+      "step": 610
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6616409420967102,
+      "learning_rate": 0.000867941127627716,
+      "loss": 0.3509,
+      "step": 611
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.1475846767425537,
+      "learning_rate": 0.0008675110934907205,
+      "loss": 0.4931,
+      "step": 612
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5931513905525208,
+      "learning_rate": 0.0008670804672200865,
+      "loss": 0.3331,
+      "step": 613
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8337993025779724,
+      "learning_rate": 0.000866649249509639,
+      "loss": 0.4516,
+      "step": 614
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.170406460762024,
+      "learning_rate": 0.0008662174410541554,
+      "loss": 0.5221,
+      "step": 615
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.1749751567840576,
+      "learning_rate": 0.0008657850425493655,
+      "loss": 0.5252,
+      "step": 616
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7522830963134766,
+      "learning_rate": 0.0008653520546919493,
+      "loss": 0.2663,
+      "step": 617
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.0195879936218262,
+      "learning_rate": 0.0008649184781795368,
+      "loss": 0.4431,
+      "step": 618
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.695127546787262,
+      "learning_rate": 0.0008644843137107057,
+      "loss": 0.3405,
+      "step": 619
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8795903325080872,
+      "learning_rate": 0.0008640495619849821,
+      "loss": 0.2708,
+      "step": 620
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7659955024719238,
+      "learning_rate": 0.0008636142237028372,
+      "loss": 0.4782,
+      "step": 621
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5857252478599548,
+      "learning_rate": 0.0008631782995656883,
+      "loss": 0.2298,
+      "step": 622
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7243017554283142,
+      "learning_rate": 0.0008627417902758956,
+      "loss": 0.3273,
+      "step": 623
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8496147990226746,
+      "learning_rate": 0.0008623046965367629,
+      "loss": 0.2599,
+      "step": 624
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7897714972496033,
+      "learning_rate": 0.000861867019052535,
+      "loss": 0.2524,
+      "step": 625
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.9007054567337036,
+      "learning_rate": 0.0008614287585283981,
+      "loss": 0.4663,
+      "step": 626
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7119572162628174,
+      "learning_rate": 0.0008609899156704768,
+      "loss": 0.3219,
+      "step": 627
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4926498830318451,
+      "learning_rate": 0.0008605504911858347,
+      "loss": 0.1927,
+      "step": 628
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8483437299728394,
+      "learning_rate": 0.0008601104857824722,
+      "loss": 0.3465,
+      "step": 629
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5934801697731018,
+      "learning_rate": 0.0008596699001693256,
+      "loss": 0.245,
+      "step": 630
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6396439671516418,
+      "learning_rate": 0.0008592287350562663,
+      "loss": 0.2867,
+      "step": 631
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.1100503206253052,
+      "learning_rate": 0.0008587869911540992,
+      "loss": 0.6473,
+      "step": 632
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8750129342079163,
+      "learning_rate": 0.0008583446691745618,
+      "loss": 0.4068,
+      "step": 633
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4340028464794159,
+      "learning_rate": 0.0008579017698303229,
+      "loss": 0.2129,
+      "step": 634
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6517631411552429,
+      "learning_rate": 0.0008574582938349817,
+      "loss": 0.3401,
+      "step": 635
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8881502151489258,
+      "learning_rate": 0.0008570142419030666,
+      "loss": 0.4902,
+      "step": 636
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5850522518157959,
+      "learning_rate": 0.0008565696147500337,
+      "loss": 0.3468,
+      "step": 637
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.985785722732544,
+      "learning_rate": 0.0008561244130922658,
+      "loss": 0.3509,
+      "step": 638
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.1336452960968018,
+      "learning_rate": 0.0008556786376470717,
+      "loss": 0.5521,
+      "step": 639
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.846519947052002,
+      "learning_rate": 0.0008552322891326845,
+      "loss": 0.3729,
+      "step": 640
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.1361637115478516,
+      "learning_rate": 0.0008547853682682604,
+      "loss": 0.6052,
+      "step": 641
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7876294255256653,
+      "learning_rate": 0.0008543378757738785,
+      "loss": 0.4139,
+      "step": 642
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.8827511668205261,
+      "learning_rate": 0.000853889812370538,
+      "loss": 0.4332,
+      "step": 643
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5149502754211426,
+      "learning_rate": 0.0008534411787801586,
+      "loss": 0.2062,
+      "step": 644
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3796164095401764,
+      "learning_rate": 0.0008529919757255782,
+      "loss": 0.1856,
+      "step": 645
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.0929725170135498,
+      "learning_rate": 0.0008525422039305529,
+      "loss": 0.6143,
+      "step": 646
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.43168944120407104,
+      "learning_rate": 0.0008520918641197542,
+      "loss": 0.2395,
+      "step": 647
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5673578977584839,
+      "learning_rate": 0.0008516409570187696,
+      "loss": 0.272,
+      "step": 648
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7821611762046814,
+      "learning_rate": 0.0008511894833541005,
+      "loss": 0.2796,
+      "step": 649
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.9285488724708557,
+      "learning_rate": 0.0008507374438531607,
+      "loss": 0.3982,
+      "step": 650
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.8097355365753174,
+      "learning_rate": 0.0008502848392442759,
+      "loss": 0.5184,
+      "step": 651
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8420424461364746,
+      "learning_rate": 0.0008498316702566827,
+      "loss": 0.5312,
+      "step": 652
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7076250910758972,
+      "learning_rate": 0.0008493779376205264,
+      "loss": 0.399,
+      "step": 653
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5107129216194153,
+      "learning_rate": 0.0008489236420668609,
+      "loss": 0.1981,
+      "step": 654
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7086421847343445,
+      "learning_rate": 0.0008484687843276469,
+      "loss": 0.2847,
+      "step": 655
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3643406331539154,
+      "learning_rate": 0.0008480133651357505,
+      "loss": 0.1907,
+      "step": 656
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5828003883361816,
+      "learning_rate": 0.0008475573852249434,
+      "loss": 0.32,
+      "step": 657
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6647655963897705,
+      "learning_rate": 0.0008471008453298998,
+      "loss": 0.3941,
+      "step": 658
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4620630741119385,
+      "learning_rate": 0.0008466437461861964,
+      "loss": 0.3344,
+      "step": 659
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.8030277490615845,
+      "learning_rate": 0.0008461860885303114,
+      "loss": 0.4826,
+      "step": 660
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.8055134415626526,
+      "learning_rate": 0.0008457278730996223,
+      "loss": 0.4639,
+      "step": 661
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.8905057907104492,
+      "learning_rate": 0.0008452691006324055,
+      "loss": 0.5489,
+      "step": 662
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7118740081787109,
+      "learning_rate": 0.0008448097718678349,
+      "loss": 0.2674,
+      "step": 663
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5584644079208374,
+      "learning_rate": 0.0008443498875459808,
+      "loss": 0.2757,
+      "step": 664
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.668904185295105,
+      "learning_rate": 0.0008438894484078086,
+      "loss": 0.481,
+      "step": 665
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6983538866043091,
+      "learning_rate": 0.0008434284551951771,
+      "loss": 0.293,
+      "step": 666
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8796665072441101,
+      "learning_rate": 0.0008429669086508389,
+      "loss": 0.1856,
+      "step": 667
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6536858081817627,
+      "learning_rate": 0.000842504809518437,
+      "loss": 0.4051,
+      "step": 668
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5601435303688049,
+      "learning_rate": 0.0008420421585425055,
+      "loss": 0.2487,
+      "step": 669
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.9737586975097656,
+      "learning_rate": 0.0008415789564684673,
+      "loss": 0.5005,
+      "step": 670
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.919513463973999,
+      "learning_rate": 0.0008411152040426331,
+      "loss": 0.4213,
+      "step": 671
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8600997924804688,
+      "learning_rate": 0.0008406509020122008,
+      "loss": 0.4536,
+      "step": 672
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5289947390556335,
+      "learning_rate": 0.0008401860511252534,
+      "loss": 0.3451,
+      "step": 673
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.2260289192199707,
+      "learning_rate": 0.0008397206521307583,
+      "loss": 0.5251,
+      "step": 674
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.671043872833252,
+      "learning_rate": 0.0008392547057785661,
+      "loss": 0.3901,
+      "step": 675
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7237336039543152,
+      "learning_rate": 0.0008387882128194093,
+      "loss": 0.3816,
+      "step": 676
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5595833659172058,
+      "learning_rate": 0.000838321174004901,
+      "loss": 0.2474,
+      "step": 677
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5163483619689941,
+      "learning_rate": 0.0008378535900875339,
+      "loss": 0.2697,
+      "step": 678
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7166603803634644,
+      "learning_rate": 0.000837385461820679,
+      "loss": 0.4622,
+      "step": 679
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8966984748840332,
+      "learning_rate": 0.0008369167899585841,
+      "loss": 0.528,
+      "step": 680
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.9994395971298218,
+      "learning_rate": 0.0008364475752563728,
+      "loss": 0.4077,
+      "step": 681
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5297536849975586,
+      "learning_rate": 0.0008359778184700439,
+      "loss": 0.3609,
+      "step": 682
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.006371259689331,
+      "learning_rate": 0.0008355075203564693,
+      "loss": 0.4598,
+      "step": 683
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.0370407104492188,
+      "learning_rate": 0.0008350366816733927,
+      "loss": 0.5666,
+      "step": 684
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7095463871955872,
+      "learning_rate": 0.0008345653031794292,
+      "loss": 0.301,
+      "step": 685
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.0551422834396362,
+      "learning_rate": 0.0008340933856340636,
+      "loss": 0.595,
+      "step": 686
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6964996457099915,
+      "learning_rate": 0.0008336209297976489,
+      "loss": 0.322,
+      "step": 687
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.48713555932044983,
+      "learning_rate": 0.000833147936431406,
+      "loss": 0.2985,
+      "step": 688
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7721118330955505,
+      "learning_rate": 0.0008326744062974212,
+      "loss": 0.363,
+      "step": 689
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0710742473602295,
+      "learning_rate": 0.0008322003401586462,
+      "loss": 0.3125,
+      "step": 690
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5489082932472229,
+      "learning_rate": 0.0008317257387788959,
+      "loss": 0.2446,
+      "step": 691
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.927073061466217,
+      "learning_rate": 0.0008312506029228478,
+      "loss": 0.5611,
+      "step": 692
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8152050971984863,
+      "learning_rate": 0.0008307749333560405,
+      "loss": 0.4448,
+      "step": 693
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.9185419082641602,
+      "learning_rate": 0.0008302987308448724,
+      "loss": 0.2668,
+      "step": 694
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.0600314140319824,
+      "learning_rate": 0.0008298219961566008,
+      "loss": 0.3617,
+      "step": 695
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 4.470431327819824,
+      "learning_rate": 0.0008293447300593402,
+      "loss": 0.2617,
+      "step": 696
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5481506586074829,
+      "learning_rate": 0.0008288669333220615,
+      "loss": 0.3159,
+      "step": 697
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.330317497253418,
+      "learning_rate": 0.0008283886067145907,
+      "loss": 0.5164,
+      "step": 698
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6495181322097778,
+      "learning_rate": 0.0008279097510076071,
+      "loss": 0.2719,
+      "step": 699
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7511414885520935,
+      "learning_rate": 0.0008274303669726426,
+      "loss": 0.4437,
+      "step": 700
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4534698724746704,
+      "learning_rate": 0.0008269504553820806,
+      "loss": 0.1816,
+      "step": 701
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 3.9947943687438965,
+      "learning_rate": 0.0008264700170091543,
+      "loss": 0.6949,
+      "step": 702
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8710176944732666,
+      "learning_rate": 0.0008259890526279459,
+      "loss": 0.5141,
+      "step": 703
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6071026921272278,
+      "learning_rate": 0.0008255075630133846,
+      "loss": 0.3386,
+      "step": 704
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8602607250213623,
+      "learning_rate": 0.0008250255489412463,
+      "loss": 0.3127,
+      "step": 705
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.1472303867340088,
+      "learning_rate": 0.0008245430111881518,
+      "loss": 0.4705,
+      "step": 706
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.527337908744812,
+      "learning_rate": 0.0008240599505315655,
+      "loss": 0.2464,
+      "step": 707
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.2686009407043457,
+      "learning_rate": 0.0008235763677497945,
+      "loss": 0.4066,
+      "step": 708
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6418237090110779,
+      "learning_rate": 0.0008230922636219871,
+      "loss": 0.3058,
+      "step": 709
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5529546141624451,
+      "learning_rate": 0.0008226076389281315,
+      "loss": 0.3277,
+      "step": 710
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6344144940376282,
+      "learning_rate": 0.0008221224944490548,
+      "loss": 0.4421,
+      "step": 711
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5650907158851624,
+      "learning_rate": 0.0008216368309664213,
+      "loss": 0.2049,
+      "step": 712
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.8746808767318726,
+      "learning_rate": 0.0008211506492627319,
+      "loss": 0.3281,
+      "step": 713
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7829561829566956,
+      "learning_rate": 0.000820663950121322,
+      "loss": 0.4204,
+      "step": 714
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6239231824874878,
+      "learning_rate": 0.0008201767343263612,
+      "loss": 0.3844,
+      "step": 715
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.8010687232017517,
+      "learning_rate": 0.000819689002662851,
+      "loss": 0.4216,
+      "step": 716
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.803775429725647,
+      "learning_rate": 0.0008192007559166247,
+      "loss": 0.5751,
+      "step": 717
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6561333537101746,
+      "learning_rate": 0.0008187119948743449,
+      "loss": 0.253,
+      "step": 718
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7546800374984741,
+      "learning_rate": 0.0008182227203235031,
+      "loss": 0.3908,
+      "step": 719
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.1675357818603516,
+      "learning_rate": 0.0008177329330524181,
+      "loss": 0.4212,
+      "step": 720
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 26.986547470092773,
+      "learning_rate": 0.000817242633850235,
+      "loss": 0.9805,
+      "step": 721
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.916666567325592,
+      "learning_rate": 0.0008167518235069235,
+      "loss": 0.3193,
+      "step": 722
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9135617613792419,
+      "learning_rate": 0.0008162605028132768,
+      "loss": 0.4579,
+      "step": 723
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9347028732299805,
+      "learning_rate": 0.0008157686725609105,
+      "loss": 0.5889,
+      "step": 724
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5087236762046814,
+      "learning_rate": 0.0008152763335422613,
+      "loss": 0.1905,
+      "step": 725
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.8403140306472778,
+      "learning_rate": 0.0008147834865505854,
+      "loss": 0.4889,
+      "step": 726
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5848077535629272,
+      "learning_rate": 0.0008142901323799578,
+      "loss": 0.3529,
+      "step": 727
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6129829287528992,
+      "learning_rate": 0.0008137962718252701,
+      "loss": 0.2712,
+      "step": 728
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.2721717357635498,
+      "learning_rate": 0.0008133019056822303,
+      "loss": 0.3922,
+      "step": 729
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6762248277664185,
+      "learning_rate": 0.0008128070347473608,
+      "loss": 0.506,
+      "step": 730
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6848029494285583,
+      "learning_rate": 0.0008123116598179971,
+      "loss": 0.4111,
+      "step": 731
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9460812211036682,
+      "learning_rate": 0.0008118157816922874,
+      "loss": 0.3721,
+      "step": 732
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.56548273563385,
+      "learning_rate": 0.0008113194011691899,
+      "loss": 0.5985,
+      "step": 733
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7197418212890625,
+      "learning_rate": 0.0008108225190484727,
+      "loss": 0.3584,
+      "step": 734
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.60785973072052,
+      "learning_rate": 0.0008103251361307119,
+      "loss": 0.3481,
+      "step": 735
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7157567739486694,
+      "learning_rate": 0.0008098272532172906,
+      "loss": 0.2598,
+      "step": 736
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6115480065345764,
+      "learning_rate": 0.0008093288711103972,
+      "loss": 0.2818,
+      "step": 737
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7690091133117676,
+      "learning_rate": 0.000808829990613025,
+      "loss": 0.4137,
+      "step": 738
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6163682341575623,
+      "learning_rate": 0.0008083306125289698,
+      "loss": 0.2826,
+      "step": 739
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6936965584754944,
+      "learning_rate": 0.0008078307376628291,
+      "loss": 0.3105,
+      "step": 740
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.46171334385871887,
+      "learning_rate": 0.0008073303668200011,
+      "loss": 0.2915,
+      "step": 741
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.1957091093063354,
+      "learning_rate": 0.000806829500806683,
+      "loss": 0.6768,
+      "step": 742
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6681724190711975,
+      "learning_rate": 0.0008063281404298699,
+      "loss": 0.3182,
+      "step": 743
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8327012062072754,
+      "learning_rate": 0.0008058262864973529,
+      "loss": 0.3312,
+      "step": 744
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6635708808898926,
+      "learning_rate": 0.0008053239398177191,
+      "loss": 0.2829,
+      "step": 745
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.902023196220398,
+      "learning_rate": 0.000804821101200349,
+      "loss": 0.3359,
+      "step": 746
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5339200496673584,
+      "learning_rate": 0.000804317771455416,
+      "loss": 0.2655,
+      "step": 747
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7631103992462158,
+      "learning_rate": 0.0008038139513938846,
+      "loss": 0.3638,
+      "step": 748
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5655779838562012,
+      "learning_rate": 0.0008033096418275092,
+      "loss": 0.2724,
+      "step": 749
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.634833812713623,
+      "learning_rate": 0.0008028048435688333,
+      "loss": 0.2684,
+      "step": 750
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8530511260032654,
+      "learning_rate": 0.0008022995574311875,
+      "loss": 0.3219,
+      "step": 751
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.9724946022033691,
+      "learning_rate": 0.0008017937842286883,
+      "loss": 0.446,
+      "step": 752
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8961754441261292,
+      "learning_rate": 0.0008012875247762372,
+      "loss": 0.2896,
+      "step": 753
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8034845590591431,
+      "learning_rate": 0.0008007807798895194,
+      "loss": 0.3136,
+      "step": 754
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7278534173965454,
+      "learning_rate": 0.0008002735503850016,
+      "loss": 0.3642,
+      "step": 755
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8983692526817322,
+      "learning_rate": 0.0007997658370799317,
+      "loss": 0.5148,
+      "step": 756
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8118635416030884,
+      "learning_rate": 0.0007992576407923372,
+      "loss": 0.63,
+      "step": 757
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5938274264335632,
+      "learning_rate": 0.0007987489623410236,
+      "loss": 0.3902,
+      "step": 758
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7841687202453613,
+      "learning_rate": 0.0007982398025455732,
+      "loss": 0.3287,
+      "step": 759
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6733991503715515,
+      "learning_rate": 0.000797730162226344,
+      "loss": 0.2722,
+      "step": 760
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7089006304740906,
+      "learning_rate": 0.0007972200422044682,
+      "loss": 0.3555,
+      "step": 761
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8278717398643494,
+      "learning_rate": 0.0007967094433018508,
+      "loss": 0.3494,
+      "step": 762
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5189366936683655,
+      "learning_rate": 0.0007961983663411684,
+      "loss": 0.258,
+      "step": 763
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.0337847471237183,
+      "learning_rate": 0.0007956868121458678,
+      "loss": 0.5117,
+      "step": 764
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6721497178077698,
+      "learning_rate": 0.000795174781540165,
+      "loss": 0.2721,
+      "step": 765
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7176830768585205,
+      "learning_rate": 0.0007946622753490433,
+      "loss": 0.3076,
+      "step": 766
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6639053821563721,
+      "learning_rate": 0.0007941492943982522,
+      "loss": 0.2655,
+      "step": 767
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.749017596244812,
+      "learning_rate": 0.0007936358395143064,
+      "loss": 0.3134,
+      "step": 768
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5415012240409851,
+      "learning_rate": 0.0007931219115244841,
+      "loss": 0.2271,
+      "step": 769
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4201126992702484,
+      "learning_rate": 0.0007926075112568258,
+      "loss": 0.1876,
+      "step": 770
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6195578575134277,
+      "learning_rate": 0.0007920926395401326,
+      "loss": 0.2392,
+      "step": 771
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6413165330886841,
+      "learning_rate": 0.000791577297203966,
+      "loss": 0.4064,
+      "step": 772
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7453102469444275,
+      "learning_rate": 0.0007910614850786447,
+      "loss": 0.2597,
+      "step": 773
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5878060460090637,
+      "learning_rate": 0.0007905452039952452,
+      "loss": 0.3192,
+      "step": 774
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.9722216725349426,
+      "learning_rate": 0.0007900284547855992,
+      "loss": 0.4181,
+      "step": 775
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.9931303262710571,
+      "learning_rate": 0.0007895112382822925,
+      "loss": 0.4329,
+      "step": 776
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6245104074478149,
+      "learning_rate": 0.0007889935553186641,
+      "loss": 0.3324,
+      "step": 777
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5244526267051697,
+      "learning_rate": 0.0007884754067288047,
+      "loss": 0.2828,
+      "step": 778
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.3127760887145996,
+      "learning_rate": 0.0007879567933475547,
+      "loss": 0.5697,
+      "step": 779
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8880525231361389,
+      "learning_rate": 0.0007874377160105036,
+      "loss": 0.6286,
+      "step": 780
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6588274240493774,
+      "learning_rate": 0.0007869181755539887,
+      "loss": 0.3326,
+      "step": 781
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6414768099784851,
+      "learning_rate": 0.0007863981728150931,
+      "loss": 0.3268,
+      "step": 782
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6813773512840271,
+      "learning_rate": 0.000785877708631645,
+      "loss": 0.4447,
+      "step": 783
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.40819892287254333,
+      "learning_rate": 0.000785356783842216,
+      "loss": 0.187,
+      "step": 784
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.49441203474998474,
+      "learning_rate": 0.0007848353992861195,
+      "loss": 0.1708,
+      "step": 785
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7067763209342957,
+      "learning_rate": 0.00078431355580341,
+      "loss": 0.2819,
+      "step": 786
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.44827333092689514,
+      "learning_rate": 0.0007837912542348818,
+      "loss": 0.2037,
+      "step": 787
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5752569437026978,
+      "learning_rate": 0.0007832684954220664,
+      "loss": 0.3028,
+      "step": 788
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6432071328163147,
+      "learning_rate": 0.0007827452802072328,
+      "loss": 0.3532,
+      "step": 789
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7072285413742065,
+      "learning_rate": 0.0007822216094333848,
+      "loss": 0.3445,
+      "step": 790
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5741270780563354,
+      "learning_rate": 0.0007816974839442604,
+      "loss": 0.2608,
+      "step": 791
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5581470727920532,
+      "learning_rate": 0.0007811729045843302,
+      "loss": 0.2678,
+      "step": 792
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4816420376300812,
+      "learning_rate": 0.0007806478721987965,
+      "loss": 0.2007,
+      "step": 793
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5070657134056091,
+      "learning_rate": 0.0007801223876335907,
+      "loss": 0.3498,
+      "step": 794
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6863113641738892,
+      "learning_rate": 0.0007795964517353734,
+      "loss": 0.2516,
+      "step": 795
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5428347587585449,
+      "learning_rate": 0.0007790700653515323,
+      "loss": 0.2523,
+      "step": 796
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.567125141620636,
+      "learning_rate": 0.0007785432293301806,
+      "loss": 0.3316,
+      "step": 797
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8222817778587341,
+      "learning_rate": 0.0007780159445201563,
+      "loss": 0.2932,
+      "step": 798
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.9620918035507202,
+      "learning_rate": 0.0007774882117710202,
+      "loss": 0.3579,
+      "step": 799
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.101112961769104,
+      "learning_rate": 0.0007769600319330552,
+      "loss": 0.3888,
+      "step": 800
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7444469332695007,
+      "learning_rate": 0.000776431405857264,
+      "loss": 0.3248,
+      "step": 801
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.526674747467041,
+      "learning_rate": 0.0007759023343953688,
+      "loss": 0.3975,
+      "step": 802
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.744870126247406,
+      "learning_rate": 0.0007753728183998093,
+      "loss": 0.296,
+      "step": 803
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7272823452949524,
+      "learning_rate": 0.0007748428587237412,
+      "loss": 0.44,
+      "step": 804
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7115190029144287,
+      "learning_rate": 0.0007743124562210351,
+      "loss": 0.289,
+      "step": 805
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.3182278871536255,
+      "learning_rate": 0.0007737816117462751,
+      "loss": 0.2631,
+      "step": 806
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8276879191398621,
+      "learning_rate": 0.0007732503261547579,
+      "loss": 0.3543,
+      "step": 807
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.889427900314331,
+      "learning_rate": 0.00077271860030249,
+      "loss": 0.3578,
+      "step": 808
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5320945382118225,
+      "learning_rate": 0.0007721864350461882,
+      "loss": 0.2123,
+      "step": 809
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.768183708190918,
+      "learning_rate": 0.0007716538312432765,
+      "loss": 0.4439,
+      "step": 810
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4858362376689911,
+      "learning_rate": 0.0007711207897518861,
+      "loss": 0.2459,
+      "step": 811
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.649494469165802,
+      "learning_rate": 0.0007705873114308528,
+      "loss": 0.3436,
+      "step": 812
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8093873858451843,
+      "learning_rate": 0.0007700533971397165,
+      "loss": 0.4,
+      "step": 813
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6140373349189758,
+      "learning_rate": 0.00076951904773872,
+      "loss": 0.3901,
+      "step": 814
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.74479740858078,
+      "learning_rate": 0.0007689842640888063,
+      "loss": 0.6698,
+      "step": 815
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7307195067405701,
+      "learning_rate": 0.0007684490470516185,
+      "loss": 0.4366,
+      "step": 816
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6703787446022034,
+      "learning_rate": 0.0007679133974894983,
+      "loss": 0.2873,
+      "step": 817
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8069244623184204,
+      "learning_rate": 0.0007673773162654836,
+      "loss": 0.5054,
+      "step": 818
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.711532473564148,
+      "learning_rate": 0.0007668408042433082,
+      "loss": 0.5735,
+      "step": 819
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.2755175828933716,
+      "learning_rate": 0.0007663038622873999,
+      "loss": 0.5877,
+      "step": 820
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5795337557792664,
+      "learning_rate": 0.0007657664912628794,
+      "loss": 0.2906,
+      "step": 821
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5580657124519348,
+      "learning_rate": 0.0007652286920355584,
+      "loss": 0.3486,
+      "step": 822
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8551422357559204,
+      "learning_rate": 0.0007646904654719385,
+      "loss": 0.5183,
+      "step": 823
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5749680399894714,
+      "learning_rate": 0.0007641518124392104,
+      "loss": 0.2185,
+      "step": 824
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.2199212312698364,
+      "learning_rate": 0.0007636127338052513,
+      "loss": 0.36,
+      "step": 825
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6131826639175415,
+      "learning_rate": 0.0007630732304386243,
+      "loss": 0.4292,
+      "step": 826
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.9495660662651062,
+      "learning_rate": 0.000762533303208577,
+      "loss": 0.3827,
+      "step": 827
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.941986620426178,
+      "learning_rate": 0.0007619929529850397,
+      "loss": 0.4047,
+      "step": 828
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6457249522209167,
+      "learning_rate": 0.0007614521806386243,
+      "loss": 0.3639,
+      "step": 829
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.9594531655311584,
+      "learning_rate": 0.000760910987040623,
+      "loss": 0.2433,
+      "step": 830
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8174279928207397,
+      "learning_rate": 0.0007603693730630066,
+      "loss": 0.4556,
+      "step": 831
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.43721091747283936,
+      "learning_rate": 0.000759827339578423,
+      "loss": 0.2289,
+      "step": 832
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.2566089630126953,
+      "learning_rate": 0.0007592848874601963,
+      "loss": 0.537,
+      "step": 833
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.554829478263855,
+      "learning_rate": 0.0007587420175823252,
+      "loss": 0.3415,
+      "step": 834
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5715281367301941,
+      "learning_rate": 0.000758198730819481,
+      "loss": 0.2604,
+      "step": 835
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.0165090560913086,
+      "learning_rate": 0.0007576550280470072,
+      "loss": 0.5407,
+      "step": 836
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7614321708679199,
+      "learning_rate": 0.000757110910140917,
+      "loss": 0.3999,
+      "step": 837
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7295865416526794,
+      "learning_rate": 0.0007565663779778933,
+      "loss": 0.3213,
+      "step": 838
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3755558133125305,
+      "learning_rate": 0.0007560214324352858,
+      "loss": 0.1316,
+      "step": 839
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7876707315444946,
+      "learning_rate": 0.0007554760743911103,
+      "loss": 0.336,
+      "step": 840
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7146762013435364,
+      "learning_rate": 0.0007549303047240474,
+      "loss": 0.4322,
+      "step": 841
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.1264396905899048,
+      "learning_rate": 0.0007543841243134408,
+      "loss": 0.5366,
+      "step": 842
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.8722646832466125,
+      "learning_rate": 0.0007538375340392961,
+      "loss": 0.312,
+      "step": 843
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 12.184880256652832,
+      "learning_rate": 0.0007532905347822791,
+      "loss": 0.6477,
+      "step": 844
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6271510124206543,
+      "learning_rate": 0.0007527431274237149,
+      "loss": 0.4178,
+      "step": 845
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6101604104042053,
+      "learning_rate": 0.0007521953128455855,
+      "loss": 0.4089,
+      "step": 846
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7042468786239624,
+      "learning_rate": 0.0007516470919305298,
+      "loss": 0.5557,
+      "step": 847
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4274675250053406,
+      "learning_rate": 0.0007510984655618407,
+      "loss": 0.2305,
+      "step": 848
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.546370267868042,
+      "learning_rate": 0.0007505494346234647,
+      "loss": 0.319,
+      "step": 849
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 3.6892476081848145,
+      "learning_rate": 0.00075,
+      "loss": 0.3993,
+      "step": 850
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7396899461746216,
+      "learning_rate": 0.0007494501625766956,
+      "loss": 0.3209,
+      "step": 851
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6485087275505066,
+      "learning_rate": 0.0007488999232394491,
+      "loss": 0.406,
+      "step": 852
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7173447012901306,
+      "learning_rate": 0.0007483492828748057,
+      "loss": 0.4093,
+      "step": 853
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.502258837223053,
+      "learning_rate": 0.0007477982423699567,
+      "loss": 0.2633,
+      "step": 854
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5773262977600098,
+      "learning_rate": 0.0007472468026127384,
+      "loss": 0.3583,
+      "step": 855
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6986655592918396,
+      "learning_rate": 0.0007466949644916301,
+      "loss": 0.305,
+      "step": 856
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7234258055686951,
+      "learning_rate": 0.0007461427288957532,
+      "loss": 0.3317,
+      "step": 857
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7346418499946594,
+      "learning_rate": 0.000745590096714869,
+      "loss": 0.361,
+      "step": 858
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5990205407142639,
+      "learning_rate": 0.0007450370688393784,
+      "loss": 0.3798,
+      "step": 859
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.97529536485672,
+      "learning_rate": 0.0007444836461603195,
+      "loss": 0.3943,
+      "step": 860
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.8161059617996216,
+      "learning_rate": 0.0007439298295693664,
+      "loss": 0.2566,
+      "step": 861
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.0157384872436523,
+      "learning_rate": 0.0007433756199588282,
+      "loss": 0.4114,
+      "step": 862
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.1874598264694214,
+      "learning_rate": 0.000742821018221647,
+      "loss": 0.514,
+      "step": 863
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5575714111328125,
+      "learning_rate": 0.0007422660252513969,
+      "loss": 0.4595,
+      "step": 864
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5573776960372925,
+      "learning_rate": 0.0007417106419422819,
+      "loss": 0.2152,
+      "step": 865
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.0349595546722412,
+      "learning_rate": 0.0007411548691891356,
+      "loss": 0.3434,
+      "step": 866
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5878648161888123,
+      "learning_rate": 0.0007405987078874186,
+      "loss": 0.283,
+      "step": 867
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5647603273391724,
+      "learning_rate": 0.0007400421589332175,
+      "loss": 0.3681,
+      "step": 868
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6117604374885559,
+      "learning_rate": 0.0007394852232232436,
+      "loss": 0.5036,
+      "step": 869
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.3398414850234985,
+      "learning_rate": 0.0007389279016548316,
+      "loss": 0.3803,
+      "step": 870
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4927322566509247,
+      "learning_rate": 0.0007383701951259375,
+      "loss": 0.3408,
+      "step": 871
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7869893908500671,
+      "learning_rate": 0.0007378121045351377,
+      "loss": 0.5704,
+      "step": 872
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5877483487129211,
+      "learning_rate": 0.0007372536307816273,
+      "loss": 0.3334,
+      "step": 873
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.8796885013580322,
+      "learning_rate": 0.0007366947747652191,
+      "loss": 0.4896,
+      "step": 874
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.906600296497345,
+      "learning_rate": 0.0007361355373863414,
+      "loss": 0.3321,
+      "step": 875
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6890631318092346,
+      "learning_rate": 0.0007355759195460371,
+      "loss": 0.3783,
+      "step": 876
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5602468848228455,
+      "learning_rate": 0.0007350159221459621,
+      "loss": 0.2623,
+      "step": 877
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7423787117004395,
+      "learning_rate": 0.0007344555460883839,
+      "loss": 0.4837,
+      "step": 878
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7250906229019165,
+      "learning_rate": 0.0007338947922761802,
+      "loss": 0.3208,
+      "step": 879
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.39499133825302124,
+      "learning_rate": 0.0007333336616128369,
+      "loss": 0.1874,
+      "step": 880
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6592676639556885,
+      "learning_rate": 0.0007327721550024475,
+      "loss": 0.3216,
+      "step": 881
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.8230254650115967,
+      "learning_rate": 0.000732210273349711,
+      "loss": 0.3677,
+      "step": 882
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.9845200777053833,
+      "learning_rate": 0.0007316480175599309,
+      "loss": 0.6492,
+      "step": 883
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.8981730937957764,
+      "learning_rate": 0.0007310853885390133,
+      "loss": 0.5073,
+      "step": 884
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6771971583366394,
+      "learning_rate": 0.0007305223871934656,
+      "loss": 0.1609,
+      "step": 885
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6244550943374634,
+      "learning_rate": 0.0007299590144303954,
+      "loss": 0.2335,
+      "step": 886
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.674512505531311,
+      "learning_rate": 0.0007293952711575086,
+      "loss": 0.2101,
+      "step": 887
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6622495651245117,
+      "learning_rate": 0.0007288311582831077,
+      "loss": 0.1537,
+      "step": 888
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6100170016288757,
+      "learning_rate": 0.0007282666767160912,
+      "loss": 0.2535,
+      "step": 889
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7179614901542664,
+      "learning_rate": 0.0007277018273659516,
+      "loss": 0.4576,
+      "step": 890
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.641441822052002,
+      "learning_rate": 0.0007271366111427734,
+      "loss": 0.3267,
+      "step": 891
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7164905071258545,
+      "learning_rate": 0.0007265710289572328,
+      "loss": 0.3561,
+      "step": 892
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8796059489250183,
+      "learning_rate": 0.0007260050817205955,
+      "loss": 0.4087,
+      "step": 893
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6368707418441772,
+      "learning_rate": 0.0007254387703447154,
+      "loss": 0.5568,
+      "step": 894
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 9.814650535583496,
+      "learning_rate": 0.0007248720957420329,
+      "loss": 0.5917,
+      "step": 895
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8121230006217957,
+      "learning_rate": 0.0007243050588255737,
+      "loss": 0.3844,
+      "step": 896
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.705636203289032,
+      "learning_rate": 0.0007237376605089477,
+      "loss": 0.455,
+      "step": 897
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7467210292816162,
+      "learning_rate": 0.000723169901706346,
+      "loss": 0.2854,
+      "step": 898
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.0960447788238525,
+      "learning_rate": 0.000722601783332542,
+      "loss": 0.3752,
+      "step": 899
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.471336305141449,
+      "learning_rate": 0.0007220333063028871,
+      "loss": 0.2488,
+      "step": 900
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6433385014533997,
+      "learning_rate": 0.0007214644715333114,
+      "loss": 0.3582,
+      "step": 901
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6260262131690979,
+      "learning_rate": 0.000720895279940321,
+      "loss": 0.3247,
+      "step": 902
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4928528666496277,
+      "learning_rate": 0.0007203257324409971,
+      "loss": 0.3257,
+      "step": 903
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6702114343643188,
+      "learning_rate": 0.000719755829952994,
+      "loss": 0.2809,
+      "step": 904
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.2252200841903687,
+      "learning_rate": 0.0007191855733945387,
+      "loss": 0.3685,
+      "step": 905
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.0061269998550415,
+      "learning_rate": 0.000718614963684428,
+      "loss": 0.5586,
+      "step": 906
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5588775277137756,
+      "learning_rate": 0.0007180440017420277,
+      "loss": 0.237,
+      "step": 907
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6254746317863464,
+      "learning_rate": 0.0007174726884872716,
+      "loss": 0.2666,
+      "step": 908
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.0091012716293335,
+      "learning_rate": 0.0007169010248406589,
+      "loss": 0.2748,
+      "step": 909
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9338001608848572,
+      "learning_rate": 0.0007163290117232541,
+      "loss": 0.2185,
+      "step": 910
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.48531028628349304,
+      "learning_rate": 0.0007157566500566843,
+      "loss": 0.24,
+      "step": 911
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7651498317718506,
+      "learning_rate": 0.000715183940763138,
+      "loss": 0.3565,
+      "step": 912
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4688432812690735,
+      "learning_rate": 0.0007146108847653641,
+      "loss": 0.2107,
+      "step": 913
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8010238409042358,
+      "learning_rate": 0.0007140374829866702,
+      "loss": 0.3548,
+      "step": 914
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5501894950866699,
+      "learning_rate": 0.0007134637363509209,
+      "loss": 0.2744,
+      "step": 915
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4454915225505829,
+      "learning_rate": 0.0007128896457825364,
+      "loss": 0.202,
+      "step": 916
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.0680900812149048,
+      "learning_rate": 0.0007123152122064909,
+      "loss": 0.3816,
+      "step": 917
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7587308287620544,
+      "learning_rate": 0.0007117404365483116,
+      "loss": 0.2774,
+      "step": 918
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.3480654954910278,
+      "learning_rate": 0.0007111653197340764,
+      "loss": 0.6987,
+      "step": 919
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.246838092803955,
+      "learning_rate": 0.0007105898626904133,
+      "loss": 0.3643,
+      "step": 920
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7033990621566772,
+      "learning_rate": 0.0007100140663444985,
+      "loss": 0.3704,
+      "step": 921
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5905662775039673,
+      "learning_rate": 0.0007094379316240544,
+      "loss": 0.2895,
+      "step": 922
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7017614841461182,
+      "learning_rate": 0.0007088614594573491,
+      "loss": 0.2395,
+      "step": 923
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2880537807941437,
+      "learning_rate": 0.0007082846507731941,
+      "loss": 0.1119,
+      "step": 924
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.0025699138641357,
+      "learning_rate": 0.0007077075065009433,
+      "loss": 0.5595,
+      "step": 925
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4431317150592804,
+      "learning_rate": 0.000707130027570491,
+      "loss": 0.186,
+      "step": 926
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5182144641876221,
+      "learning_rate": 0.000706552214912271,
+      "loss": 0.2601,
+      "step": 927
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6188678741455078,
+      "learning_rate": 0.0007059740694572545,
+      "loss": 0.2679,
+      "step": 928
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6590104103088379,
+      "learning_rate": 0.0007053955921369493,
+      "loss": 0.3128,
+      "step": 929
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6747900247573853,
+      "learning_rate": 0.0007048167838833977,
+      "loss": 0.8975,
+      "step": 930
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5337473750114441,
+      "learning_rate": 0.0007042376456291751,
+      "loss": 0.2287,
+      "step": 931
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.9899834990501404,
+      "learning_rate": 0.0007036581783073887,
+      "loss": 0.3741,
+      "step": 932
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6095447540283203,
+      "learning_rate": 0.0007030783828516759,
+      "loss": 0.3395,
+      "step": 933
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7213789224624634,
+      "learning_rate": 0.0007024982601962027,
+      "loss": 0.371,
+      "step": 934
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.1294257640838623,
+      "learning_rate": 0.0007019178112756625,
+      "loss": 0.5472,
+      "step": 935
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7760403752326965,
+      "learning_rate": 0.0007013370370252739,
+      "loss": 0.323,
+      "step": 936
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5592139363288879,
+      "learning_rate": 0.0007007559383807804,
+      "loss": 0.3818,
+      "step": 937
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4886627793312073,
+      "learning_rate": 0.0007001745162784476,
+      "loss": 0.2675,
+      "step": 938
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.562885046005249,
+      "learning_rate": 0.0006995927716550623,
+      "loss": 0.2206,
+      "step": 939
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4581106901168823,
+      "learning_rate": 0.0006990107054479312,
+      "loss": 0.2054,
+      "step": 940
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7962608337402344,
+      "learning_rate": 0.000698428318594879,
+      "loss": 0.4223,
+      "step": 941
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4981114864349365,
+      "learning_rate": 0.0006978456120342468,
+      "loss": 0.178,
+      "step": 942
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6693292260169983,
+      "learning_rate": 0.0006972625867048914,
+      "loss": 0.2234,
+      "step": 943
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5983363389968872,
+      "learning_rate": 0.0006966792435461827,
+      "loss": 0.3419,
+      "step": 944
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.9164005517959595,
+      "learning_rate": 0.0006960955834980027,
+      "loss": 0.4963,
+      "step": 945
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6697097420692444,
+      "learning_rate": 0.0006955116075007443,
+      "loss": 0.4033,
+      "step": 946
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6218885183334351,
+      "learning_rate": 0.000694927316495309,
+      "loss": 0.267,
+      "step": 947
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.1508067846298218,
+      "learning_rate": 0.0006943427114231063,
+      "loss": 0.2959,
+      "step": 948
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6373894810676575,
+      "learning_rate": 0.0006937577932260515,
+      "loss": 0.3024,
+      "step": 949
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6286529898643494,
+      "learning_rate": 0.0006931725628465643,
+      "loss": 0.2992,
+      "step": 950
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8749977350234985,
+      "learning_rate": 0.0006925870212275676,
+      "loss": 0.3954,
+      "step": 951
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5200191140174866,
+      "learning_rate": 0.0006920011693124857,
+      "loss": 0.213,
+      "step": 952
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6564799547195435,
+      "learning_rate": 0.0006914150080452428,
+      "loss": 0.3027,
+      "step": 953
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6447715163230896,
+      "learning_rate": 0.0006908285383702616,
+      "loss": 0.2473,
+      "step": 954
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.540440559387207,
+      "learning_rate": 0.0006902417612324615,
+      "loss": 0.357,
+      "step": 955
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.36145466566085815,
+      "learning_rate": 0.0006896546775772576,
+      "loss": 0.1948,
+      "step": 956
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.790409505367279,
+      "learning_rate": 0.0006890672883505588,
+      "loss": 0.4841,
+      "step": 957
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.552435576915741,
+      "learning_rate": 0.0006884795944987661,
+      "loss": 0.3725,
+      "step": 958
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.48317885398864746,
+      "learning_rate": 0.0006878915969687714,
+      "loss": 0.1892,
+      "step": 959
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8355458974838257,
+      "learning_rate": 0.0006873032967079561,
+      "loss": 0.4814,
+      "step": 960
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7481607794761658,
+      "learning_rate": 0.0006867146946641891,
+      "loss": 0.2625,
+      "step": 961
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7298922538757324,
+      "learning_rate": 0.0006861257917858257,
+      "loss": 0.328,
+      "step": 962
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3932209014892578,
+      "learning_rate": 0.0006855365890217057,
+      "loss": 0.3029,
+      "step": 963
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6357635259628296,
+      "learning_rate": 0.0006849470873211522,
+      "loss": 0.3682,
+      "step": 964
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8700315952301025,
+      "learning_rate": 0.0006843572876339704,
+      "loss": 0.4866,
+      "step": 965
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7602806091308594,
+      "learning_rate": 0.0006837671909104447,
+      "loss": 0.3381,
+      "step": 966
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8669351935386658,
+      "learning_rate": 0.0006831767981013388,
+      "loss": 0.3958,
+      "step": 967
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5393077731132507,
+      "learning_rate": 0.0006825861101578931,
+      "loss": 0.2087,
+      "step": 968
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.43480342626571655,
+      "learning_rate": 0.0006819951280318237,
+      "loss": 0.2586,
+      "step": 969
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9570690393447876,
+      "learning_rate": 0.0006814038526753205,
+      "loss": 0.6008,
+      "step": 970
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7020831108093262,
+      "learning_rate": 0.000680812285041046,
+      "loss": 0.4712,
+      "step": 971
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5883210301399231,
+      "learning_rate": 0.0006802204260821339,
+      "loss": 0.3104,
+      "step": 972
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7088538408279419,
+      "learning_rate": 0.0006796282767521869,
+      "loss": 0.2714,
+      "step": 973
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5097481608390808,
+      "learning_rate": 0.0006790358380052751,
+      "loss": 0.2626,
+      "step": 974
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5309572219848633,
+      "learning_rate": 0.0006784431107959359,
+      "loss": 0.2939,
+      "step": 975
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6540642976760864,
+      "learning_rate": 0.0006778500960791709,
+      "loss": 0.3523,
+      "step": 976
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6825799942016602,
+      "learning_rate": 0.0006772567948104452,
+      "loss": 0.4607,
+      "step": 977
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4674563705921173,
+      "learning_rate": 0.0006766632079456852,
+      "loss": 0.249,
+      "step": 978
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.629218339920044,
+      "learning_rate": 0.0006760693364412776,
+      "loss": 0.3355,
+      "step": 979
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7523155808448792,
+      "learning_rate": 0.0006754751812540679,
+      "loss": 0.5164,
+      "step": 980
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5138117074966431,
+      "learning_rate": 0.0006748807433413586,
+      "loss": 0.2989,
+      "step": 981
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.42316293716430664,
+      "learning_rate": 0.0006742860236609076,
+      "loss": 0.1871,
+      "step": 982
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6735039353370667,
+      "learning_rate": 0.000673691023170927,
+      "loss": 0.3488,
+      "step": 983
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5802951455116272,
+      "learning_rate": 0.0006730957428300813,
+      "loss": 0.4043,
+      "step": 984
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8631969094276428,
+      "learning_rate": 0.0006725001835974853,
+      "loss": 0.3532,
+      "step": 985
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4295097291469574,
+      "learning_rate": 0.0006719043464327042,
+      "loss": 0.2798,
+      "step": 986
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8754927515983582,
+      "learning_rate": 0.0006713082322957503,
+      "loss": 0.4136,
+      "step": 987
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3174607455730438,
+      "learning_rate": 0.0006707118421470822,
+      "loss": 0.1687,
+      "step": 988
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.1024227142333984,
+      "learning_rate": 0.0006701151769476033,
+      "loss": 0.1626,
+      "step": 989
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6028107404708862,
+      "learning_rate": 0.0006695182376586602,
+      "loss": 0.3035,
+      "step": 990
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5161485075950623,
+      "learning_rate": 0.0006689210252420415,
+      "loss": 0.3299,
+      "step": 991
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5899966359138489,
+      "learning_rate": 0.0006683235406599749,
+      "loss": 0.3688,
+      "step": 992
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.9813715219497681,
+      "learning_rate": 0.0006677257848751275,
+      "loss": 0.4816,
+      "step": 993
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6310505867004395,
+      "learning_rate": 0.0006671277588506029,
+      "loss": 0.2877,
+      "step": 994
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5029193758964539,
+      "learning_rate": 0.0006665294635499404,
+      "loss": 0.2528,
+      "step": 995
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4148341119289398,
+      "learning_rate": 0.000665930899937113,
+      "loss": 0.2259,
+      "step": 996
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.41644999384880066,
+      "learning_rate": 0.0006653320689765257,
+      "loss": 0.1825,
+      "step": 997
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.697913646697998,
+      "learning_rate": 0.0006647329716330148,
+      "loss": 0.2706,
+      "step": 998
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.739029049873352,
+      "learning_rate": 0.0006641336088718456,
+      "loss": 0.3116,
+      "step": 999
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6200854182243347,
+      "learning_rate": 0.0006635339816587109,
+      "loss": 0.2932,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6857175827026367,
+      "learning_rate": 0.0006629340909597297,
+      "loss": 0.3276,
+      "step": 1001
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.44770586490631104,
+      "learning_rate": 0.0006623339377414456,
+      "loss": 0.2501,
+      "step": 1002
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8567302823066711,
+      "learning_rate": 0.0006617335229708249,
+      "loss": 0.4944,
+      "step": 1003
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7527105808258057,
+      "learning_rate": 0.0006611328476152556,
+      "loss": 0.2944,
+      "step": 1004
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.623694121837616,
+      "learning_rate": 0.0006605319126425454,
+      "loss": 0.4682,
+      "step": 1005
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6465189456939697,
+      "learning_rate": 0.0006599307190209205,
+      "loss": 0.2177,
+      "step": 1006
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5936398506164551,
+      "learning_rate": 0.0006593292677190234,
+      "loss": 0.3225,
+      "step": 1007
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3749469816684723,
+      "learning_rate": 0.0006587275597059125,
+      "loss": 0.1401,
+      "step": 1008
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4972422122955322,
+      "learning_rate": 0.000658125595951059,
+      "loss": 0.2089,
+      "step": 1009
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6559800505638123,
+      "learning_rate": 0.0006575233774243465,
+      "loss": 0.3401,
+      "step": 1010
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5021636486053467,
+      "learning_rate": 0.0006569209050960691,
+      "loss": 0.2253,
+      "step": 1011
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5802257657051086,
+      "learning_rate": 0.0006563181799369301,
+      "loss": 0.224,
+      "step": 1012
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6598565578460693,
+      "learning_rate": 0.0006557152029180398,
+      "loss": 0.1903,
+      "step": 1013
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5926521420478821,
+      "learning_rate": 0.0006551119750109141,
+      "loss": 0.3264,
+      "step": 1014
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.2590575218200684,
+      "learning_rate": 0.0006545084971874737,
+      "loss": 0.9175,
+      "step": 1015
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3434202969074249,
+      "learning_rate": 0.0006539047704200417,
+      "loss": 0.1696,
+      "step": 1016
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.37384888529777527,
+      "learning_rate": 0.000653300795681342,
+      "loss": 0.1739,
+      "step": 1017
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9205343127250671,
+      "learning_rate": 0.0006526965739444987,
+      "loss": 0.4146,
+      "step": 1018
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.710533082485199,
+      "learning_rate": 0.0006520921061830332,
+      "loss": 0.3488,
+      "step": 1019
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.1363492012023926,
+      "learning_rate": 0.0006514873933708637,
+      "loss": 0.4286,
+      "step": 1020
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6555688977241516,
+      "learning_rate": 0.0006508824364823031,
+      "loss": 0.3362,
+      "step": 1021
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.705757200717926,
+      "learning_rate": 0.0006502772364920573,
+      "loss": 0.3078,
+      "step": 1022
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8903169631958008,
+      "learning_rate": 0.0006496717943752243,
+      "loss": 0.4568,
+      "step": 1023
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6346259117126465,
+      "learning_rate": 0.0006490661111072923,
+      "loss": 0.2078,
+      "step": 1024
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9185580611228943,
+      "learning_rate": 0.0006484601876641375,
+      "loss": 0.3738,
+      "step": 1025
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.810040295124054,
+      "learning_rate": 0.0006478540250220234,
+      "loss": 0.3349,
+      "step": 1026
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6751656532287598,
+      "learning_rate": 0.0006472476241575989,
+      "loss": 0.3282,
+      "step": 1027
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1449875831604004,
+      "learning_rate": 0.0006466409860478966,
+      "loss": 0.2653,
+      "step": 1028
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6683035492897034,
+      "learning_rate": 0.0006460341116703317,
+      "loss": 0.3304,
+      "step": 1029
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7053085565567017,
+      "learning_rate": 0.0006454270020026995,
+      "loss": 0.3319,
+      "step": 1030
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7800212502479553,
+      "learning_rate": 0.0006448196580231748,
+      "loss": 0.3518,
+      "step": 1031
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.687646746635437,
+      "learning_rate": 0.0006442120807103101,
+      "loss": 0.4122,
+      "step": 1032
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.787562370300293,
+      "learning_rate": 0.0006436042710430332,
+      "loss": 0.4071,
+      "step": 1033
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4035986363887787,
+      "learning_rate": 0.0006429962300006467,
+      "loss": 0.1494,
+      "step": 1034
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5613333582878113,
+      "learning_rate": 0.0006423879585628261,
+      "loss": 0.3232,
+      "step": 1035
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6273500323295593,
+      "learning_rate": 0.0006417794577096179,
+      "loss": 0.3071,
+      "step": 1036
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.006776213645935,
+      "learning_rate": 0.0006411707284214383,
+      "loss": 0.6485,
+      "step": 1037
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7093353271484375,
+      "learning_rate": 0.0006405617716790713,
+      "loss": 0.3788,
+      "step": 1038
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7205580472946167,
+      "learning_rate": 0.0006399525884636681,
+      "loss": 0.4857,
+      "step": 1039
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6377617120742798,
+      "learning_rate": 0.0006393431797567439,
+      "loss": 0.5022,
+      "step": 1040
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.9162612557411194,
+      "learning_rate": 0.0006387335465401776,
+      "loss": 0.2979,
+      "step": 1041
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.2079604864120483,
+      "learning_rate": 0.0006381236897962102,
+      "loss": 0.2724,
+      "step": 1042
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.44193050265312195,
+      "learning_rate": 0.0006375136105074422,
+      "loss": 0.2158,
+      "step": 1043
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.81155925989151,
+      "learning_rate": 0.0006369033096568329,
+      "loss": 0.5243,
+      "step": 1044
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4662119448184967,
+      "learning_rate": 0.0006362927882276989,
+      "loss": 0.2229,
+      "step": 1045
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.46297112107276917,
+      "learning_rate": 0.0006356820472037119,
+      "loss": 0.1577,
+      "step": 1046
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7655563354492188,
+      "learning_rate": 0.0006350710875688972,
+      "loss": 0.3543,
+      "step": 1047
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7138177752494812,
+      "learning_rate": 0.0006344599103076329,
+      "loss": 0.2724,
+      "step": 1048
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8312786221504211,
+      "learning_rate": 0.0006338485164046471,
+      "loss": 0.3259,
+      "step": 1049
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6465335488319397,
+      "learning_rate": 0.0006332369068450174,
+      "loss": 0.3343,
+      "step": 1050
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7901415824890137,
+      "learning_rate": 0.0006326250826141688,
+      "loss": 0.4592,
+      "step": 1051
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7137051224708557,
+      "learning_rate": 0.0006320130446978722,
+      "loss": 0.2757,
+      "step": 1052
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7401980757713318,
+      "learning_rate": 0.0006314007940822425,
+      "loss": 0.2634,
+      "step": 1053
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.42048975825309753,
+      "learning_rate": 0.0006307883317537375,
+      "loss": 0.2208,
+      "step": 1054
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 9.062432289123535,
+      "learning_rate": 0.0006301756586991561,
+      "loss": 0.2374,
+      "step": 1055
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5630722045898438,
+      "learning_rate": 0.0006295627759056368,
+      "loss": 0.2619,
+      "step": 1056
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8039603233337402,
+      "learning_rate": 0.0006289496843606559,
+      "loss": 0.3318,
+      "step": 1057
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5242823362350464,
+      "learning_rate": 0.0006283363850520262,
+      "loss": 0.1903,
+      "step": 1058
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.0117770433425903,
+      "learning_rate": 0.0006277228789678953,
+      "loss": 0.4221,
+      "step": 1059
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5779409408569336,
+      "learning_rate": 0.0006271091670967436,
+      "loss": 0.2959,
+      "step": 1060
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5154205560684204,
+      "learning_rate": 0.0006264952504273831,
+      "loss": 0.1525,
+      "step": 1061
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.47642436623573303,
+      "learning_rate": 0.0006258811299489563,
+      "loss": 0.1893,
+      "step": 1062
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6867002248764038,
+      "learning_rate": 0.0006252668066509335,
+      "loss": 0.3566,
+      "step": 1063
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.49086835980415344,
+      "learning_rate": 0.0006246522815231121,
+      "loss": 0.2622,
+      "step": 1064
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7697747945785522,
+      "learning_rate": 0.0006240375555556145,
+      "loss": 0.3574,
+      "step": 1065
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7349634170532227,
+      "learning_rate": 0.0006234226297388869,
+      "loss": 0.3951,
+      "step": 1066
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5543031692504883,
+      "learning_rate": 0.0006228075050636972,
+      "loss": 0.1921,
+      "step": 1067
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.436019629240036,
+      "learning_rate": 0.0006221921825211342,
+      "loss": 0.2155,
+      "step": 1068
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7876712679862976,
+      "learning_rate": 0.0006215766631026048,
+      "loss": 0.2347,
+      "step": 1069
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5204719305038452,
+      "learning_rate": 0.0006209609477998338,
+      "loss": 0.2291,
+      "step": 1070
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.0805472135543823,
+      "learning_rate": 0.0006203450376048614,
+      "loss": 0.417,
+      "step": 1071
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6691756248474121,
+      "learning_rate": 0.0006197289335100412,
+      "loss": 0.2576,
+      "step": 1072
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5985954999923706,
+      "learning_rate": 0.0006191126365080402,
+      "loss": 0.2914,
+      "step": 1073
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4913211464881897,
+      "learning_rate": 0.0006184961475918355,
+      "loss": 0.2108,
+      "step": 1074
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7180914282798767,
+      "learning_rate": 0.0006178794677547138,
+      "loss": 0.4348,
+      "step": 1075
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6727098822593689,
+      "learning_rate": 0.000617262597990269,
+      "loss": 0.2813,
+      "step": 1076
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6296001672744751,
+      "learning_rate": 0.0006166455392924014,
+      "loss": 0.2873,
+      "step": 1077
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4847506284713745,
+      "learning_rate": 0.0006160282926553158,
+      "loss": 0.2055,
+      "step": 1078
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.49828800559043884,
+      "learning_rate": 0.0006154108590735193,
+      "loss": 0.2938,
+      "step": 1079
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5496594309806824,
+      "learning_rate": 0.0006147932395418205,
+      "loss": 0.272,
+      "step": 1080
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.2711024284362793,
+      "learning_rate": 0.0006141754350553279,
+      "loss": 0.1236,
+      "step": 1081
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5817246437072754,
+      "learning_rate": 0.0006135574466094475,
+      "loss": 0.2972,
+      "step": 1082
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7663928270339966,
+      "learning_rate": 0.0006129392751998817,
+      "loss": 0.4816,
+      "step": 1083
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7487969398498535,
+      "learning_rate": 0.0006123209218226281,
+      "loss": 0.3031,
+      "step": 1084
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7048330307006836,
+      "learning_rate": 0.0006117023874739772,
+      "loss": 0.2627,
+      "step": 1085
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4851953089237213,
+      "learning_rate": 0.0006110836731505111,
+      "loss": 0.2327,
+      "step": 1086
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6638092398643494,
+      "learning_rate": 0.0006104647798491021,
+      "loss": 0.5013,
+      "step": 1087
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.9075124263763428,
+      "learning_rate": 0.0006098457085669105,
+      "loss": 0.579,
+      "step": 1088
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7261365652084351,
+      "learning_rate": 0.0006092264603013836,
+      "loss": 0.4446,
+      "step": 1089
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5950862765312195,
+      "learning_rate": 0.0006086070360502539,
+      "loss": 0.3303,
+      "step": 1090
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.680767834186554,
+      "learning_rate": 0.0006079874368115373,
+      "loss": 0.3472,
+      "step": 1091
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4536004364490509,
+      "learning_rate": 0.0006073676635835317,
+      "loss": 0.2564,
+      "step": 1092
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6102024912834167,
+      "learning_rate": 0.0006067477173648153,
+      "loss": 0.2334,
+      "step": 1093
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7941223382949829,
+      "learning_rate": 0.000606127599154245,
+      "loss": 0.3999,
+      "step": 1094
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5325322151184082,
+      "learning_rate": 0.0006055073099509549,
+      "loss": 0.2069,
+      "step": 1095
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5478134751319885,
+      "learning_rate": 0.0006048868507543547,
+      "loss": 0.3168,
+      "step": 1096
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.41208842396736145,
+      "learning_rate": 0.0006042662225641276,
+      "loss": 0.2102,
+      "step": 1097
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8149394392967224,
+      "learning_rate": 0.0006036454263802297,
+      "loss": 0.4137,
+      "step": 1098
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.45352303981781006,
+      "learning_rate": 0.000603024463202887,
+      "loss": 0.1989,
+      "step": 1099
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4395929276943207,
+      "learning_rate": 0.0006024033340325954,
+      "loss": 0.1931,
+      "step": 1100
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7390046119689941,
+      "learning_rate": 0.0006017820398701174,
+      "loss": 0.4257,
+      "step": 1101
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.1061534881591797,
+      "learning_rate": 0.0006011605817164822,
+      "loss": 0.5192,
+      "step": 1102
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.1110053062438965,
+      "learning_rate": 0.0006005389605729824,
+      "loss": 0.2405,
+      "step": 1103
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5650463700294495,
+      "learning_rate": 0.0005999171774411737,
+      "loss": 0.155,
+      "step": 1104
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.9587744474411011,
+      "learning_rate": 0.0005992952333228728,
+      "loss": 0.2304,
+      "step": 1105
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.48374903202056885,
+      "learning_rate": 0.0005986731292201555,
+      "loss": 0.2443,
+      "step": 1106
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7335638999938965,
+      "learning_rate": 0.0005980508661353556,
+      "loss": 0.2879,
+      "step": 1107
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7504791617393494,
+      "learning_rate": 0.0005974284450710631,
+      "loss": 0.2635,
+      "step": 1108
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5602500438690186,
+      "learning_rate": 0.0005968058670301222,
+      "loss": 0.2894,
+      "step": 1109
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5755776166915894,
+      "learning_rate": 0.0005961831330156305,
+      "loss": 0.2117,
+      "step": 1110
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.45893555879592896,
+      "learning_rate": 0.0005955602440309365,
+      "loss": 0.2299,
+      "step": 1111
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8198021054267883,
+      "learning_rate": 0.0005949372010796383,
+      "loss": 0.3457,
+      "step": 1112
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.46037590503692627,
+      "learning_rate": 0.0005943140051655828,
+      "loss": 0.2684,
+      "step": 1113
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.3972904682159424,
+      "learning_rate": 0.0005936906572928624,
+      "loss": 0.397,
+      "step": 1114
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4555102586746216,
+      "learning_rate": 0.0005930671584658151,
+      "loss": 0.3316,
+      "step": 1115
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.32504335045814514,
+      "learning_rate": 0.0005924435096890216,
+      "loss": 0.2003,
+      "step": 1116
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5859614014625549,
+      "learning_rate": 0.0005918197119673046,
+      "loss": 0.3145,
+      "step": 1117
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7345462441444397,
+      "learning_rate": 0.0005911957663057263,
+      "loss": 0.3688,
+      "step": 1118
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5362555384635925,
+      "learning_rate": 0.0005905716737095879,
+      "loss": 0.2342,
+      "step": 1119
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6702203154563904,
+      "learning_rate": 0.000589947435184427,
+      "loss": 0.3797,
+      "step": 1120
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6967395544052124,
+      "learning_rate": 0.000589323051736016,
+      "loss": 0.4463,
+      "step": 1121
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5258362293243408,
+      "learning_rate": 0.0005886985243703612,
+      "loss": 0.2057,
+      "step": 1122
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5999728441238403,
+      "learning_rate": 0.0005880738540937008,
+      "loss": 0.3147,
+      "step": 1123
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5848231911659241,
+      "learning_rate": 0.0005874490419125033,
+      "loss": 0.2619,
+      "step": 1124
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.745381236076355,
+      "learning_rate": 0.0005868240888334653,
+      "loss": 0.4726,
+      "step": 1125
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6087539792060852,
+      "learning_rate": 0.0005861989958635109,
+      "loss": 0.4574,
+      "step": 1126
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5028085708618164,
+      "learning_rate": 0.0005855737640097897,
+      "loss": 0.2873,
+      "step": 1127
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4994693994522095,
+      "learning_rate": 0.0005849483942796747,
+      "loss": 0.2897,
+      "step": 1128
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8390994668006897,
+      "learning_rate": 0.0005843228876807613,
+      "loss": 0.1937,
+      "step": 1129
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7523295283317566,
+      "learning_rate": 0.0005836972452208654,
+      "loss": 0.4566,
+      "step": 1130
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.641046404838562,
+      "learning_rate": 0.0005830714679080215,
+      "loss": 0.3694,
+      "step": 1131
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5842282772064209,
+      "learning_rate": 0.0005824455567504817,
+      "loss": 0.4673,
+      "step": 1132
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6116966605186462,
+      "learning_rate": 0.0005818195127567136,
+      "loss": 0.309,
+      "step": 1133
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5021530389785767,
+      "learning_rate": 0.0005811933369353991,
+      "loss": 0.266,
+      "step": 1134
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4892033040523529,
+      "learning_rate": 0.0005805670302954321,
+      "loss": 0.2321,
+      "step": 1135
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5554322600364685,
+      "learning_rate": 0.0005799405938459175,
+      "loss": 0.2465,
+      "step": 1136
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.790320634841919,
+      "learning_rate": 0.0005793140285961692,
+      "loss": 0.3438,
+      "step": 1137
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7685120701789856,
+      "learning_rate": 0.0005786873355557089,
+      "loss": 0.3195,
+      "step": 1138
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3911312520503998,
+      "learning_rate": 0.0005780605157342641,
+      "loss": 0.1886,
+      "step": 1139
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7720545530319214,
+      "learning_rate": 0.0005774335701417662,
+      "loss": 0.3284,
+      "step": 1140
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.535527765750885,
+      "learning_rate": 0.0005768064997883497,
+      "loss": 0.1989,
+      "step": 1141
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4850839078426361,
+      "learning_rate": 0.00057617930568435,
+      "loss": 0.1878,
+      "step": 1142
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7033033967018127,
+      "learning_rate": 0.0005755519888403018,
+      "loss": 0.3514,
+      "step": 1143
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.682561993598938,
+      "learning_rate": 0.0005749245502669375,
+      "loss": 0.3662,
+      "step": 1144
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3520594835281372,
+      "learning_rate": 0.0005742969909751859,
+      "loss": 0.0995,
+      "step": 1145
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7896652817726135,
+      "learning_rate": 0.00057366931197617,
+      "loss": 0.4088,
+      "step": 1146
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.0011906623840332,
+      "learning_rate": 0.0005730415142812059,
+      "loss": 0.4352,
+      "step": 1147
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6589980721473694,
+      "learning_rate": 0.0005724135989018007,
+      "loss": 0.3361,
+      "step": 1148
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.1270843744277954,
+      "learning_rate": 0.0005717855668496513,
+      "loss": 0.3916,
+      "step": 1149
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7868853807449341,
+      "learning_rate": 0.0005711574191366427,
+      "loss": 0.3683,
+      "step": 1150
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5050237774848938,
+      "learning_rate": 0.0005705291567748459,
+      "loss": 0.2211,
+      "step": 1151
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.43563565611839294,
+      "learning_rate": 0.0005699007807765168,
+      "loss": 0.2213,
+      "step": 1152
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.558112382888794,
+      "learning_rate": 0.0005692722921540945,
+      "loss": 0.309,
+      "step": 1153
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4513910114765167,
+      "learning_rate": 0.0005686436919201996,
+      "loss": 0.2655,
+      "step": 1154
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6573309898376465,
+      "learning_rate": 0.0005680149810876322,
+      "loss": 0.2222,
+      "step": 1155
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5102400779724121,
+      "learning_rate": 0.0005673861606693707,
+      "loss": 0.2969,
+      "step": 1156
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5109910368919373,
+      "learning_rate": 0.0005667572316785705,
+      "loss": 0.2779,
+      "step": 1157
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4165787100791931,
+      "learning_rate": 0.0005661281951285613,
+      "loss": 0.2325,
+      "step": 1158
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5661754012107849,
+      "learning_rate": 0.0005654990520328465,
+      "loss": 0.2473,
+      "step": 1159
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.6070654392242432,
+      "learning_rate": 0.0005648698034051009,
+      "loss": 0.3427,
+      "step": 1160
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.727352499961853,
+      "learning_rate": 0.0005642404502591697,
+      "loss": 0.4848,
+      "step": 1161
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.602330207824707,
+      "learning_rate": 0.0005636109936090661,
+      "loss": 0.3468,
+      "step": 1162
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.9016145467758179,
+      "learning_rate": 0.0005629814344689705,
+      "loss": 0.3198,
+      "step": 1163
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5985903143882751,
+      "learning_rate": 0.000562351773853228,
+      "loss": 0.3764,
+      "step": 1164
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4263474643230438,
+      "learning_rate": 0.0005617220127763474,
+      "loss": 0.18,
+      "step": 1165
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5322577357292175,
+      "learning_rate": 0.0005610921522529994,
+      "loss": 0.2341,
+      "step": 1166
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.661536693572998,
+      "learning_rate": 0.0005604621932980147,
+      "loss": 0.3024,
+      "step": 1167
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5718533396720886,
+      "learning_rate": 0.0005598321369263829,
+      "loss": 0.3225,
+      "step": 1168
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6254765391349792,
+      "learning_rate": 0.0005592019841532506,
+      "loss": 0.2488,
+      "step": 1169
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7568292617797852,
+      "learning_rate": 0.0005585717359939192,
+      "loss": 0.2914,
+      "step": 1170
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6458844542503357,
+      "learning_rate": 0.0005579413934638444,
+      "loss": 0.1698,
+      "step": 1171
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.9667510986328125,
+      "learning_rate": 0.0005573109575786333,
+      "loss": 0.292,
+      "step": 1172
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8906057476997375,
+      "learning_rate": 0.0005566804293540443,
+      "loss": 0.3747,
+      "step": 1173
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5640904903411865,
+      "learning_rate": 0.0005560498098059837,
+      "loss": 0.3475,
+      "step": 1174
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6816082000732422,
+      "learning_rate": 0.0005554190999505056,
+      "loss": 0.3221,
+      "step": 1175
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5217597484588623,
+      "learning_rate": 0.000554788300803809,
+      "loss": 0.2479,
+      "step": 1176
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.516711950302124,
+      "learning_rate": 0.0005541574133822374,
+      "loss": 0.3157,
+      "step": 1177
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.1447367668151855,
+      "learning_rate": 0.0005535264387022759,
+      "loss": 0.2371,
+      "step": 1178
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.48891252279281616,
+      "learning_rate": 0.0005528953777805507,
+      "loss": 0.1903,
+      "step": 1179
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5924687385559082,
+      "learning_rate": 0.0005522642316338268,
+      "loss": 0.324,
+      "step": 1180
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8512484431266785,
+      "learning_rate": 0.0005516330012790062,
+      "loss": 0.3608,
+      "step": 1181
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4188278317451477,
+      "learning_rate": 0.0005510016877331271,
+      "loss": 0.1869,
+      "step": 1182
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7830476760864258,
+      "learning_rate": 0.0005503702920133614,
+      "loss": 0.2588,
+      "step": 1183
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.48327332735061646,
+      "learning_rate": 0.0005497388151370135,
+      "loss": 0.3356,
+      "step": 1184
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6262472867965698,
+      "learning_rate": 0.0005491072581215186,
+      "loss": 0.2921,
+      "step": 1185
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4407556653022766,
+      "learning_rate": 0.0005484756219844408,
+      "loss": 0.227,
+      "step": 1186
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4841013550758362,
+      "learning_rate": 0.0005478439077434719,
+      "loss": 0.286,
+      "step": 1187
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.596108615398407,
+      "learning_rate": 0.0005472121164164295,
+      "loss": 0.2544,
+      "step": 1188
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.472913533449173,
+      "learning_rate": 0.0005465802490212554,
+      "loss": 0.1894,
+      "step": 1189
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.37586137652397156,
+      "learning_rate": 0.0005459483065760138,
+      "loss": 0.1652,
+      "step": 1190
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4654616117477417,
+      "learning_rate": 0.0005453162900988901,
+      "loss": 0.119,
+      "step": 1191
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5111765265464783,
+      "learning_rate": 0.000544684200608189,
+      "loss": 0.1525,
+      "step": 1192
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4907207787036896,
+      "learning_rate": 0.0005440520391223323,
+      "loss": 0.2373,
+      "step": 1193
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3408985733985901,
+      "learning_rate": 0.0005434198066598584,
+      "loss": 0.1619,
+      "step": 1194
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5993598103523254,
+      "learning_rate": 0.0005427875042394199,
+      "loss": 0.3208,
+      "step": 1195
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3779894709587097,
+      "learning_rate": 0.000542155132879782,
+      "loss": 0.1663,
+      "step": 1196
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.405900776386261,
+      "learning_rate": 0.000541522693599821,
+      "loss": 0.2059,
+      "step": 1197
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6929959058761597,
+      "learning_rate": 0.0005408901874185226,
+      "loss": 0.2689,
+      "step": 1198
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.45032864809036255,
+      "learning_rate": 0.0005402576153549804,
+      "loss": 0.2118,
+      "step": 1199
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.636280357837677,
+      "learning_rate": 0.0005396249784283942,
+      "loss": 0.24,
+      "step": 1200
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5864680409431458,
+      "learning_rate": 0.0005389922776580682,
+      "loss": 0.3213,
+      "step": 1201
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7503579258918762,
+      "learning_rate": 0.0005383595140634093,
+      "loss": 0.483,
+      "step": 1202
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.642006516456604,
+      "learning_rate": 0.0005377266886639259,
+      "loss": 0.3626,
+      "step": 1203
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4304860830307007,
+      "learning_rate": 0.0005370938024792261,
+      "loss": 0.1779,
+      "step": 1204
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7060337066650391,
+      "learning_rate": 0.0005364608565290155,
+      "loss": 0.4376,
+      "step": 1205
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5312660336494446,
+      "learning_rate": 0.0005358278518330959,
+      "loss": 0.2754,
+      "step": 1206
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5639314651489258,
+      "learning_rate": 0.0005351947894113645,
+      "loss": 0.2266,
+      "step": 1207
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3418157398700714,
+      "learning_rate": 0.0005345616702838111,
+      "loss": 0.1661,
+      "step": 1208
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4649193584918976,
+      "learning_rate": 0.0005339284954705165,
+      "loss": 0.2371,
+      "step": 1209
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.48832380771636963,
+      "learning_rate": 0.000533295265991652,
+      "loss": 0.1857,
+      "step": 1210
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5488593578338623,
+      "learning_rate": 0.0005326619828674761,
+      "loss": 0.2879,
+      "step": 1211
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.34766995906829834,
+      "learning_rate": 0.0005320286471183343,
+      "loss": 0.1384,
+      "step": 1212
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.46045982837677,
+      "learning_rate": 0.0005313952597646568,
+      "loss": 0.2463,
+      "step": 1213
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5941422581672668,
+      "learning_rate": 0.0005307618218269569,
+      "loss": 0.373,
+      "step": 1214
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5424063205718994,
+      "learning_rate": 0.0005301283343258293,
+      "loss": 0.2206,
+      "step": 1215
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.676394522190094,
+      "learning_rate": 0.0005294947982819488,
+      "loss": 0.3758,
+      "step": 1216
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4859623610973358,
+      "learning_rate": 0.0005288612147160681,
+      "loss": 0.3241,
+      "step": 1217
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.075390100479126,
+      "learning_rate": 0.0005282275846490168,
+      "loss": 0.4276,
+      "step": 1218
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.9063329696655273,
+      "learning_rate": 0.0005275939091016992,
+      "loss": 0.353,
+      "step": 1219
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.40442991256713867,
+      "learning_rate": 0.000526960189095093,
+      "loss": 0.2104,
+      "step": 1220
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4046921133995056,
+      "learning_rate": 0.0005263264256502474,
+      "loss": 0.1738,
+      "step": 1221
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5788969397544861,
+      "learning_rate": 0.0005256926197882815,
+      "loss": 0.2863,
+      "step": 1222
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5465753674507141,
+      "learning_rate": 0.0005250587725303831,
+      "loss": 0.3564,
+      "step": 1223
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.880646288394928,
+      "learning_rate": 0.0005244248848978067,
+      "loss": 0.4183,
+      "step": 1224
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.1397321224212646,
+      "learning_rate": 0.0005237909579118712,
+      "loss": 0.5054,
+      "step": 1225
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.41163769364356995,
+      "learning_rate": 0.0005231569925939596,
+      "loss": 0.1253,
+      "step": 1226
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.43419840931892395,
+      "learning_rate": 0.0005225229899655162,
+      "loss": 0.2107,
+      "step": 1227
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4516807794570923,
+      "learning_rate": 0.0005218889510480461,
+      "loss": 0.2081,
+      "step": 1228
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5743427872657776,
+      "learning_rate": 0.0005212548768631118,
+      "loss": 0.1781,
+      "step": 1229
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7555286288261414,
+      "learning_rate": 0.0005206207684323337,
+      "loss": 0.3213,
+      "step": 1230
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6755029559135437,
+      "learning_rate": 0.0005199866267773868,
+      "loss": 0.3372,
+      "step": 1231
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3037530481815338,
+      "learning_rate": 0.0005193524529199994,
+      "loss": 0.1408,
+      "step": 1232
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6245545744895935,
+      "learning_rate": 0.0005187182478819523,
+      "loss": 0.2401,
+      "step": 1233
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5972506999969482,
+      "learning_rate": 0.0005180840126850763,
+      "loss": 0.303,
+      "step": 1234
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5332531929016113,
+      "learning_rate": 0.0005174497483512506,
+      "loss": 0.3186,
+      "step": 1235
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4802628457546234,
+      "learning_rate": 0.0005168154559024014,
+      "loss": 0.322,
+      "step": 1236
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8202114105224609,
+      "learning_rate": 0.0005161811363605006,
+      "loss": 0.5496,
+      "step": 1237
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5084963440895081,
+      "learning_rate": 0.0005155467907475632,
+      "loss": 0.2578,
+      "step": 1238
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.48286205530166626,
+      "learning_rate": 0.0005149124200856466,
+      "loss": 0.2741,
+      "step": 1239
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5392525792121887,
+      "learning_rate": 0.0005142780253968481,
+      "loss": 0.2331,
+      "step": 1240
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.43944358825683594,
+      "learning_rate": 0.0005136436077033044,
+      "loss": 0.2476,
+      "step": 1241
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.48144862055778503,
+      "learning_rate": 0.0005130091680271887,
+      "loss": 0.3003,
+      "step": 1242
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4356020987033844,
+      "learning_rate": 0.0005123747073907098,
+      "loss": 0.27,
+      "step": 1243
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6104761362075806,
+      "learning_rate": 0.0005117402268161102,
+      "loss": 0.2668,
+      "step": 1244
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.0429234504699707,
+      "learning_rate": 0.0005111057273256647,
+      "loss": 0.3699,
+      "step": 1245
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.3734066486358643,
+      "learning_rate": 0.0005104712099416785,
+      "loss": 0.269,
+      "step": 1246
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5260823965072632,
+      "learning_rate": 0.0005098366756864856,
+      "loss": 0.2444,
+      "step": 1247
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7035953998565674,
+      "learning_rate": 0.000509202125582447,
+      "loss": 0.3183,
+      "step": 1248
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6811387538909912,
+      "learning_rate": 0.0005085675606519497,
+      "loss": 0.3385,
+      "step": 1249
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.665605902671814,
+      "learning_rate": 0.000507932981917404,
+      "loss": 0.1629,
+      "step": 1250
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5908010005950928,
+      "learning_rate": 0.0005072983904012429,
+      "loss": 0.2936,
+      "step": 1251
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.46022671461105347,
+      "learning_rate": 0.0005066637871259201,
+      "loss": 0.2344,
+      "step": 1252
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5479859709739685,
+      "learning_rate": 0.0005060291731139076,
+      "loss": 0.1937,
+      "step": 1253
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4093743860721588,
+      "learning_rate": 0.0005053945493876952,
+      "loss": 0.1542,
+      "step": 1254
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.608150064945221,
+      "learning_rate": 0.0005047599169697884,
+      "loss": 0.2398,
+      "step": 1255
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.40389272570610046,
+      "learning_rate": 0.0005041252768827063,
+      "loss": 0.2108,
+      "step": 1256
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6458195447921753,
+      "learning_rate": 0.0005034906301489807,
+      "loss": 0.4656,
+      "step": 1257
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5573038458824158,
+      "learning_rate": 0.0005028559777911542,
+      "loss": 0.4223,
+      "step": 1258
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4716726243495941,
+      "learning_rate": 0.0005022213208317781,
+      "loss": 0.2804,
+      "step": 1259
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5809752941131592,
+      "learning_rate": 0.0005015866602934111,
+      "loss": 0.4401,
+      "step": 1260
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4886329174041748,
+      "learning_rate": 0.0005009519971986183,
+      "loss": 0.2818,
+      "step": 1261
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4117079973220825,
+      "learning_rate": 0.0005003173325699681,
+      "loss": 0.2516,
+      "step": 1262
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.43445149064064026,
+      "learning_rate": 0.000499682667430032,
+      "loss": 0.1577,
+      "step": 1263
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7662147879600525,
+      "learning_rate": 0.0004990480028013818,
+      "loss": 0.3354,
+      "step": 1264
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5708134174346924,
+      "learning_rate": 0.0004984133397065889,
+      "loss": 0.2256,
+      "step": 1265
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4830529987812042,
+      "learning_rate": 0.000497778679168222,
+      "loss": 0.3649,
+      "step": 1266
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6902855634689331,
+      "learning_rate": 0.0004971440222088459,
+      "loss": 0.2366,
+      "step": 1267
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.098472237586975,
+      "learning_rate": 0.0004965093698510193,
+      "loss": 0.202,
+      "step": 1268
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7373398542404175,
+      "learning_rate": 0.0004958747231172938,
+      "loss": 0.41,
+      "step": 1269
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.641361653804779,
+      "learning_rate": 0.0004952400830302117,
+      "loss": 0.2673,
+      "step": 1270
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6290469765663147,
+      "learning_rate": 0.0004946054506123048,
+      "loss": 0.3866,
+      "step": 1271
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4022032618522644,
+      "learning_rate": 0.0004939708268860924,
+      "loss": 0.1209,
+      "step": 1272
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5088335275650024,
+      "learning_rate": 0.0004933362128740799,
+      "loss": 0.2698,
+      "step": 1273
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7188237905502319,
+      "learning_rate": 0.000492701609598757,
+      "loss": 0.4464,
+      "step": 1274
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.489655077457428,
+      "learning_rate": 0.000492067018082596,
+      "loss": 0.1886,
+      "step": 1275
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.2303135395050049,
+      "learning_rate": 0.0004914324393480503,
+      "loss": 0.2671,
+      "step": 1276
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6081627607345581,
+      "learning_rate": 0.000490797874417553,
+      "loss": 0.2816,
+      "step": 1277
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4613741636276245,
+      "learning_rate": 0.0004901633243135144,
+      "loss": 0.2379,
+      "step": 1278
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6592218279838562,
+      "learning_rate": 0.0004895287900583215,
+      "loss": 0.2814,
+      "step": 1279
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4331430196762085,
+      "learning_rate": 0.0004888942726743353,
+      "loss": 0.1605,
+      "step": 1280
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.33474603295326233,
+      "learning_rate": 0.0004882597731838898,
+      "loss": 0.1684,
+      "step": 1281
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6644952297210693,
+      "learning_rate": 0.0004876252926092903,
+      "loss": 0.1712,
+      "step": 1282
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6033799648284912,
+      "learning_rate": 0.00048699083197281136,
+      "loss": 0.2723,
+      "step": 1283
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5661482214927673,
+      "learning_rate": 0.0004863563922966957,
+      "loss": 0.3553,
+      "step": 1284
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.1966214179992676,
+      "learning_rate": 0.0004857219746031519,
+      "loss": 0.4529,
+      "step": 1285
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5893396735191345,
+      "learning_rate": 0.00048508757991435367,
+      "loss": 0.4109,
+      "step": 1286
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.40645307302474976,
+      "learning_rate": 0.00048445320925243697,
+      "loss": 0.0951,
+      "step": 1287
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7006975412368774,
+      "learning_rate": 0.00048381886363949955,
+      "loss": 0.3212,
+      "step": 1288
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6233726739883423,
+      "learning_rate": 0.00048318454409759874,
+      "loss": 0.3029,
+      "step": 1289
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6424432992935181,
+      "learning_rate": 0.0004825502516487497,
+      "loss": 0.3589,
+      "step": 1290
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.31011345982551575,
+      "learning_rate": 0.0004819159873149239,
+      "loss": 0.1441,
+      "step": 1291
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7138365507125854,
+      "learning_rate": 0.00048128175211804787,
+      "loss": 0.2447,
+      "step": 1292
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7503759860992432,
+      "learning_rate": 0.0004806475470800008,
+      "loss": 0.4808,
+      "step": 1293
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4752475321292877,
+      "learning_rate": 0.00048001337322261346,
+      "loss": 0.2209,
+      "step": 1294
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8876127600669861,
+      "learning_rate": 0.00047937923156766646,
+      "loss": 0.4843,
+      "step": 1295
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6003799438476562,
+      "learning_rate": 0.0004787451231368882,
+      "loss": 0.2717,
+      "step": 1296
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.38364455103874207,
+      "learning_rate": 0.00047811104895195407,
+      "loss": 0.2551,
+      "step": 1297
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7016602754592896,
+      "learning_rate": 0.0004774770100344838,
+      "loss": 0.3163,
+      "step": 1298
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.43174615502357483,
+      "learning_rate": 0.00047684300740604053,
+      "loss": 0.2146,
+      "step": 1299
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.13893461227417,
+      "learning_rate": 0.0004762090420881289,
+      "loss": 0.2955,
+      "step": 1300
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5727802515029907,
+      "learning_rate": 0.00047557511510219335,
+      "loss": 0.1952,
+      "step": 1301
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6121445894241333,
+      "learning_rate": 0.00047494122746961686,
+      "loss": 0.2016,
+      "step": 1302
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6245537996292114,
+      "learning_rate": 0.00047430738021171854,
+      "loss": 0.2754,
+      "step": 1303
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5940746665000916,
+      "learning_rate": 0.0004736735743497528,
+      "loss": 0.2447,
+      "step": 1304
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5351815223693848,
+      "learning_rate": 0.0004730398109049071,
+      "loss": 0.2483,
+      "step": 1305
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4627023935317993,
+      "learning_rate": 0.00047240609089830085,
+      "loss": 0.2473,
+      "step": 1306
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5777115225791931,
+      "learning_rate": 0.00047177241535098323,
+      "loss": 0.2581,
+      "step": 1307
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6496902704238892,
+      "learning_rate": 0.00047113878528393197,
+      "loss": 0.3204,
+      "step": 1308
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7833969593048096,
+      "learning_rate": 0.00047050520171805136,
+      "loss": 0.2561,
+      "step": 1309
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5576673746109009,
+      "learning_rate": 0.00046987166567417086,
+      "loss": 0.3482,
+      "step": 1310
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.3298665285110474,
+      "learning_rate": 0.00046923817817304324,
+      "loss": 0.4108,
+      "step": 1311
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7219784259796143,
+      "learning_rate": 0.0004686047402353433,
+      "loss": 0.2342,
+      "step": 1312
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5603151917457581,
+      "learning_rate": 0.0004679713528816658,
+      "loss": 0.1948,
+      "step": 1313
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7777407765388489,
+      "learning_rate": 0.000467338017132524,
+      "loss": 0.4624,
+      "step": 1314
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7171744704246521,
+      "learning_rate": 0.00046670473400834805,
+      "loss": 0.2208,
+      "step": 1315
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5864310264587402,
+      "learning_rate": 0.0004660715045294834,
+      "loss": 0.2268,
+      "step": 1316
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8083603978157043,
+      "learning_rate": 0.0004654383297161889,
+      "loss": 0.2929,
+      "step": 1317
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5472362041473389,
+      "learning_rate": 0.00046480521058863544,
+      "loss": 0.3279,
+      "step": 1318
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4493251144886017,
+      "learning_rate": 0.0004641721481669041,
+      "loss": 0.1958,
+      "step": 1319
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3722875416278839,
+      "learning_rate": 0.00046353914347098467,
+      "loss": 0.1623,
+      "step": 1320
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6715302467346191,
+      "learning_rate": 0.00046290619752077395,
+      "loss": 0.3057,
+      "step": 1321
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6635205745697021,
+      "learning_rate": 0.00046227331133607397,
+      "loss": 0.2967,
+      "step": 1322
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8103947043418884,
+      "learning_rate": 0.0004616404859365907,
+      "loss": 0.4898,
+      "step": 1323
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.39107760787010193,
+      "learning_rate": 0.00046100772234193186,
+      "loss": 0.1688,
+      "step": 1324
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5381836295127869,
+      "learning_rate": 0.00046037502157160573,
+      "loss": 0.3015,
+      "step": 1325
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4255870282649994,
+      "learning_rate": 0.0004597423846450196,
+      "loss": 0.1967,
+      "step": 1326
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5862542390823364,
+      "learning_rate": 0.0004591098125814776,
+      "loss": 0.2559,
+      "step": 1327
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5621623396873474,
+      "learning_rate": 0.00045847730640017926,
+      "loss": 0.2037,
+      "step": 1328
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.48296311497688293,
+      "learning_rate": 0.00045784486712021824,
+      "loss": 0.1194,
+      "step": 1329
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5260604619979858,
+      "learning_rate": 0.0004572124957605803,
+      "loss": 0.1854,
+      "step": 1330
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6724199056625366,
+      "learning_rate": 0.00045658019334014174,
+      "loss": 0.3862,
+      "step": 1331
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4905756115913391,
+      "learning_rate": 0.00045594796087766786,
+      "loss": 0.2571,
+      "step": 1332
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6839861869812012,
+      "learning_rate": 0.00045531579939181124,
+      "loss": 0.1709,
+      "step": 1333
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.37646812200546265,
+      "learning_rate": 0.00045468370990111,
+      "loss": 0.1486,
+      "step": 1334
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7496247887611389,
+      "learning_rate": 0.00045405169342398633,
+      "loss": 0.3277,
+      "step": 1335
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5747839212417603,
+      "learning_rate": 0.00045341975097874484,
+      "loss": 0.2919,
+      "step": 1336
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7497596740722656,
+      "learning_rate": 0.00045278788358357064,
+      "loss": 0.2849,
+      "step": 1337
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6464429497718811,
+      "learning_rate": 0.0004521560922565282,
+      "loss": 0.3538,
+      "step": 1338
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6545172929763794,
+      "learning_rate": 0.0004515243780155593,
+      "loss": 0.2111,
+      "step": 1339
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5676270723342896,
+      "learning_rate": 0.00045089274187848144,
+      "loss": 0.2321,
+      "step": 1340
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.0442204475402832,
+      "learning_rate": 0.00045026118486298655,
+      "loss": 0.4811,
+      "step": 1341
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8982413411140442,
+      "learning_rate": 0.00044962970798663866,
+      "loss": 0.2513,
+      "step": 1342
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.38883501291275024,
+      "learning_rate": 0.00044899831226687294,
+      "loss": 0.2045,
+      "step": 1343
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5009539127349854,
+      "learning_rate": 0.00044836699872099383,
+      "loss": 0.2307,
+      "step": 1344
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.04543936252594,
+      "learning_rate": 0.00044773576836617336,
+      "loss": 0.4725,
+      "step": 1345
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4325162470340729,
+      "learning_rate": 0.00044710462221944934,
+      "loss": 0.3319,
+      "step": 1346
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5937920212745667,
+      "learning_rate": 0.0004464735612977242,
+      "loss": 0.3165,
+      "step": 1347
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.411224901676178,
+      "learning_rate": 0.0004458425866177627,
+      "loss": 0.1737,
+      "step": 1348
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.1483572721481323,
+      "learning_rate": 0.000445211699196191,
+      "loss": 0.3306,
+      "step": 1349
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.507117509841919,
+      "learning_rate": 0.00044458090004949454,
+      "loss": 0.1584,
+      "step": 1350
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.166082739830017,
+      "learning_rate": 0.00044395019019401633,
+      "loss": 0.3882,
+      "step": 1351
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8233304619789124,
+      "learning_rate": 0.0004433195706459558,
+      "loss": 0.3211,
+      "step": 1352
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5203681588172913,
+      "learning_rate": 0.00044268904242136673,
+      "loss": 0.2806,
+      "step": 1353
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.852720856666565,
+      "learning_rate": 0.0004420586065361558,
+      "loss": 0.3201,
+      "step": 1354
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.49043864011764526,
+      "learning_rate": 0.0004414282640060809,
+      "loss": 0.1337,
+      "step": 1355
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5777467489242554,
+      "learning_rate": 0.0004407980158467495,
+      "loss": 0.2153,
+      "step": 1356
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.564673125743866,
+      "learning_rate": 0.0004401678630736172,
+      "loss": 0.2008,
+      "step": 1357
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7502429485321045,
+      "learning_rate": 0.0004395378067019854,
+      "loss": 0.413,
+      "step": 1358
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7711302042007446,
+      "learning_rate": 0.0004389078477470008,
+      "loss": 0.536,
+      "step": 1359
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5109966993331909,
+      "learning_rate": 0.00043827798722365264,
+      "loss": 0.3362,
+      "step": 1360
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7068579792976379,
+      "learning_rate": 0.000437648226146772,
+      "loss": 0.2148,
+      "step": 1361
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4624647796154022,
+      "learning_rate": 0.00043701856553102946,
+      "loss": 0.2565,
+      "step": 1362
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5238811373710632,
+      "learning_rate": 0.0004363890063909338,
+      "loss": 0.3739,
+      "step": 1363
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7186887860298157,
+      "learning_rate": 0.0004357595497408303,
+      "loss": 0.2197,
+      "step": 1364
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4615837633609772,
+      "learning_rate": 0.0004351301965948991,
+      "loss": 0.238,
+      "step": 1365
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.42300844192504883,
+      "learning_rate": 0.00043450094796715355,
+      "loss": 0.1776,
+      "step": 1366
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5931771397590637,
+      "learning_rate": 0.0004338718048714387,
+      "loss": 0.1834,
+      "step": 1367
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.3082271814346313,
+      "learning_rate": 0.0004332427683214295,
+      "loss": 0.5538,
+      "step": 1368
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5966417193412781,
+      "learning_rate": 0.0004326138393306292,
+      "loss": 0.3886,
+      "step": 1369
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5978918075561523,
+      "learning_rate": 0.000431985018912368,
+      "loss": 0.2342,
+      "step": 1370
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2906109690666199,
+      "learning_rate": 0.0004313563080798006,
+      "loss": 0.1231,
+      "step": 1371
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6116635203361511,
+      "learning_rate": 0.00043072770784590564,
+      "loss": 0.2673,
+      "step": 1372
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6135227084159851,
+      "learning_rate": 0.00043009921922348336,
+      "loss": 0.2486,
+      "step": 1373
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.9338779449462891,
+      "learning_rate": 0.00042947084322515435,
+      "loss": 0.2926,
+      "step": 1374
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8346027731895447,
+      "learning_rate": 0.0004288425808633575,
+      "loss": 0.5292,
+      "step": 1375
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.45036184787750244,
+      "learning_rate": 0.0004282144331503488,
+      "loss": 0.1852,
+      "step": 1376
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5766432881355286,
+      "learning_rate": 0.00042758640109819945,
+      "loss": 0.2296,
+      "step": 1377
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7593908309936523,
+      "learning_rate": 0.00042695848571879425,
+      "loss": 0.4153,
+      "step": 1378
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.0980818271636963,
+      "learning_rate": 0.00042633068802383007,
+      "loss": 0.3768,
+      "step": 1379
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6042316555976868,
+      "learning_rate": 0.00042570300902481425,
+      "loss": 0.2257,
+      "step": 1380
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5592933893203735,
+      "learning_rate": 0.00042507544973306255,
+      "loss": 0.3085,
+      "step": 1381
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.801805853843689,
+      "learning_rate": 0.00042444801115969834,
+      "loss": 0.1576,
+      "step": 1382
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7482402920722961,
+      "learning_rate": 0.00042382069431565005,
+      "loss": 0.2982,
+      "step": 1383
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.183913230895996,
+      "learning_rate": 0.00042319350021165035,
+      "loss": 0.3519,
+      "step": 1384
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5274115204811096,
+      "learning_rate": 0.0004225664298582339,
+      "loss": 0.3167,
+      "step": 1385
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.9177464246749878,
+      "learning_rate": 0.00042193948426573605,
+      "loss": 0.5167,
+      "step": 1386
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6432636976242065,
+      "learning_rate": 0.00042131266444429106,
+      "loss": 0.2184,
+      "step": 1387
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5422487854957581,
+      "learning_rate": 0.0004206859714038308,
+      "loss": 0.2485,
+      "step": 1388
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8615155816078186,
+      "learning_rate": 0.0004200594061540826,
+      "loss": 0.3277,
+      "step": 1389
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6886224746704102,
+      "learning_rate": 0.000419432969704568,
+      "loss": 0.2691,
+      "step": 1390
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4622880220413208,
+      "learning_rate": 0.000418806663064601,
+      "loss": 0.2283,
+      "step": 1391
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3486861288547516,
+      "learning_rate": 0.0004181804872432864,
+      "loss": 0.1676,
+      "step": 1392
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5075287222862244,
+      "learning_rate": 0.0004175544432495184,
+      "loss": 0.302,
+      "step": 1393
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5341110229492188,
+      "learning_rate": 0.0004169285320919786,
+      "loss": 0.1924,
+      "step": 1394
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.31910768151283264,
+      "learning_rate": 0.0004163027547791347,
+      "loss": 0.1497,
+      "step": 1395
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5273780226707458,
+      "learning_rate": 0.00041567711231923875,
+      "loss": 0.1872,
+      "step": 1396
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4046824872493744,
+      "learning_rate": 0.00041505160572032535,
+      "loss": 0.1936,
+      "step": 1397
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.40604498982429504,
+      "learning_rate": 0.0004144262359902104,
+      "loss": 0.1878,
+      "step": 1398
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7499468922615051,
+      "learning_rate": 0.00041380100413648916,
+      "loss": 0.3451,
+      "step": 1399
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.9113187789916992,
+      "learning_rate": 0.00041317591116653486,
+      "loss": 0.3477,
+      "step": 1400
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4917261302471161,
+      "learning_rate": 0.0004125509580874969,
+      "loss": 0.2748,
+      "step": 1401
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4847176671028137,
+      "learning_rate": 0.00041192614590629915,
+      "loss": 0.2679,
+      "step": 1402
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.35232701897621155,
+      "learning_rate": 0.00041130147562963885,
+      "loss": 0.2412,
+      "step": 1403
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5438411831855774,
+      "learning_rate": 0.00041067694826398406,
+      "loss": 0.4306,
+      "step": 1404
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.39578789472579956,
+      "learning_rate": 0.00041005256481557305,
+      "loss": 0.146,
+      "step": 1405
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5750135779380798,
+      "learning_rate": 0.000409428326290412,
+      "loss": 0.2974,
+      "step": 1406
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5082857012748718,
+      "learning_rate": 0.00040880423369427354,
+      "loss": 0.1875,
+      "step": 1407
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7963637709617615,
+      "learning_rate": 0.00040818028803269547,
+      "loss": 0.3594,
+      "step": 1408
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6540344953536987,
+      "learning_rate": 0.0004075564903109784,
+      "loss": 0.2519,
+      "step": 1409
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.46947067975997925,
+      "learning_rate": 0.000406932841534185,
+      "loss": 0.2008,
+      "step": 1410
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.507051944732666,
+      "learning_rate": 0.0004063093427071376,
+      "loss": 0.2424,
+      "step": 1411
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5720059275627136,
+      "learning_rate": 0.00040568599483441746,
+      "loss": 0.2553,
+      "step": 1412
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3568060100078583,
+      "learning_rate": 0.00040506279892036187,
+      "loss": 0.1673,
+      "step": 1413
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.515558123588562,
+      "learning_rate": 0.0004044397559690638,
+      "loss": 0.2361,
+      "step": 1414
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4425044357776642,
+      "learning_rate": 0.0004038168669843697,
+      "loss": 0.175,
+      "step": 1415
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.46418333053588867,
+      "learning_rate": 0.00040319413296987784,
+      "loss": 0.248,
+      "step": 1416
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.27793437242507935,
+      "learning_rate": 0.00040257155492893706,
+      "loss": 0.1278,
+      "step": 1417
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3570486009120941,
+      "learning_rate": 0.00040194913386464444,
+      "loss": 0.1123,
+      "step": 1418
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4627930819988251,
+      "learning_rate": 0.0004013268707798447,
+      "loss": 0.2518,
+      "step": 1419
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.47083809971809387,
+      "learning_rate": 0.00040070476667712743,
+      "loss": 0.1379,
+      "step": 1420
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5171617269515991,
+      "learning_rate": 0.00040008282255882645,
+      "loss": 0.189,
+      "step": 1421
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5554575324058533,
+      "learning_rate": 0.0003994610394270178,
+      "loss": 0.2609,
+      "step": 1422
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5172342658042908,
+      "learning_rate": 0.000398839418283518,
+      "loss": 0.284,
+      "step": 1423
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6821736693382263,
+      "learning_rate": 0.00039821796012988263,
+      "loss": 0.2747,
+      "step": 1424
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6868269443511963,
+      "learning_rate": 0.0003975966659674047,
+      "loss": 0.1921,
+      "step": 1425
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6725178360939026,
+      "learning_rate": 0.0003969755367971131,
+      "loss": 0.3997,
+      "step": 1426
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4366752803325653,
+      "learning_rate": 0.00039635457361977046,
+      "loss": 0.1723,
+      "step": 1427
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2950134873390198,
+      "learning_rate": 0.00039573377743587246,
+      "loss": 0.1687,
+      "step": 1428
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4947931468486786,
+      "learning_rate": 0.00039511314924564546,
+      "loss": 0.1597,
+      "step": 1429
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5434812903404236,
+      "learning_rate": 0.0003944926900490452,
+      "loss": 0.3293,
+      "step": 1430
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7365113496780396,
+      "learning_rate": 0.00039387240084575516,
+      "loss": 0.2958,
+      "step": 1431
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.45515209436416626,
+      "learning_rate": 0.00039325228263518486,
+      "loss": 0.2315,
+      "step": 1432
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.33248576521873474,
+      "learning_rate": 0.0003926323364164684,
+      "loss": 0.1175,
+      "step": 1433
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.9997519850730896,
+      "learning_rate": 0.00039201256318846273,
+      "loss": 0.2283,
+      "step": 1434
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7527481317520142,
+      "learning_rate": 0.0003913929639497462,
+      "loss": 0.5307,
+      "step": 1435
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6624760627746582,
+      "learning_rate": 0.0003907735396986165,
+      "loss": 0.3762,
+      "step": 1436
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4882284998893738,
+      "learning_rate": 0.0003901542914330896,
+      "loss": 0.2123,
+      "step": 1437
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6611272692680359,
+      "learning_rate": 0.00038953522015089804,
+      "loss": 0.261,
+      "step": 1438
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6852008104324341,
+      "learning_rate": 0.000388916326849489,
+      "loss": 0.2228,
+      "step": 1439
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5258339047431946,
+      "learning_rate": 0.0003882976125260229,
+      "loss": 0.1829,
+      "step": 1440
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7073531150817871,
+      "learning_rate": 0.000387679078177372,
+      "loss": 0.2106,
+      "step": 1441
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7238799333572388,
+      "learning_rate": 0.0003870607248001184,
+      "loss": 0.2313,
+      "step": 1442
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3475869596004486,
+      "learning_rate": 0.0003864425533905527,
+      "loss": 0.2229,
+      "step": 1443
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5526435971260071,
+      "learning_rate": 0.0003858245649446721,
+      "loss": 0.2492,
+      "step": 1444
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8399673104286194,
+      "learning_rate": 0.0003852067604581794,
+      "loss": 0.3551,
+      "step": 1445
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6057975888252258,
+      "learning_rate": 0.00038458914092648073,
+      "loss": 0.301,
+      "step": 1446
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7111477851867676,
+      "learning_rate": 0.0003839717073446842,
+      "loss": 0.302,
+      "step": 1447
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.47930166125297546,
+      "learning_rate": 0.00038335446070759857,
+      "loss": 0.2207,
+      "step": 1448
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3944559693336487,
+      "learning_rate": 0.0003827374020097311,
+      "loss": 0.2047,
+      "step": 1449
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8423956036567688,
+      "learning_rate": 0.0003821205322452863,
+      "loss": 0.2447,
+      "step": 1450
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5238419771194458,
+      "learning_rate": 0.0003815038524081646,
+      "loss": 0.2289,
+      "step": 1451
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5598323941230774,
+      "learning_rate": 0.0003808873634919599,
+      "loss": 0.2733,
+      "step": 1452
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5247355699539185,
+      "learning_rate": 0.0003802710664899588,
+      "loss": 0.2699,
+      "step": 1453
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.48935574293136597,
+      "learning_rate": 0.00037965496239513873,
+      "loss": 0.3503,
+      "step": 1454
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7752866744995117,
+      "learning_rate": 0.0003790390522001662,
+      "loss": 0.3131,
+      "step": 1455
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.335162878036499,
+      "learning_rate": 0.0003784233368973952,
+      "loss": 0.1236,
+      "step": 1456
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.485786497592926,
+      "learning_rate": 0.00037780781747886593,
+      "loss": 0.2247,
+      "step": 1457
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5163724422454834,
+      "learning_rate": 0.000377192494936303,
+      "loss": 0.2504,
+      "step": 1458
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6366633772850037,
+      "learning_rate": 0.0003765773702611134,
+      "loss": 0.1899,
+      "step": 1459
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.4217761754989624,
+      "learning_rate": 0.0003759624444443858,
+      "loss": 0.6331,
+      "step": 1460
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.48011329770088196,
+      "learning_rate": 0.00037534771847688817,
+      "loss": 0.193,
+      "step": 1461
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7198631763458252,
+      "learning_rate": 0.00037473319334906677,
+      "loss": 0.2232,
+      "step": 1462
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5556163191795349,
+      "learning_rate": 0.00037411887005104396,
+      "loss": 0.2937,
+      "step": 1463
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6626385450363159,
+      "learning_rate": 0.00037350474957261706,
+      "loss": 0.2776,
+      "step": 1464
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5172512531280518,
+      "learning_rate": 0.00037289083290325663,
+      "loss": 0.1982,
+      "step": 1465
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8622768521308899,
+      "learning_rate": 0.0003722771210321048,
+      "loss": 0.3825,
+      "step": 1466
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7267595529556274,
+      "learning_rate": 0.0003716636149479737,
+      "loss": 0.1659,
+      "step": 1467
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7739598751068115,
+      "learning_rate": 0.0003710503156393441,
+      "loss": 0.2446,
+      "step": 1468
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5506634712219238,
+      "learning_rate": 0.00037043722409436334,
+      "loss": 0.225,
+      "step": 1469
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6424477696418762,
+      "learning_rate": 0.00036982434130084397,
+      "loss": 0.2803,
+      "step": 1470
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8989078998565674,
+      "learning_rate": 0.0003692116682462626,
+      "loss": 0.172,
+      "step": 1471
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5391958355903625,
+      "learning_rate": 0.00036859920591775756,
+      "loss": 0.2264,
+      "step": 1472
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.1317509412765503,
+      "learning_rate": 0.00036798695530212776,
+      "loss": 0.3652,
+      "step": 1473
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7309859991073608,
+      "learning_rate": 0.00036737491738583116,
+      "loss": 0.3042,
+      "step": 1474
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4718625247478485,
+      "learning_rate": 0.0003667630931549826,
+      "loss": 0.2351,
+      "step": 1475
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4488973021507263,
+      "learning_rate": 0.000366151483595353,
+      "loss": 0.2051,
+      "step": 1476
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.46878764033317566,
+      "learning_rate": 0.00036554008969236717,
+      "loss": 0.2368,
+      "step": 1477
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5311998128890991,
+      "learning_rate": 0.00036492891243110283,
+      "loss": 0.2422,
+      "step": 1478
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5397355556488037,
+      "learning_rate": 0.0003643179527962882,
+      "loss": 0.2679,
+      "step": 1479
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4746856689453125,
+      "learning_rate": 0.00036370721177230114,
+      "loss": 0.2444,
+      "step": 1480
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.36967629194259644,
+      "learning_rate": 0.0003630966903431671,
+      "loss": 0.2298,
+      "step": 1481
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7091696262359619,
+      "learning_rate": 0.0003624863894925579,
+      "loss": 0.3186,
+      "step": 1482
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.45452743768692017,
+      "learning_rate": 0.00036187631020378986,
+      "loss": 0.2172,
+      "step": 1483
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5324978828430176,
+      "learning_rate": 0.0003612664534598224,
+      "loss": 0.1893,
+      "step": 1484
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5635752081871033,
+      "learning_rate": 0.0003606568202432562,
+      "loss": 0.3216,
+      "step": 1485
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.473518043756485,
+      "learning_rate": 0.0003600474115363319,
+      "loss": 0.2215,
+      "step": 1486
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.42318835854530334,
+      "learning_rate": 0.0003594382283209286,
+      "loss": 0.2263,
+      "step": 1487
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.9782075881958008,
+      "learning_rate": 0.0003588292715785617,
+      "loss": 0.4143,
+      "step": 1488
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.38577574491500854,
+      "learning_rate": 0.00035822054229038206,
+      "loss": 0.1478,
+      "step": 1489
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.487304151058197,
+      "learning_rate": 0.00035761204143717383,
+      "loss": 0.2271,
+      "step": 1490
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3951490521430969,
+      "learning_rate": 0.00035700376999935334,
+      "loss": 0.1593,
+      "step": 1491
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3283596634864807,
+      "learning_rate": 0.00035639572895696686,
+      "loss": 0.1573,
+      "step": 1492
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.64703369140625,
+      "learning_rate": 0.00035578791928968993,
+      "loss": 0.2678,
+      "step": 1493
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6910591125488281,
+      "learning_rate": 0.0003551803419768251,
+      "loss": 0.4373,
+      "step": 1494
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6167072653770447,
+      "learning_rate": 0.00035457299799730046,
+      "loss": 0.176,
+      "step": 1495
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6198394894599915,
+      "learning_rate": 0.00035396588832966826,
+      "loss": 0.1899,
+      "step": 1496
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 10.932646751403809,
+      "learning_rate": 0.0003533590139521033,
+      "loss": 0.3948,
+      "step": 1497
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4788089692592621,
+      "learning_rate": 0.00035275237584240127,
+      "loss": 0.1975,
+      "step": 1498
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6831006407737732,
+      "learning_rate": 0.0003521459749779768,
+      "loss": 0.3469,
+      "step": 1499
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.42612120509147644,
+      "learning_rate": 0.00035153981233586274,
+      "loss": 0.1611,
+      "step": 1500
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.529823899269104,
+      "learning_rate": 0.00035093388889270786,
+      "loss": 0.2013,
+      "step": 1501
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.39637184143066406,
+      "learning_rate": 0.00035032820562477574,
+      "loss": 0.152,
+      "step": 1502
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.48494189977645874,
+      "learning_rate": 0.00034972276350794286,
+      "loss": 0.1742,
+      "step": 1503
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5656492710113525,
+      "learning_rate": 0.0003491175635176972,
+      "loss": 0.2319,
+      "step": 1504
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8223924040794373,
+      "learning_rate": 0.0003485126066291364,
+      "loss": 0.194,
+      "step": 1505
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4379829466342926,
+      "learning_rate": 0.00034790789381696684,
+      "loss": 0.1482,
+      "step": 1506
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6520694494247437,
+      "learning_rate": 0.00034730342605550135,
+      "loss": 0.3363,
+      "step": 1507
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8732770085334778,
+      "learning_rate": 0.000346699204318658,
+      "loss": 0.3599,
+      "step": 1508
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3645210862159729,
+      "learning_rate": 0.00034609522957995844,
+      "loss": 0.1702,
+      "step": 1509
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4755183160305023,
+      "learning_rate": 0.00034549150281252633,
+      "loss": 0.2243,
+      "step": 1510
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.580884575843811,
+      "learning_rate": 0.00034488802498908595,
+      "loss": 0.1953,
+      "step": 1511
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5562728047370911,
+      "learning_rate": 0.00034428479708196037,
+      "loss": 0.1415,
+      "step": 1512
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3049013912677765,
+      "learning_rate": 0.00034368182006307005,
+      "loss": 0.122,
+      "step": 1513
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8780558109283447,
+      "learning_rate": 0.00034307909490393095,
+      "loss": 0.2993,
+      "step": 1514
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6014513969421387,
+      "learning_rate": 0.0003424766225756537,
+      "loss": 0.3613,
+      "step": 1515
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4829213321208954,
+      "learning_rate": 0.0003418744040489412,
+      "loss": 0.2743,
+      "step": 1516
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.41878432035446167,
+      "learning_rate": 0.0003412724402940876,
+      "loss": 0.125,
+      "step": 1517
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6898136734962463,
+      "learning_rate": 0.0003406707322809766,
+      "loss": 0.2462,
+      "step": 1518
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3427504599094391,
+      "learning_rate": 0.0003400692809790796,
+      "loss": 0.0933,
+      "step": 1519
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7385209202766418,
+      "learning_rate": 0.0003394680873574546,
+      "loss": 0.1282,
+      "step": 1520
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.518543004989624,
+      "learning_rate": 0.00033886715238474455,
+      "loss": 0.1785,
+      "step": 1521
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5221672058105469,
+      "learning_rate": 0.0003382664770291752,
+      "loss": 0.239,
+      "step": 1522
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6485975384712219,
+      "learning_rate": 0.00033766606225855457,
+      "loss": 0.2451,
+      "step": 1523
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.533948540687561,
+      "learning_rate": 0.00033706590904027036,
+      "loss": 0.224,
+      "step": 1524
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.633888304233551,
+      "learning_rate": 0.0003364660183412892,
+      "loss": 0.3252,
+      "step": 1525
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7050673961639404,
+      "learning_rate": 0.0003358663911281544,
+      "loss": 0.6138,
+      "step": 1526
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.9656645059585571,
+      "learning_rate": 0.0003352670283669852,
+      "loss": 0.4752,
+      "step": 1527
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.994947612285614,
+      "learning_rate": 0.0003346679310234744,
+      "loss": 0.4276,
+      "step": 1528
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.45203593373298645,
+      "learning_rate": 0.00033406910006288716,
+      "loss": 0.1679,
+      "step": 1529
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7085717916488647,
+      "learning_rate": 0.00033347053645005966,
+      "loss": 0.2134,
+      "step": 1530
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5689020752906799,
+      "learning_rate": 0.00033287224114939706,
+      "loss": 0.187,
+      "step": 1531
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5220786333084106,
+      "learning_rate": 0.00033227421512487253,
+      "loss": 0.2715,
+      "step": 1532
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6286534070968628,
+      "learning_rate": 0.0003316764593400251,
+      "loss": 0.3098,
+      "step": 1533
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.678643524646759,
+      "learning_rate": 0.00033107897475795854,
+      "loss": 0.4205,
+      "step": 1534
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5814206600189209,
+      "learning_rate": 0.00033048176234133967,
+      "loss": 0.3251,
+      "step": 1535
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5203486084938049,
+      "learning_rate": 0.0003298848230523967,
+      "loss": 0.3803,
+      "step": 1536
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5527113676071167,
+      "learning_rate": 0.00032928815785291787,
+      "loss": 0.2944,
+      "step": 1537
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6000751256942749,
+      "learning_rate": 0.00032869176770424977,
+      "loss": 0.3112,
+      "step": 1538
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.9565652012825012,
+      "learning_rate": 0.00032809565356729575,
+      "loss": 0.3485,
+      "step": 1539
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6817490458488464,
+      "learning_rate": 0.0003274998164025148,
+      "loss": 0.3043,
+      "step": 1540
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5999687314033508,
+      "learning_rate": 0.000326904257169919,
+      "loss": 0.1968,
+      "step": 1541
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8395372629165649,
+      "learning_rate": 0.0003263089768290731,
+      "loss": 0.2711,
+      "step": 1542
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4859904944896698,
+      "learning_rate": 0.0003257139763390925,
+      "loss": 0.4411,
+      "step": 1543
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.860249400138855,
+      "learning_rate": 0.0003251192566586416,
+      "loss": 0.1938,
+      "step": 1544
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5148289203643799,
+      "learning_rate": 0.0003245248187459323,
+      "loss": 0.2146,
+      "step": 1545
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.38087597489356995,
+      "learning_rate": 0.0003239306635587226,
+      "loss": 0.1635,
+      "step": 1546
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3393816649913788,
+      "learning_rate": 0.0003233367920543151,
+      "loss": 0.1888,
+      "step": 1547
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8856964707374573,
+      "learning_rate": 0.00032274320518955497,
+      "loss": 0.3934,
+      "step": 1548
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.1225167512893677,
+      "learning_rate": 0.0003221499039208291,
+      "loss": 0.4337,
+      "step": 1549
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.662632405757904,
+      "learning_rate": 0.00032155688920406414,
+      "loss": 0.3098,
+      "step": 1550
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5616256594657898,
+      "learning_rate": 0.00032096416199472495,
+      "loss": 0.3081,
+      "step": 1551
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7170083522796631,
+      "learning_rate": 0.00032037172324781327,
+      "loss": 0.3596,
+      "step": 1552
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4767475724220276,
+      "learning_rate": 0.00031977957391786616,
+      "loss": 0.1512,
+      "step": 1553
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6480745673179626,
+      "learning_rate": 0.00031918771495895393,
+      "loss": 0.2658,
+      "step": 1554
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8991349339485168,
+      "learning_rate": 0.00031859614732467957,
+      "loss": 0.2753,
+      "step": 1555
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5432215332984924,
+      "learning_rate": 0.00031800487196817647,
+      "loss": 0.2153,
+      "step": 1556
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6638306379318237,
+      "learning_rate": 0.00031741388984210707,
+      "loss": 0.2948,
+      "step": 1557
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3066766560077667,
+      "learning_rate": 0.0003168232018986613,
+      "loss": 0.1251,
+      "step": 1558
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5557448267936707,
+      "learning_rate": 0.0003162328090895554,
+      "loss": 0.2891,
+      "step": 1559
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5230395197868347,
+      "learning_rate": 0.0003156427123660297,
+      "loss": 0.1447,
+      "step": 1560
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5600555539131165,
+      "learning_rate": 0.0003150529126788477,
+      "loss": 0.2682,
+      "step": 1561
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3566872477531433,
+      "learning_rate": 0.00031446341097829443,
+      "loss": 0.1101,
+      "step": 1562
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4270867109298706,
+      "learning_rate": 0.00031387420821417443,
+      "loss": 0.2118,
+      "step": 1563
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6850987076759338,
+      "learning_rate": 0.000313285305335811,
+      "loss": 0.1825,
+      "step": 1564
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.438902497291565,
+      "learning_rate": 0.00031269670329204396,
+      "loss": 0.637,
+      "step": 1565
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8428890109062195,
+      "learning_rate": 0.00031210840303122863,
+      "loss": 0.2703,
+      "step": 1566
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6533735394477844,
+      "learning_rate": 0.00031152040550123396,
+      "loss": 0.3271,
+      "step": 1567
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.49693766236305237,
+      "learning_rate": 0.00031093271164944113,
+      "loss": 0.1709,
+      "step": 1568
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.603986382484436,
+      "learning_rate": 0.0003103453224227424,
+      "loss": 0.141,
+      "step": 1569
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4767712950706482,
+      "learning_rate": 0.0003097582387675385,
+      "loss": 0.1573,
+      "step": 1570
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.41160711646080017,
+      "learning_rate": 0.00030917146162973846,
+      "loss": 0.1692,
+      "step": 1571
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.39470282196998596,
+      "learning_rate": 0.0003085849919547572,
+      "loss": 0.1922,
+      "step": 1572
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4779236614704132,
+      "learning_rate": 0.00030799883068751435,
+      "loss": 0.2349,
+      "step": 1573
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7107570171356201,
+      "learning_rate": 0.00030741297877243237,
+      "loss": 0.2524,
+      "step": 1574
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5750626921653748,
+      "learning_rate": 0.00030682743715343565,
+      "loss": 0.357,
+      "step": 1575
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.42093855142593384,
+      "learning_rate": 0.0003062422067739485,
+      "loss": 0.113,
+      "step": 1576
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.39594075083732605,
+      "learning_rate": 0.0003056572885768937,
+      "loss": 0.1357,
+      "step": 1577
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3834647834300995,
+      "learning_rate": 0.000305072683504691,
+      "loss": 0.1212,
+      "step": 1578
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4627830684185028,
+      "learning_rate": 0.0003044883924992557,
+      "loss": 0.1738,
+      "step": 1579
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.9418272972106934,
+      "learning_rate": 0.00030390441650199725,
+      "loss": 0.2303,
+      "step": 1580
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7691028714179993,
+      "learning_rate": 0.00030332075645381727,
+      "loss": 0.3347,
+      "step": 1581
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4376651346683502,
+      "learning_rate": 0.0003027374132951085,
+      "loss": 0.1426,
+      "step": 1582
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.9362401962280273,
+      "learning_rate": 0.0003021543879657533,
+      "loss": 0.1852,
+      "step": 1583
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7820633053779602,
+      "learning_rate": 0.0003015716814051213,
+      "loss": 0.2391,
+      "step": 1584
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.30439847707748413,
+      "learning_rate": 0.00030098929455206903,
+      "loss": 0.1503,
+      "step": 1585
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.44292593002319336,
+      "learning_rate": 0.0003004072283449379,
+      "loss": 0.187,
+      "step": 1586
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6317681670188904,
+      "learning_rate": 0.0002998254837215526,
+      "loss": 0.3051,
+      "step": 1587
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4450269341468811,
+      "learning_rate": 0.0002992440616192197,
+      "loss": 0.1526,
+      "step": 1588
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4756679832935333,
+      "learning_rate": 0.00029866296297472616,
+      "loss": 0.1575,
+      "step": 1589
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6380780935287476,
+      "learning_rate": 0.00029808218872433767,
+      "loss": 0.1951,
+      "step": 1590
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3304156959056854,
+      "learning_rate": 0.00029750173980379736,
+      "loss": 0.0512,
+      "step": 1591
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4558219313621521,
+      "learning_rate": 0.0002969216171483242,
+      "loss": 0.2728,
+      "step": 1592
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4633806347846985,
+      "learning_rate": 0.00029634182169261135,
+      "loss": 0.2216,
+      "step": 1593
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4375460147857666,
+      "learning_rate": 0.000295762354370825,
+      "loss": 0.208,
+      "step": 1594
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5668496489524841,
+      "learning_rate": 0.0002951832161166024,
+      "loss": 0.3033,
+      "step": 1595
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6970236897468567,
+      "learning_rate": 0.00029460440786305075,
+      "loss": 0.321,
+      "step": 1596
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4035065472126007,
+      "learning_rate": 0.00029402593054274557,
+      "loss": 0.1852,
+      "step": 1597
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.49647217988967896,
+      "learning_rate": 0.00029344778508772916,
+      "loss": 0.2425,
+      "step": 1598
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5142238736152649,
+      "learning_rate": 0.0002928699724295091,
+      "loss": 0.3149,
+      "step": 1599
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.37981370091438293,
+      "learning_rate": 0.0002922924934990568,
+      "loss": 0.1787,
+      "step": 1600
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5125280618667603,
+      "learning_rate": 0.00029171534922680596,
+      "loss": 0.242,
+      "step": 1601
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6091323494911194,
+      "learning_rate": 0.0002911385405426511,
+      "loss": 0.2082,
+      "step": 1602
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3661910593509674,
+      "learning_rate": 0.00029056206837594565,
+      "loss": 0.1355,
+      "step": 1603
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.488540381193161,
+      "learning_rate": 0.00028998593365550174,
+      "loss": 0.1531,
+      "step": 1604
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5429328680038452,
+      "learning_rate": 0.0002894101373095867,
+      "loss": 0.2578,
+      "step": 1605
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5442932844161987,
+      "learning_rate": 0.0002888346802659238,
+      "loss": 0.2868,
+      "step": 1606
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4877355098724365,
+      "learning_rate": 0.0002882595634516886,
+      "loss": 0.1355,
+      "step": 1607
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.539988100528717,
+      "learning_rate": 0.00028768478779350927,
+      "loss": 0.2603,
+      "step": 1608
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.41667425632476807,
+      "learning_rate": 0.00028711035421746366,
+      "loss": 0.1347,
+      "step": 1609
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4320918917655945,
+      "learning_rate": 0.0002865362636490791,
+      "loss": 0.1489,
+      "step": 1610
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5243663787841797,
+      "learning_rate": 0.00028596251701332973,
+      "loss": 0.1469,
+      "step": 1611
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6771276593208313,
+      "learning_rate": 0.00028538911523463595,
+      "loss": 0.3281,
+      "step": 1612
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.413829505443573,
+      "learning_rate": 0.00028481605923686207,
+      "loss": 0.144,
+      "step": 1613
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6589773297309875,
+      "learning_rate": 0.0002842433499433158,
+      "loss": 0.2448,
+      "step": 1614
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5140525698661804,
+      "learning_rate": 0.00028367098827674573,
+      "loss": 0.1954,
+      "step": 1615
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4005924165248871,
+      "learning_rate": 0.00028309897515934105,
+      "loss": 0.1429,
+      "step": 1616
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.0166994333267212,
+      "learning_rate": 0.0002825273115127286,
+      "loss": 0.2242,
+      "step": 1617
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6818990111351013,
+      "learning_rate": 0.0002819559982579723,
+      "loss": 0.2054,
+      "step": 1618
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7678354978561401,
+      "learning_rate": 0.0002813850363155722,
+      "loss": 0.231,
+      "step": 1619
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.497450053691864,
+      "learning_rate": 0.00028081442660546124,
+      "loss": 0.1985,
+      "step": 1620
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.3372050523757935,
+      "learning_rate": 0.00028024417004700596,
+      "loss": 0.3307,
+      "step": 1621
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7201959490776062,
+      "learning_rate": 0.00027967426755900294,
+      "loss": 0.3064,
+      "step": 1622
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.42270874977111816,
+      "learning_rate": 0.0002791047200596791,
+      "loss": 0.1597,
+      "step": 1623
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.29078662395477295,
+      "learning_rate": 0.0002785355284666886,
+      "loss": 0.08,
+      "step": 1624
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7525697946548462,
+      "learning_rate": 0.0002779666936971129,
+      "loss": 0.3776,
+      "step": 1625
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7718998193740845,
+      "learning_rate": 0.00027739821666745816,
+      "loss": 0.2306,
+      "step": 1626
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8091120719909668,
+      "learning_rate": 0.00027683009829365413,
+      "loss": 0.3035,
+      "step": 1627
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.0184810161590576,
+      "learning_rate": 0.0002762623394910525,
+      "loss": 0.2608,
+      "step": 1628
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3765392005443573,
+      "learning_rate": 0.0002756949411744264,
+      "loss": 0.1554,
+      "step": 1629
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8755757808685303,
+      "learning_rate": 0.0002751279042579672,
+      "loss": 0.2188,
+      "step": 1630
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5236902236938477,
+      "learning_rate": 0.00027456122965528474,
+      "loss": 0.0866,
+      "step": 1631
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.43244364857673645,
+      "learning_rate": 0.0002739949182794045,
+      "loss": 0.2027,
+      "step": 1632
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.1580415964126587,
+      "learning_rate": 0.00027342897104276734,
+      "loss": 0.3209,
+      "step": 1633
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4686545431613922,
+      "learning_rate": 0.0002728633888572267,
+      "loss": 0.2776,
+      "step": 1634
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.34226664900779724,
+      "learning_rate": 0.00027229817263404863,
+      "loss": 0.1192,
+      "step": 1635
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.905350923538208,
+      "learning_rate": 0.0002717333232839088,
+      "loss": 0.4047,
+      "step": 1636
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.43198248744010925,
+      "learning_rate": 0.0002711688417168924,
+      "loss": 0.1338,
+      "step": 1637
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.58113694190979,
+      "learning_rate": 0.00027060472884249143,
+      "loss": 0.2179,
+      "step": 1638
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5295534133911133,
+      "learning_rate": 0.00027004098556960453,
+      "loss": 0.1874,
+      "step": 1639
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.57394939661026,
+      "learning_rate": 0.0002694776128065345,
+      "loss": 0.2616,
+      "step": 1640
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5774073600769043,
+      "learning_rate": 0.0002689146114609868,
+      "loss": 0.234,
+      "step": 1641
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.283901572227478,
+      "learning_rate": 0.00026835198244006924,
+      "loss": 0.4036,
+      "step": 1642
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6678780913352966,
+      "learning_rate": 0.00026778972665028903,
+      "loss": 0.3199,
+      "step": 1643
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5236244797706604,
+      "learning_rate": 0.0002672278449975527,
+      "loss": 0.2841,
+      "step": 1644
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5693409442901611,
+      "learning_rate": 0.00026666633838716316,
+      "loss": 0.2082,
+      "step": 1645
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7569113969802856,
+      "learning_rate": 0.00026610520772381996,
+      "loss": 0.2048,
+      "step": 1646
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.479211688041687,
+      "learning_rate": 0.000265544453911616,
+      "loss": 0.2126,
+      "step": 1647
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4093790650367737,
+      "learning_rate": 0.00026498407785403795,
+      "loss": 0.1619,
+      "step": 1648
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.48506611585617065,
+      "learning_rate": 0.0002644240804539629,
+      "loss": 0.2059,
+      "step": 1649
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.45267125964164734,
+      "learning_rate": 0.0002638644626136587,
+      "loss": 0.1555,
+      "step": 1650
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4000639021396637,
+      "learning_rate": 0.00026330522523478083,
+      "loss": 0.1957,
+      "step": 1651
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5435691475868225,
+      "learning_rate": 0.0002627463692183727,
+      "loss": 0.3312,
+      "step": 1652
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8608721494674683,
+      "learning_rate": 0.00026218789546486235,
+      "loss": 0.239,
+      "step": 1653
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.38197168707847595,
+      "learning_rate": 0.00026162980487406257,
+      "loss": 0.1573,
+      "step": 1654
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.9562878012657166,
+      "learning_rate": 0.00026107209834516854,
+      "loss": 0.2148,
+      "step": 1655
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.40003103017807007,
+      "learning_rate": 0.0002605147767767564,
+      "loss": 0.1589,
+      "step": 1656
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.9941704869270325,
+      "learning_rate": 0.0002599578410667827,
+      "loss": 0.5321,
+      "step": 1657
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4456992447376251,
+      "learning_rate": 0.00025940129211258144,
+      "loss": 0.2307,
+      "step": 1658
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.49607011675834656,
+      "learning_rate": 0.0002588451308108645,
+      "loss": 0.3218,
+      "step": 1659
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7094647288322449,
+      "learning_rate": 0.000258289358057718,
+      "loss": 0.1852,
+      "step": 1660
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6584123969078064,
+      "learning_rate": 0.0002577339747486033,
+      "loss": 0.2638,
+      "step": 1661
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5190715193748474,
+      "learning_rate": 0.00025717898177835295,
+      "loss": 0.1801,
+      "step": 1662
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.35140207409858704,
+      "learning_rate": 0.0002566243800411719,
+      "loss": 0.1694,
+      "step": 1663
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.44930392503738403,
+      "learning_rate": 0.0002560701704306336,
+      "loss": 0.1852,
+      "step": 1664
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.404563844203949,
+      "learning_rate": 0.00025551635383968066,
+      "loss": 0.1084,
+      "step": 1665
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5529747009277344,
+      "learning_rate": 0.0002549629311606215,
+      "loss": 0.3302,
+      "step": 1666
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.758722186088562,
+      "learning_rate": 0.000254409903285131,
+      "loss": 0.2027,
+      "step": 1667
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.2261089086532593,
+      "learning_rate": 0.0002538572711042469,
+      "loss": 0.3965,
+      "step": 1668
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.41045820713043213,
+      "learning_rate": 0.00025330503550837,
+      "loss": 0.1821,
+      "step": 1669
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.442655086517334,
+      "learning_rate": 0.0002527531973872617,
+      "loss": 0.1997,
+      "step": 1670
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6825320720672607,
+      "learning_rate": 0.0002522017576300434,
+      "loss": 0.3314,
+      "step": 1671
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3625807762145996,
+      "learning_rate": 0.0002516507171251944,
+      "loss": 0.1375,
+      "step": 1672
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.35880157351493835,
+      "learning_rate": 0.00025110007676055107,
+      "loss": 0.1173,
+      "step": 1673
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.31605076789855957,
+      "learning_rate": 0.0002505498374233044,
+      "loss": 0.1225,
+      "step": 1674
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.48455095291137695,
+      "learning_rate": 0.0002500000000000001,
+      "loss": 0.1861,
+      "step": 1675
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3977052867412567,
+      "learning_rate": 0.00024945056537653543,
+      "loss": 0.1707,
+      "step": 1676
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.46795257925987244,
+      "learning_rate": 0.0002489015344381596,
+      "loss": 0.1903,
+      "step": 1677
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5739725828170776,
+      "learning_rate": 0.0002483529080694705,
+      "loss": 0.1455,
+      "step": 1678
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5169729590415955,
+      "learning_rate": 0.00024780468715441455,
+      "loss": 0.1514,
+      "step": 1679
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5376644134521484,
+      "learning_rate": 0.0002472568725762853,
+      "loss": 0.2807,
+      "step": 1680
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5058024525642395,
+      "learning_rate": 0.0002467094652177209,
+      "loss": 0.1754,
+      "step": 1681
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.45354458689689636,
+      "learning_rate": 0.000246162465960704,
+      "loss": 0.1852,
+      "step": 1682
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6148148775100708,
+      "learning_rate": 0.00024561587568655926,
+      "loss": 0.2798,
+      "step": 1683
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4131772220134735,
+      "learning_rate": 0.00024506969527595275,
+      "loss": 0.1745,
+      "step": 1684
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5810641050338745,
+      "learning_rate": 0.00024452392560888974,
+      "loss": 0.251,
+      "step": 1685
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6910896897315979,
+      "learning_rate": 0.0002439785675647143,
+      "loss": 0.1726,
+      "step": 1686
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.37305381894111633,
+      "learning_rate": 0.00024343362202210666,
+      "loss": 0.0968,
+      "step": 1687
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8382201790809631,
+      "learning_rate": 0.00024288908985908302,
+      "loss": 0.269,
+      "step": 1688
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5739067792892456,
+      "learning_rate": 0.00024234497195299288,
+      "loss": 0.1913,
+      "step": 1689
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5062401294708252,
+      "learning_rate": 0.00024180126918051909,
+      "loss": 0.2063,
+      "step": 1690
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6002269983291626,
+      "learning_rate": 0.00024125798241767484,
+      "loss": 0.2001,
+      "step": 1691
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5738735198974609,
+      "learning_rate": 0.00024071511253980366,
+      "loss": 0.2078,
+      "step": 1692
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6711931228637695,
+      "learning_rate": 0.000240172660421577,
+      "loss": 0.3019,
+      "step": 1693
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5227352976799011,
+      "learning_rate": 0.0002396306269369935,
+      "loss": 0.2161,
+      "step": 1694
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.84751296043396,
+      "learning_rate": 0.00023908901295937712,
+      "loss": 0.2445,
+      "step": 1695
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2735321819782257,
+      "learning_rate": 0.00023854781936137576,
+      "loss": 0.0846,
+      "step": 1696
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5112348794937134,
+      "learning_rate": 0.00023800704701496051,
+      "loss": 0.2115,
+      "step": 1697
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7543137669563293,
+      "learning_rate": 0.00023746669679142312,
+      "loss": 0.2015,
+      "step": 1698
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.47386452555656433,
+      "learning_rate": 0.00023692676956137582,
+      "loss": 0.2206,
+      "step": 1699
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4818185567855835,
+      "learning_rate": 0.00023638726619474876,
+      "loss": 0.1663,
+      "step": 1700
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.9844948053359985,
+      "learning_rate": 0.0002358481875607897,
+      "loss": 0.2561,
+      "step": 1701
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8404081463813782,
+      "learning_rate": 0.00023530953452806141,
+      "loss": 0.1825,
+      "step": 1702
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5917308330535889,
+      "learning_rate": 0.00023477130796444173,
+      "loss": 0.1811,
+      "step": 1703
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5200976133346558,
+      "learning_rate": 0.00023423350873712058,
+      "loss": 0.2036,
+      "step": 1704
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6934298276901245,
+      "learning_rate": 0.00023369613771260007,
+      "loss": 0.3614,
+      "step": 1705
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8989263772964478,
+      "learning_rate": 0.0002331591957566917,
+      "loss": 0.241,
+      "step": 1706
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5434936881065369,
+      "learning_rate": 0.0002326226837345164,
+      "loss": 0.3054,
+      "step": 1707
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.262990802526474,
+      "learning_rate": 0.00023208660251050156,
+      "loss": 0.1141,
+      "step": 1708
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5658536553382874,
+      "learning_rate": 0.00023155095294838136,
+      "loss": 0.1571,
+      "step": 1709
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3867456912994385,
+      "learning_rate": 0.0002310157359111938,
+      "loss": 0.1179,
+      "step": 1710
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4275045692920685,
+      "learning_rate": 0.0002304809522612802,
+      "loss": 0.1608,
+      "step": 1711
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8404039144515991,
+      "learning_rate": 0.00022994660286028347,
+      "loss": 0.3762,
+      "step": 1712
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.605236291885376,
+      "learning_rate": 0.00022941268856914744,
+      "loss": 0.242,
+      "step": 1713
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5649933815002441,
+      "learning_rate": 0.00022887921024811403,
+      "loss": 0.2379,
+      "step": 1714
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.35869643092155457,
+      "learning_rate": 0.0002283461687567236,
+      "loss": 0.2195,
+      "step": 1715
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.46537691354751587,
+      "learning_rate": 0.00022781356495381184,
+      "loss": 0.1562,
+      "step": 1716
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.664360523223877,
+      "learning_rate": 0.00022728139969751006,
+      "loss": 0.167,
+      "step": 1717
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5735337734222412,
+      "learning_rate": 0.00022674967384524238,
+      "loss": 0.1901,
+      "step": 1718
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4495382010936737,
+      "learning_rate": 0.00022621838825372491,
+      "loss": 0.1868,
+      "step": 1719
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5063818693161011,
+      "learning_rate": 0.00022568754377896516,
+      "loss": 0.1858,
+      "step": 1720
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.509235680103302,
+      "learning_rate": 0.00022515714127625897,
+      "loss": 0.1905,
+      "step": 1721
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6686124801635742,
+      "learning_rate": 0.00022462718160019086,
+      "loss": 0.2351,
+      "step": 1722
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5115715265274048,
+      "learning_rate": 0.00022409766560463118,
+      "loss": 0.1761,
+      "step": 1723
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7995520830154419,
+      "learning_rate": 0.00022356859414273612,
+      "loss": 0.1947,
+      "step": 1724
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8036087155342102,
+      "learning_rate": 0.00022303996806694487,
+      "loss": 0.1972,
+      "step": 1725
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5572486519813538,
+      "learning_rate": 0.0002225117882289799,
+      "loss": 0.1958,
+      "step": 1726
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5391748547554016,
+      "learning_rate": 0.00022198405547984375,
+      "loss": 0.2085,
+      "step": 1727
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7222040891647339,
+      "learning_rate": 0.00022145677066981944,
+      "loss": 0.2021,
+      "step": 1728
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5539470911026001,
+      "learning_rate": 0.00022092993464846772,
+      "loss": 0.166,
+      "step": 1729
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6876240372657776,
+      "learning_rate": 0.00022040354826462666,
+      "loss": 0.2154,
+      "step": 1730
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4845176637172699,
+      "learning_rate": 0.00021987761236640934,
+      "loss": 0.2085,
+      "step": 1731
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4913284480571747,
+      "learning_rate": 0.00021935212780120367,
+      "loss": 0.1673,
+      "step": 1732
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.2622206509113312,
+      "learning_rate": 0.00021882709541566996,
+      "loss": 0.0954,
+      "step": 1733
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.39314186573028564,
+      "learning_rate": 0.0002183025160557398,
+      "loss": 0.2219,
+      "step": 1734
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.33111441135406494,
+      "learning_rate": 0.00021777839056661552,
+      "loss": 0.1113,
+      "step": 1735
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6581352949142456,
+      "learning_rate": 0.00021725471979276733,
+      "loss": 0.39,
+      "step": 1736
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5542645454406738,
+      "learning_rate": 0.00021673150457793373,
+      "loss": 0.2658,
+      "step": 1737
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5880759954452515,
+      "learning_rate": 0.0002162087457651183,
+      "loss": 0.2082,
+      "step": 1738
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3208143413066864,
+      "learning_rate": 0.00021568644419659005,
+      "loss": 0.1535,
+      "step": 1739
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.47254377603530884,
+      "learning_rate": 0.0002151646007138806,
+      "loss": 0.3424,
+      "step": 1740
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.4164371490478516,
+      "learning_rate": 0.0002146432161577842,
+      "loss": 0.7138,
+      "step": 1741
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4458087980747223,
+      "learning_rate": 0.00021412229136835497,
+      "loss": 0.1608,
+      "step": 1742
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.32590383291244507,
+      "learning_rate": 0.0002136018271849069,
+      "loss": 0.1408,
+      "step": 1743
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.38898277282714844,
+      "learning_rate": 0.00021308182444601127,
+      "loss": 0.1244,
+      "step": 1744
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2702401876449585,
+      "learning_rate": 0.0002125622839894964,
+      "loss": 0.1014,
+      "step": 1745
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5780808329582214,
+      "learning_rate": 0.00021204320665244532,
+      "loss": 0.2349,
+      "step": 1746
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5132737755775452,
+      "learning_rate": 0.0002115245932711954,
+      "loss": 0.2566,
+      "step": 1747
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.45856451988220215,
+      "learning_rate": 0.00021100644468133572,
+      "loss": 0.2446,
+      "step": 1748
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6223645806312561,
+      "learning_rate": 0.00021048876171770754,
+      "loss": 0.191,
+      "step": 1749
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6456278562545776,
+      "learning_rate": 0.00020997154521440098,
+      "loss": 0.2461,
+      "step": 1750
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7013435363769531,
+      "learning_rate": 0.00020945479600475482,
+      "loss": 0.3973,
+      "step": 1751
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.41678425669670105,
+      "learning_rate": 0.00020893851492135535,
+      "loss": 0.1694,
+      "step": 1752
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.36316898465156555,
+      "learning_rate": 0.00020842270279603403,
+      "loss": 0.155,
+      "step": 1753
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4271506667137146,
+      "learning_rate": 0.00020790736045986734,
+      "loss": 0.2377,
+      "step": 1754
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6778103113174438,
+      "learning_rate": 0.0002073924887431744,
+      "loss": 0.2586,
+      "step": 1755
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.413908451795578,
+      "learning_rate": 0.00020687808847551608,
+      "loss": 0.1569,
+      "step": 1756
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.32623785734176636,
+      "learning_rate": 0.00020636416048569372,
+      "loss": 0.121,
+      "step": 1757
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.432521253824234,
+      "learning_rate": 0.00020585070560174808,
+      "loss": 0.2269,
+      "step": 1758
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6957417130470276,
+      "learning_rate": 0.0002053377246509569,
+      "loss": 0.2335,
+      "step": 1759
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7314872741699219,
+      "learning_rate": 0.00020482521845983521,
+      "loss": 0.1688,
+      "step": 1760
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6433822512626648,
+      "learning_rate": 0.00020431318785413228,
+      "loss": 0.2144,
+      "step": 1761
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.533151388168335,
+      "learning_rate": 0.00020380163365883188,
+      "loss": 0.1716,
+      "step": 1762
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.42821067571640015,
+      "learning_rate": 0.00020329055669814934,
+      "loss": 0.1793,
+      "step": 1763
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.617129921913147,
+      "learning_rate": 0.00020277995779553192,
+      "loss": 0.2596,
+      "step": 1764
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4524945616722107,
+      "learning_rate": 0.00020226983777365604,
+      "loss": 0.2271,
+      "step": 1765
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5111883878707886,
+      "learning_rate": 0.0002017601974544269,
+      "loss": 0.251,
+      "step": 1766
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4560619294643402,
+      "learning_rate": 0.0002012510376589764,
+      "loss": 0.1574,
+      "step": 1767
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7710248231887817,
+      "learning_rate": 0.00020074235920766287,
+      "loss": 0.2514,
+      "step": 1768
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4914345145225525,
+      "learning_rate": 0.00020023416292006829,
+      "loss": 0.2017,
+      "step": 1769
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3698033094406128,
+      "learning_rate": 0.00019972644961499853,
+      "loss": 0.2467,
+      "step": 1770
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7500527501106262,
+      "learning_rate": 0.00019921922011048067,
+      "loss": 0.2409,
+      "step": 1771
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6877529621124268,
+      "learning_rate": 0.0001987124752237628,
+      "loss": 0.2254,
+      "step": 1772
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.46513083577156067,
+      "learning_rate": 0.00019820621577131188,
+      "loss": 0.1627,
+      "step": 1773
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4881384074687958,
+      "learning_rate": 0.00019770044256881258,
+      "loss": 0.202,
+      "step": 1774
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4608810544013977,
+      "learning_rate": 0.00019719515643116677,
+      "loss": 0.148,
+      "step": 1775
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.566243052482605,
+      "learning_rate": 0.00019669035817249076,
+      "loss": 0.2242,
+      "step": 1776
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5340837836265564,
+      "learning_rate": 0.00019618604860611555,
+      "loss": 0.2695,
+      "step": 1777
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3891966938972473,
+      "learning_rate": 0.000195682228544584,
+      "loss": 0.1192,
+      "step": 1778
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6433067321777344,
+      "learning_rate": 0.00019517889879965106,
+      "loss": 0.3514,
+      "step": 1779
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5262988209724426,
+      "learning_rate": 0.0001946760601822809,
+      "loss": 0.2493,
+      "step": 1780
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5620471239089966,
+      "learning_rate": 0.00019417371350264717,
+      "loss": 0.1995,
+      "step": 1781
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.29241397976875305,
+      "learning_rate": 0.00019367185957013024,
+      "loss": 0.1497,
+      "step": 1782
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.39494335651397705,
+      "learning_rate": 0.00019317049919331702,
+      "loss": 0.1394,
+      "step": 1783
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.35671600699424744,
+      "learning_rate": 0.0001926696331799988,
+      "loss": 0.1286,
+      "step": 1784
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5461544394493103,
+      "learning_rate": 0.00019216926233717085,
+      "loss": 0.1422,
+      "step": 1785
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.41858693957328796,
+      "learning_rate": 0.00019166938747103014,
+      "loss": 0.1294,
+      "step": 1786
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8093598484992981,
+      "learning_rate": 0.00019117000938697492,
+      "loss": 0.3223,
+      "step": 1787
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.39399078488349915,
+      "learning_rate": 0.0001906711288896028,
+      "loss": 0.155,
+      "step": 1788
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5278188586235046,
+      "learning_rate": 0.00019017274678270946,
+      "loss": 0.1909,
+      "step": 1789
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5749332308769226,
+      "learning_rate": 0.00018967486386928817,
+      "loss": 0.2275,
+      "step": 1790
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.9052302837371826,
+      "learning_rate": 0.0001891774809515273,
+      "loss": 0.2647,
+      "step": 1791
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5835824608802795,
+      "learning_rate": 0.00018868059883081012,
+      "loss": 0.2442,
+      "step": 1792
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.44197699427604675,
+      "learning_rate": 0.00018818421830771255,
+      "loss": 0.1602,
+      "step": 1793
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7265429496765137,
+      "learning_rate": 0.00018768834018200288,
+      "loss": 0.252,
+      "step": 1794
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6587382555007935,
+      "learning_rate": 0.00018719296525263924,
+      "loss": 0.2566,
+      "step": 1795
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3740403950214386,
+      "learning_rate": 0.0001866980943177699,
+      "loss": 0.1501,
+      "step": 1796
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5355677604675293,
+      "learning_rate": 0.00018620372817473003,
+      "loss": 0.2259,
+      "step": 1797
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6001706719398499,
+      "learning_rate": 0.00018570986762004243,
+      "loss": 0.2107,
+      "step": 1798
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7202430367469788,
+      "learning_rate": 0.00018521651344941464,
+      "loss": 0.2612,
+      "step": 1799
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6108411550521851,
+      "learning_rate": 0.0001847236664577389,
+      "loss": 0.2684,
+      "step": 1800
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.43881139159202576,
+      "learning_rate": 0.0001842313274390896,
+      "loss": 0.2232,
+      "step": 1801
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.47748687863349915,
+      "learning_rate": 0.00018373949718672344,
+      "loss": 0.2381,
+      "step": 1802
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.27354443073272705,
+      "learning_rate": 0.00018324817649307667,
+      "loss": 0.1162,
+      "step": 1803
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6101070046424866,
+      "learning_rate": 0.00018275736614976518,
+      "loss": 0.1861,
+      "step": 1804
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6663463115692139,
+      "learning_rate": 0.00018226706694758193,
+      "loss": 0.166,
+      "step": 1805
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6441879868507385,
+      "learning_rate": 0.00018177727967649704,
+      "loss": 0.2859,
+      "step": 1806
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5756428837776184,
+      "learning_rate": 0.00018128800512565513,
+      "loss": 0.1952,
+      "step": 1807
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.9916976690292358,
+      "learning_rate": 0.00018079924408337538,
+      "loss": 0.2634,
+      "step": 1808
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3379478454589844,
+      "learning_rate": 0.0001803109973371489,
+      "loss": 0.1281,
+      "step": 1809
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6759968996047974,
+      "learning_rate": 0.0001798232656736389,
+      "loss": 0.2954,
+      "step": 1810
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5103501677513123,
+      "learning_rate": 0.00017933604987867814,
+      "loss": 0.2922,
+      "step": 1811
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.49126997590065,
+      "learning_rate": 0.00017884935073726822,
+      "loss": 0.2018,
+      "step": 1812
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4552325904369354,
+      "learning_rate": 0.0001783631690335788,
+      "loss": 0.2457,
+      "step": 1813
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.41360795497894287,
+      "learning_rate": 0.0001778775055509453,
+      "loss": 0.1749,
+      "step": 1814
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7234411835670471,
+      "learning_rate": 0.00017739236107186857,
+      "loss": 0.2177,
+      "step": 1815
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5025150775909424,
+      "learning_rate": 0.00017690773637801295,
+      "loss": 0.2095,
+      "step": 1816
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.40785863995552063,
+      "learning_rate": 0.00017642363225020557,
+      "loss": 0.112,
+      "step": 1817
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5719576478004456,
+      "learning_rate": 0.00017594004946843456,
+      "loss": 0.2146,
+      "step": 1818
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6989184021949768,
+      "learning_rate": 0.00017545698881184834,
+      "loss": 0.3492,
+      "step": 1819
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5603457689285278,
+      "learning_rate": 0.00017497445105875377,
+      "loss": 0.2121,
+      "step": 1820
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5093814730644226,
+      "learning_rate": 0.00017449243698661554,
+      "loss": 0.1803,
+      "step": 1821
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6010790467262268,
+      "learning_rate": 0.00017401094737205415,
+      "loss": 0.2151,
+      "step": 1822
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5446028113365173,
+      "learning_rate": 0.00017352998299084572,
+      "loss": 0.1364,
+      "step": 1823
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8248989582061768,
+      "learning_rate": 0.00017304954461791938,
+      "loss": 0.202,
+      "step": 1824
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5732833743095398,
+      "learning_rate": 0.0001725696330273575,
+      "loss": 0.2422,
+      "step": 1825
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.44641679525375366,
+      "learning_rate": 0.00017209024899239296,
+      "loss": 0.1533,
+      "step": 1826
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4699559807777405,
+      "learning_rate": 0.00017161139328540932,
+      "loss": 0.1537,
+      "step": 1827
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8324146270751953,
+      "learning_rate": 0.0001711330666779385,
+      "loss": 0.2741,
+      "step": 1828
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.479613333940506,
+      "learning_rate": 0.00017065526994065972,
+      "loss": 0.1411,
+      "step": 1829
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.38805508613586426,
+      "learning_rate": 0.00017017800384339925,
+      "loss": 0.1537,
+      "step": 1830
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6732274889945984,
+      "learning_rate": 0.00016970126915512757,
+      "loss": 0.1509,
+      "step": 1831
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7822940945625305,
+      "learning_rate": 0.0001692250666439596,
+      "loss": 0.2121,
+      "step": 1832
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.64760822057724,
+      "learning_rate": 0.00016874939707715216,
+      "loss": 0.2731,
+      "step": 1833
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.2728150188922882,
+      "learning_rate": 0.00016827426122110412,
+      "loss": 0.1311,
+      "step": 1834
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3668431043624878,
+      "learning_rate": 0.00016779965984135375,
+      "loss": 0.1351,
+      "step": 1835
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6833921074867249,
+      "learning_rate": 0.00016732559370257881,
+      "loss": 0.362,
+      "step": 1836
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7363101840019226,
+      "learning_rate": 0.00016685206356859401,
+      "loss": 0.2317,
+      "step": 1837
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4533635377883911,
+      "learning_rate": 0.00016637907020235115,
+      "loss": 0.191,
+      "step": 1838
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3575870990753174,
+      "learning_rate": 0.00016590661436593663,
+      "loss": 0.1433,
+      "step": 1839
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.39928552508354187,
+      "learning_rate": 0.00016543469682057105,
+      "loss": 0.2191,
+      "step": 1840
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5755822062492371,
+      "learning_rate": 0.00016496331832660744,
+      "loss": 0.2379,
+      "step": 1841
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.47988295555114746,
+      "learning_rate": 0.00016449247964353092,
+      "loss": 0.2464,
+      "step": 1842
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7936148047447205,
+      "learning_rate": 0.00016402218152995607,
+      "loss": 0.2892,
+      "step": 1843
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.206404447555542,
+      "learning_rate": 0.0001635524247436273,
+      "loss": 0.3224,
+      "step": 1844
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5721861720085144,
+      "learning_rate": 0.00016308321004141607,
+      "loss": 0.2347,
+      "step": 1845
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3795848786830902,
+      "learning_rate": 0.0001626145381793212,
+      "loss": 0.12,
+      "step": 1846
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4160403311252594,
+      "learning_rate": 0.0001621464099124661,
+      "loss": 0.1518,
+      "step": 1847
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.41612157225608826,
+      "learning_rate": 0.00016167882599509904,
+      "loss": 0.1933,
+      "step": 1848
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6765478849411011,
+      "learning_rate": 0.0001612117871805907,
+      "loss": 0.2949,
+      "step": 1849
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7778146266937256,
+      "learning_rate": 0.00016074529422143398,
+      "loss": 0.227,
+      "step": 1850
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5555291771888733,
+      "learning_rate": 0.00016027934786924186,
+      "loss": 0.2736,
+      "step": 1851
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.29629385471343994,
+      "learning_rate": 0.0001598139488747467,
+      "loss": 0.1285,
+      "step": 1852
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6322994232177734,
+      "learning_rate": 0.00015934909798779933,
+      "loss": 0.2328,
+      "step": 1853
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5554819107055664,
+      "learning_rate": 0.00015888479595736694,
+      "loss": 0.1722,
+      "step": 1854
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4540086090564728,
+      "learning_rate": 0.00015842104353153285,
+      "loss": 0.1764,
+      "step": 1855
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4212754964828491,
+      "learning_rate": 0.00015795784145749452,
+      "loss": 0.1199,
+      "step": 1856
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.455199271440506,
+      "learning_rate": 0.00015749519048156307,
+      "loss": 0.1745,
+      "step": 1857
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.284004271030426,
+      "learning_rate": 0.00015703309134916117,
+      "loss": 0.1328,
+      "step": 1858
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6243090033531189,
+      "learning_rate": 0.00015657154480482293,
+      "loss": 0.2105,
+      "step": 1859
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5791494250297546,
+      "learning_rate": 0.0001561105515921915,
+      "loss": 0.2661,
+      "step": 1860
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.45375970005989075,
+      "learning_rate": 0.00015565011245401928,
+      "loss": 0.1987,
+      "step": 1861
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5331457853317261,
+      "learning_rate": 0.0001551902281321651,
+      "loss": 0.1933,
+      "step": 1862
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5254542231559753,
+      "learning_rate": 0.0001547308993675946,
+      "loss": 0.2163,
+      "step": 1863
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5726532936096191,
+      "learning_rate": 0.00015427212690037774,
+      "loss": 0.1649,
+      "step": 1864
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8186507821083069,
+      "learning_rate": 0.00015381391146968864,
+      "loss": 0.2106,
+      "step": 1865
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6159462332725525,
+      "learning_rate": 0.00015335625381380363,
+      "loss": 0.1243,
+      "step": 1866
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4144650101661682,
+      "learning_rate": 0.00015289915467010028,
+      "loss": 0.2022,
+      "step": 1867
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4135063588619232,
+      "learning_rate": 0.00015244261477505676,
+      "loss": 0.1061,
+      "step": 1868
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.43412768840789795,
+      "learning_rate": 0.00015198663486424947,
+      "loss": 0.103,
+      "step": 1869
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8138027191162109,
+      "learning_rate": 0.00015153121567235335,
+      "loss": 0.3794,
+      "step": 1870
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4338344633579254,
+      "learning_rate": 0.00015107635793313912,
+      "loss": 0.1355,
+      "step": 1871
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5566967129707336,
+      "learning_rate": 0.00015062206237947363,
+      "loss": 0.4475,
+      "step": 1872
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6075343489646912,
+      "learning_rate": 0.00015016832974331724,
+      "loss": 0.1471,
+      "step": 1873
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6454697847366333,
+      "learning_rate": 0.00014971516075572405,
+      "loss": 0.1986,
+      "step": 1874
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4726046323776245,
+      "learning_rate": 0.00014926255614683932,
+      "loss": 0.162,
+      "step": 1875
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5148561000823975,
+      "learning_rate": 0.00014881051664589957,
+      "loss": 0.3156,
+      "step": 1876
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.32184407114982605,
+      "learning_rate": 0.00014835904298123026,
+      "loss": 0.1712,
+      "step": 1877
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.9718727469444275,
+      "learning_rate": 0.0001479081358802458,
+      "loss": 0.2574,
+      "step": 1878
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.5344032049179077,
+      "learning_rate": 0.00014745779606944716,
+      "loss": 0.5984,
+      "step": 1879
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.32570624351501465,
+      "learning_rate": 0.0001470080242744218,
+      "loss": 0.0851,
+      "step": 1880
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5076736807823181,
+      "learning_rate": 0.00014655882121984137,
+      "loss": 0.2674,
+      "step": 1881
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5638168454170227,
+      "learning_rate": 0.00014611018762946215,
+      "loss": 0.3874,
+      "step": 1882
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4560689926147461,
+      "learning_rate": 0.00014566212422612157,
+      "loss": 0.1709,
+      "step": 1883
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5444175004959106,
+      "learning_rate": 0.00014521463173173966,
+      "loss": 0.2326,
+      "step": 1884
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.493664413690567,
+      "learning_rate": 0.00014476771086731566,
+      "loss": 0.1884,
+      "step": 1885
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5806133151054382,
+      "learning_rate": 0.00014432136235292847,
+      "loss": 0.2034,
+      "step": 1886
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7597365379333496,
+      "learning_rate": 0.00014387558690773429,
+      "loss": 0.2378,
+      "step": 1887
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3772962987422943,
+      "learning_rate": 0.00014343038524996642,
+      "loss": 0.0901,
+      "step": 1888
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.41952499747276306,
+      "learning_rate": 0.00014298575809693355,
+      "loss": 0.1585,
+      "step": 1889
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5933902263641357,
+      "learning_rate": 0.00014254170616501827,
+      "loss": 0.1736,
+      "step": 1890
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5449525713920593,
+      "learning_rate": 0.00014209823016967722,
+      "loss": 0.1975,
+      "step": 1891
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.867601752281189,
+      "learning_rate": 0.0001416553308254383,
+      "loss": 0.2421,
+      "step": 1892
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6079220175743103,
+      "learning_rate": 0.00014121300884590099,
+      "loss": 0.2398,
+      "step": 1893
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4159815013408661,
+      "learning_rate": 0.00014077126494373377,
+      "loss": 0.1342,
+      "step": 1894
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.43129321932792664,
+      "learning_rate": 0.00014033009983067452,
+      "loss": 0.271,
+      "step": 1895
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5323105454444885,
+      "learning_rate": 0.00013988951421752788,
+      "loss": 0.1468,
+      "step": 1896
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7480573654174805,
+      "learning_rate": 0.0001394495088141654,
+      "loss": 0.2045,
+      "step": 1897
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6981969475746155,
+      "learning_rate": 0.00013901008432952322,
+      "loss": 0.1712,
+      "step": 1898
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3247562348842621,
+      "learning_rate": 0.00013857124147160206,
+      "loss": 0.0994,
+      "step": 1899
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.660347580909729,
+      "learning_rate": 0.0001381329809474649,
+      "loss": 0.2229,
+      "step": 1900
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7265562415122986,
+      "learning_rate": 0.00013769530346323722,
+      "loss": 0.3079,
+      "step": 1901
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6675474643707275,
+      "learning_rate": 0.00013725820972410437,
+      "loss": 0.1848,
+      "step": 1902
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3202802836894989,
+      "learning_rate": 0.00013682170043431176,
+      "loss": 0.0916,
+      "step": 1903
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4072101414203644,
+      "learning_rate": 0.00013638577629716264,
+      "loss": 0.1242,
+      "step": 1904
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5636209845542908,
+      "learning_rate": 0.00013595043801501794,
+      "loss": 0.4426,
+      "step": 1905
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6380083560943604,
+      "learning_rate": 0.00013551568628929433,
+      "loss": 0.2283,
+      "step": 1906
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.39248985052108765,
+      "learning_rate": 0.00013508152182046336,
+      "loss": 0.1714,
+      "step": 1907
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.341219037771225,
+      "learning_rate": 0.00013464794530805075,
+      "loss": 0.1751,
+      "step": 1908
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5375643372535706,
+      "learning_rate": 0.0001342149574506345,
+      "loss": 0.1822,
+      "step": 1909
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6201104521751404,
+      "learning_rate": 0.00013378255894584462,
+      "loss": 0.1867,
+      "step": 1910
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.600313127040863,
+      "learning_rate": 0.000133350750490361,
+      "loss": 0.2193,
+      "step": 1911
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6699445247650146,
+      "learning_rate": 0.00013291953277991348,
+      "loss": 0.2115,
+      "step": 1912
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6619518995285034,
+      "learning_rate": 0.00013248890650927948,
+      "loss": 0.1441,
+      "step": 1913
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6789380311965942,
+      "learning_rate": 0.00013205887237228398,
+      "loss": 0.1201,
+      "step": 1914
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.9208201766014099,
+      "learning_rate": 0.00013162943106179747,
+      "loss": 0.2017,
+      "step": 1915
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4957718253135681,
+      "learning_rate": 0.00013120058326973583,
+      "loss": 0.1847,
+      "step": 1916
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5457937717437744,
+      "learning_rate": 0.00013077232968705804,
+      "loss": 0.1766,
+      "step": 1917
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5686308741569519,
+      "learning_rate": 0.00013034467100376623,
+      "loss": 0.168,
+      "step": 1918
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3089151680469513,
+      "learning_rate": 0.00012991760790890362,
+      "loss": 0.1012,
+      "step": 1919
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8191972970962524,
+      "learning_rate": 0.00012949114109055414,
+      "loss": 0.3339,
+      "step": 1920
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5798625349998474,
+      "learning_rate": 0.00012906527123584082,
+      "loss": 0.1902,
+      "step": 1921
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.39634591341018677,
+      "learning_rate": 0.00012863999903092472,
+      "loss": 0.1234,
+      "step": 1922
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.9583735466003418,
+      "learning_rate": 0.00012821532516100448,
+      "loss": 0.2119,
+      "step": 1923
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.38397395610809326,
+      "learning_rate": 0.00012779125031031413,
+      "loss": 0.1434,
+      "step": 1924
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5622944831848145,
+      "learning_rate": 0.00012736777516212267,
+      "loss": 0.279,
+      "step": 1925
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.32028254866600037,
+      "learning_rate": 0.00012694490039873337,
+      "loss": 0.1056,
+      "step": 1926
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5751548409461975,
+      "learning_rate": 0.00012652262670148135,
+      "loss": 0.2237,
+      "step": 1927
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5107061862945557,
+      "learning_rate": 0.00012610095475073412,
+      "loss": 0.1955,
+      "step": 1928
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.672981858253479,
+      "learning_rate": 0.00012567988522588908,
+      "loss": 0.2878,
+      "step": 1929
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.020228147506714,
+      "learning_rate": 0.00012525941880537307,
+      "loss": 0.194,
+      "step": 1930
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.45506563782691956,
+      "learning_rate": 0.0001248395561666415,
+      "loss": 0.1749,
+      "step": 1931
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5512291789054871,
+      "learning_rate": 0.0001244202979861766,
+      "loss": 0.3374,
+      "step": 1932
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.38538306951522827,
+      "learning_rate": 0.00012400164493948712,
+      "loss": 0.2432,
+      "step": 1933
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4216156005859375,
+      "learning_rate": 0.00012358359770110633,
+      "loss": 0.1518,
+      "step": 1934
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.35135912895202637,
+      "learning_rate": 0.0001231661569445919,
+      "loss": 0.1726,
+      "step": 1935
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5512171387672424,
+      "learning_rate": 0.00012274932334252386,
+      "loss": 0.1376,
+      "step": 1936
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4129510223865509,
+      "learning_rate": 0.00012233309756650457,
+      "loss": 0.1608,
+      "step": 1937
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5901585817337036,
+      "learning_rate": 0.00012191748028715632,
+      "loss": 0.2986,
+      "step": 1938
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.42592722177505493,
+      "learning_rate": 0.00012150247217412185,
+      "loss": 0.1714,
+      "step": 1939
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4816395044326782,
+      "learning_rate": 0.00012108807389606158,
+      "loss": 0.0955,
+      "step": 1940
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7335021495819092,
+      "learning_rate": 0.00012067428612065407,
+      "loss": 0.6466,
+      "step": 1941
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.35399383306503296,
+      "learning_rate": 0.00012026110951459362,
+      "loss": 0.096,
+      "step": 1942
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3159615397453308,
+      "learning_rate": 0.00011984854474359041,
+      "loss": 0.058,
+      "step": 1943
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5437485575675964,
+      "learning_rate": 0.00011943659247236838,
+      "loss": 0.2867,
+      "step": 1944
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5627810955047607,
+      "learning_rate": 0.00011902525336466464,
+      "loss": 0.1891,
+      "step": 1945
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.0886915922164917,
+      "learning_rate": 0.00011861452808322876,
+      "loss": 0.4004,
+      "step": 1946
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4651523232460022,
+      "learning_rate": 0.00011820441728982073,
+      "loss": 0.1487,
+      "step": 1947
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5138458013534546,
+      "learning_rate": 0.00011779492164521116,
+      "loss": 0.2347,
+      "step": 1948
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.41944369673728943,
+      "learning_rate": 0.00011738604180917889,
+      "loss": 0.1381,
+      "step": 1949
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.35789477825164795,
+      "learning_rate": 0.00011697777844051105,
+      "loss": 0.0931,
+      "step": 1950
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.46110159158706665,
+      "learning_rate": 0.00011657013219700107,
+      "loss": 0.2152,
+      "step": 1951
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.31931230425834656,
+      "learning_rate": 0.00011616310373544864,
+      "loss": 0.1747,
+      "step": 1952
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4100174307823181,
+      "learning_rate": 0.00011575669371165748,
+      "loss": 0.216,
+      "step": 1953
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6284206509590149,
+      "learning_rate": 0.00011535090278043537,
+      "loss": 0.1972,
+      "step": 1954
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7394257187843323,
+      "learning_rate": 0.00011494573159559212,
+      "loss": 0.3093,
+      "step": 1955
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6470516920089722,
+      "learning_rate": 0.00011454118080993964,
+      "loss": 0.4517,
+      "step": 1956
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5364859700202942,
+      "learning_rate": 0.00011413725107528956,
+      "loss": 0.1973,
+      "step": 1957
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6607769727706909,
+      "learning_rate": 0.0001137339430424535,
+      "loss": 0.214,
+      "step": 1958
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5710316300392151,
+      "learning_rate": 0.00011333125736124084,
+      "loss": 0.1464,
+      "step": 1959
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4962694048881531,
+      "learning_rate": 0.00011292919468045875,
+      "loss": 0.2211,
+      "step": 1960
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.48959726095199585,
+      "learning_rate": 0.00011252775564791024,
+      "loss": 0.1756,
+      "step": 1961
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5554394721984863,
+      "learning_rate": 0.0001121269409103935,
+      "loss": 0.2451,
+      "step": 1962
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5216475129127502,
+      "learning_rate": 0.00011172675111370123,
+      "loss": 0.1971,
+      "step": 1963
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6767510175704956,
+      "learning_rate": 0.00011132718690261867,
+      "loss": 0.2502,
+      "step": 1964
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2966114580631256,
+      "learning_rate": 0.00011092824892092374,
+      "loss": 0.1321,
+      "step": 1965
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.44479796290397644,
+      "learning_rate": 0.00011052993781138477,
+      "loss": 0.156,
+      "step": 1966
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7554308772087097,
+      "learning_rate": 0.00011013225421576078,
+      "loss": 0.1718,
+      "step": 1967
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5758110880851746,
+      "learning_rate": 0.00010973519877479876,
+      "loss": 0.2898,
+      "step": 1968
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7163851261138916,
+      "learning_rate": 0.00010933877212823462,
+      "loss": 0.4187,
+      "step": 1969
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.192113399505615,
+      "learning_rate": 0.00010894297491479043,
+      "loss": 0.3834,
+      "step": 1970
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.35814329981803894,
+      "learning_rate": 0.00010854780777217466,
+      "loss": 0.0851,
+      "step": 1971
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7150161862373352,
+      "learning_rate": 0.00010815327133708014,
+      "loss": 0.2421,
+      "step": 1972
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.25390350818634033,
+      "learning_rate": 0.00010775936624518396,
+      "loss": 0.0688,
+      "step": 1973
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.9474075436592102,
+      "learning_rate": 0.00010736609313114548,
+      "loss": 0.1584,
+      "step": 1974
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5088942646980286,
+      "learning_rate": 0.00010697345262860636,
+      "loss": 0.1802,
+      "step": 1975
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.9616951942443848,
+      "learning_rate": 0.00010658144537018843,
+      "loss": 0.2257,
+      "step": 1976
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3800095319747925,
+      "learning_rate": 0.00010619007198749386,
+      "loss": 0.1012,
+      "step": 1977
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6980737447738647,
+      "learning_rate": 0.0001057993331111029,
+      "loss": 0.2983,
+      "step": 1978
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4341868460178375,
+      "learning_rate": 0.00010540922937057407,
+      "loss": 0.1238,
+      "step": 1979
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4616117775440216,
+      "learning_rate": 0.00010501976139444191,
+      "loss": 0.1084,
+      "step": 1980
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.38700833916664124,
+      "learning_rate": 0.00010463092981021732,
+      "loss": 0.0814,
+      "step": 1981
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5624858736991882,
+      "learning_rate": 0.00010424273524438521,
+      "loss": 0.1622,
+      "step": 1982
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.303281545639038,
+      "learning_rate": 0.00010385517832240471,
+      "loss": 0.4667,
+      "step": 1983
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.46048659086227417,
+      "learning_rate": 0.0001034682596687071,
+      "loss": 0.1555,
+      "step": 1984
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3692218065261841,
+      "learning_rate": 0.00010308197990669538,
+      "loss": 0.1314,
+      "step": 1985
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.563901960849762,
+      "learning_rate": 0.00010269633965874348,
+      "loss": 0.1798,
+      "step": 1986
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8000838756561279,
+      "learning_rate": 0.00010231133954619448,
+      "loss": 0.2241,
+      "step": 1987
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.43821343779563904,
+      "learning_rate": 0.00010192698018936059,
+      "loss": 0.2289,
+      "step": 1988
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.42093101143836975,
+      "learning_rate": 0.00010154326220752108,
+      "loss": 0.1787,
+      "step": 1989
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6172183752059937,
+      "learning_rate": 0.00010116018621892236,
+      "loss": 0.1326,
+      "step": 1990
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7973692417144775,
+      "learning_rate": 0.00010077775284077601,
+      "loss": 0.1886,
+      "step": 1991
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5027013421058655,
+      "learning_rate": 0.00010039596268925866,
+      "loss": 0.2924,
+      "step": 1992
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6461924910545349,
+      "learning_rate": 0.0001000148163795101,
+      "loss": 0.2076,
+      "step": 1993
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.35654205083847046,
+      "learning_rate": 9.963431452563332e-05,
+      "loss": 0.1127,
+      "step": 1994
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.45396995544433594,
+      "learning_rate": 9.925445774069231e-05,
+      "loss": 0.2252,
+      "step": 1995
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6571694016456604,
+      "learning_rate": 9.887524663671244e-05,
+      "loss": 0.2783,
+      "step": 1996
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5617998838424683,
+      "learning_rate": 9.849668182467808e-05,
+      "loss": 0.1878,
+      "step": 1997
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3240343928337097,
+      "learning_rate": 9.811876391453295e-05,
+      "loss": 0.1295,
+      "step": 1998
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4116528630256653,
+      "learning_rate": 9.774149351517775e-05,
+      "loss": 0.1987,
+      "step": 1999
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.347501277923584,
+      "learning_rate": 9.736487123447069e-05,
+      "loss": 0.1531,
+      "step": 2000
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.44534531235694885,
+      "learning_rate": 9.698889767922514e-05,
+      "loss": 0.1165,
+      "step": 2001
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.48863258957862854,
+      "learning_rate": 9.661357345520938e-05,
+      "loss": 0.187,
+      "step": 2002
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.49325940012931824,
+      "learning_rate": 9.623889916714579e-05,
+      "loss": 0.2751,
+      "step": 2003
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5226061940193176,
+      "learning_rate": 9.586487541870908e-05,
+      "loss": 0.1124,
+      "step": 2004
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4200877547264099,
+      "learning_rate": 9.549150281252633e-05,
+      "loss": 0.1735,
+      "step": 2005
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6373294591903687,
+      "learning_rate": 9.511878195017499e-05,
+      "loss": 0.2072,
+      "step": 2006
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5899367928504944,
+      "learning_rate": 9.474671343218294e-05,
+      "loss": 0.1242,
+      "step": 2007
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.445956826210022,
+      "learning_rate": 9.437529785802646e-05,
+      "loss": 0.1833,
+      "step": 2008
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7763081192970276,
+      "learning_rate": 9.400453582613034e-05,
+      "loss": 0.1798,
+      "step": 2009
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5080363154411316,
+      "learning_rate": 9.363442793386607e-05,
+      "loss": 0.1876,
+      "step": 2010
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7089261412620544,
+      "learning_rate": 9.326497477755114e-05,
+      "loss": 0.2242,
+      "step": 2011
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6344239115715027,
+      "learning_rate": 9.289617695244818e-05,
+      "loss": 0.2086,
+      "step": 2012
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5408456325531006,
+      "learning_rate": 9.25280350527643e-05,
+      "loss": 0.2245,
+      "step": 2013
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4833788275718689,
+      "learning_rate": 9.216054967164917e-05,
+      "loss": 0.1626,
+      "step": 2014
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.518933117389679,
+      "learning_rate": 9.179372140119524e-05,
+      "loss": 0.1829,
+      "step": 2015
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6058471202850342,
+      "learning_rate": 9.142755083243575e-05,
+      "loss": 0.1452,
+      "step": 2016
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3924574851989746,
+      "learning_rate": 9.106203855534478e-05,
+      "loss": 0.1201,
+      "step": 2017
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5917443037033081,
+      "learning_rate": 9.069718515883523e-05,
+      "loss": 0.1957,
+      "step": 2018
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.31607523560523987,
+      "learning_rate": 9.033299123075884e-05,
+      "loss": 0.111,
+      "step": 2019
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.43478962779045105,
+      "learning_rate": 8.996945735790446e-05,
+      "loss": 0.1495,
+      "step": 2020
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.37127575278282166,
+      "learning_rate": 8.960658412599781e-05,
+      "loss": 0.1215,
+      "step": 2021
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5545438528060913,
+      "learning_rate": 8.924437211969983e-05,
+      "loss": 0.2668,
+      "step": 2022
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4196215569972992,
+      "learning_rate": 8.888282192260644e-05,
+      "loss": 0.2006,
+      "step": 2023
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5784420967102051,
+      "learning_rate": 8.852193411724702e-05,
+      "loss": 0.1922,
+      "step": 2024
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.44451579451560974,
+      "learning_rate": 8.816170928508365e-05,
+      "loss": 0.1498,
+      "step": 2025
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.38538575172424316,
+      "learning_rate": 8.78021480065106e-05,
+      "loss": 0.1218,
+      "step": 2026
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.43418964743614197,
+      "learning_rate": 8.744325086085247e-05,
+      "loss": 0.1567,
+      "step": 2027
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.43733420968055725,
+      "learning_rate": 8.708501842636441e-05,
+      "loss": 0.1467,
+      "step": 2028
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6950334310531616,
+      "learning_rate": 8.672745128022997e-05,
+      "loss": 0.305,
+      "step": 2029
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.426617294549942,
+      "learning_rate": 8.637054999856148e-05,
+      "loss": 0.1772,
+      "step": 2030
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6103867292404175,
+      "learning_rate": 8.601431515639768e-05,
+      "loss": 0.1324,
+      "step": 2031
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.45348817110061646,
+      "learning_rate": 8.565874732770429e-05,
+      "loss": 0.1439,
+      "step": 2032
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.41567617654800415,
+      "learning_rate": 8.53038470853716e-05,
+      "loss": 0.1973,
+      "step": 2033
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.38819947838783264,
+      "learning_rate": 8.494961500121501e-05,
+      "loss": 0.0899,
+      "step": 2034
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6108384728431702,
+      "learning_rate": 8.459605164597267e-05,
+      "loss": 0.3177,
+      "step": 2035
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4565349519252777,
+      "learning_rate": 8.424315758930596e-05,
+      "loss": 0.1635,
+      "step": 2036
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3525888919830322,
+      "learning_rate": 8.389093339979726e-05,
+      "loss": 0.125,
+      "step": 2037
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5254977941513062,
+      "learning_rate": 8.353937964495028e-05,
+      "loss": 0.1162,
+      "step": 2038
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.6196539402008057,
+      "learning_rate": 8.318849689118802e-05,
+      "loss": 0.2196,
+      "step": 2039
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.37363407015800476,
+      "learning_rate": 8.283828570385238e-05,
+      "loss": 0.1476,
+      "step": 2040
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5632930994033813,
+      "learning_rate": 8.248874664720374e-05,
+      "loss": 0.1532,
+      "step": 2041
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5555292367935181,
+      "learning_rate": 8.213988028441893e-05,
+      "loss": 0.1485,
+      "step": 2042
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5303210020065308,
+      "learning_rate": 8.179168717759144e-05,
+      "loss": 0.2145,
+      "step": 2043
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.45957398414611816,
+      "learning_rate": 8.144416788772957e-05,
+      "loss": 0.207,
+      "step": 2044
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.6085442304611206,
+      "learning_rate": 8.109732297475635e-05,
+      "loss": 0.1432,
+      "step": 2045
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5784573554992676,
+      "learning_rate": 8.075115299750796e-05,
+      "loss": 0.1688,
+      "step": 2046
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7543654441833496,
+      "learning_rate": 8.040565851373333e-05,
+      "loss": 0.1503,
+      "step": 2047
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5422773361206055,
+      "learning_rate": 8.006084008009284e-05,
+      "loss": 0.2202,
+      "step": 2048
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5515322089195251,
+      "learning_rate": 7.971669825215788e-05,
+      "loss": 0.2292,
+      "step": 2049
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3613136410713196,
+      "learning_rate": 7.937323358440934e-05,
+      "loss": 0.1419,
+      "step": 2050
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.438313364982605,
+      "learning_rate": 7.903044663023756e-05,
+      "loss": 0.1134,
+      "step": 2051
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3975502848625183,
+      "learning_rate": 7.868833794194047e-05,
+      "loss": 0.1168,
+      "step": 2052
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.6506795883178711,
+      "learning_rate": 7.834690807072342e-05,
+      "loss": 0.2133,
+      "step": 2053
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8236292600631714,
+      "learning_rate": 7.800615756669782e-05,
+      "loss": 0.2317,
+      "step": 2054
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.665313720703125,
+      "learning_rate": 7.766608697888095e-05,
+      "loss": 0.2816,
+      "step": 2055
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5899403095245361,
+      "learning_rate": 7.732669685519406e-05,
+      "loss": 0.2123,
+      "step": 2056
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.43901926279067993,
+      "learning_rate": 7.698798774246257e-05,
+      "loss": 0.1749,
+      "step": 2057
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5694443583488464,
+      "learning_rate": 7.664996018641412e-05,
+      "loss": 0.2204,
+      "step": 2058
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.44464337825775146,
+      "learning_rate": 7.631261473167878e-05,
+      "loss": 0.2591,
+      "step": 2059
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.6509461998939514,
+      "learning_rate": 7.597595192178702e-05,
+      "loss": 0.2535,
+      "step": 2060
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.512502133846283,
+      "learning_rate": 7.563997229917002e-05,
+      "loss": 0.1869,
+      "step": 2061
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4892568290233612,
+      "learning_rate": 7.530467640515782e-05,
+      "loss": 0.282,
+      "step": 2062
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5519267916679382,
+      "learning_rate": 7.497006477997875e-05,
+      "loss": 0.1617,
+      "step": 2063
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5957995653152466,
+      "learning_rate": 7.46361379627592e-05,
+      "loss": 0.2236,
+      "step": 2064
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.40321362018585205,
+      "learning_rate": 7.430289649152156e-05,
+      "loss": 0.1161,
+      "step": 2065
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3922900855541229,
+      "learning_rate": 7.397034090318456e-05,
+      "loss": 0.1254,
+      "step": 2066
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.41142556071281433,
+      "learning_rate": 7.363847173356119e-05,
+      "loss": 0.1143,
+      "step": 2067
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.48004427552223206,
+      "learning_rate": 7.330728951735915e-05,
+      "loss": 0.1948,
+      "step": 2068
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.46935975551605225,
+      "learning_rate": 7.297679478817881e-05,
+      "loss": 0.1762,
+      "step": 2069
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5462561249732971,
+      "learning_rate": 7.264698807851328e-05,
+      "loss": 0.2714,
+      "step": 2070
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.18997442722320557,
+      "learning_rate": 7.23178699197467e-05,
+      "loss": 0.0725,
+      "step": 2071
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.32573890686035156,
+      "learning_rate": 7.19894408421542e-05,
+      "loss": 0.1376,
+      "step": 2072
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.2656877040863037,
+      "learning_rate": 7.166170137490035e-05,
+      "loss": 0.0745,
+      "step": 2073
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5588958859443665,
+      "learning_rate": 7.133465204603895e-05,
+      "loss": 0.2026,
+      "step": 2074
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8618218302726746,
+      "learning_rate": 7.100829338251146e-05,
+      "loss": 0.4059,
+      "step": 2075
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6156420707702637,
+      "learning_rate": 7.068262591014695e-05,
+      "loss": 0.2089,
+      "step": 2076
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6823580265045166,
+      "learning_rate": 7.035765015366047e-05,
+      "loss": 0.307,
+      "step": 2077
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4598240554332733,
+      "learning_rate": 7.003336663665293e-05,
+      "loss": 0.2564,
+      "step": 2078
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3518538475036621,
+      "learning_rate": 6.970977588160964e-05,
+      "loss": 0.1067,
+      "step": 2079
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.44161489605903625,
+      "learning_rate": 6.938687840989971e-05,
+      "loss": 0.1661,
+      "step": 2080
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.43504780530929565,
+      "learning_rate": 6.906467474177558e-05,
+      "loss": 0.1497,
+      "step": 2081
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.058307409286499,
+      "learning_rate": 6.874316539637127e-05,
+      "loss": 0.4472,
+      "step": 2082
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.41559553146362305,
+      "learning_rate": 6.842235089170273e-05,
+      "loss": 0.1465,
+      "step": 2083
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.783308207988739,
+      "learning_rate": 6.81022317446659e-05,
+      "loss": 0.2605,
+      "step": 2084
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.2718304395675659,
+      "learning_rate": 6.778280847103668e-05,
+      "loss": 0.084,
+      "step": 2085
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5049070715904236,
+      "learning_rate": 6.746408158546946e-05,
+      "loss": 0.2404,
+      "step": 2086
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5881661772727966,
+      "learning_rate": 6.7146051601497e-05,
+      "loss": 0.3096,
+      "step": 2087
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4110894203186035,
+      "learning_rate": 6.682871903152887e-05,
+      "loss": 0.1338,
+      "step": 2088
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.35153526067733765,
+      "learning_rate": 6.651208438685119e-05,
+      "loss": 0.0712,
+      "step": 2089
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6093126535415649,
+      "learning_rate": 6.619614817762538e-05,
+      "loss": 0.1584,
+      "step": 2090
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6412477493286133,
+      "learning_rate": 6.588091091288784e-05,
+      "loss": 0.2309,
+      "step": 2091
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3465915322303772,
+      "learning_rate": 6.556637310054841e-05,
+      "loss": 0.104,
+      "step": 2092
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3677145540714264,
+      "learning_rate": 6.52525352473905e-05,
+      "loss": 0.1334,
+      "step": 2093
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.40056315064430237,
+      "learning_rate": 6.493939785906927e-05,
+      "loss": 0.1496,
+      "step": 2094
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.48610106110572815,
+      "learning_rate": 6.462696144011149e-05,
+      "loss": 0.1705,
+      "step": 2095
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6832398772239685,
+      "learning_rate": 6.431522649391447e-05,
+      "loss": 0.2715,
+      "step": 2096
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5608856081962585,
+      "learning_rate": 6.40041935227455e-05,
+      "loss": 0.1526,
+      "step": 2097
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8478699922561646,
+      "learning_rate": 6.36938630277405e-05,
+      "loss": 0.2149,
+      "step": 2098
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8250143527984619,
+      "learning_rate": 6.338423550890405e-05,
+      "loss": 0.2673,
+      "step": 2099
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3759734034538269,
+      "learning_rate": 6.307531146510753e-05,
+      "loss": 0.1724,
+      "step": 2100
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7483788728713989,
+      "learning_rate": 6.276709139408938e-05,
+      "loss": 0.2199,
+      "step": 2101
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.496183842420578,
+      "learning_rate": 6.245957579245348e-05,
+      "loss": 0.1806,
+      "step": 2102
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6234924793243408,
+      "learning_rate": 6.21527651556687e-05,
+      "loss": 0.1656,
+      "step": 2103
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4628143608570099,
+      "learning_rate": 6.184665997806832e-05,
+      "loss": 0.1944,
+      "step": 2104
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3779745399951935,
+      "learning_rate": 6.154126075284855e-05,
+      "loss": 0.1688,
+      "step": 2105
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6127331256866455,
+      "learning_rate": 6.123656797206872e-05,
+      "loss": 0.2157,
+      "step": 2106
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7860340476036072,
+      "learning_rate": 6.093258212664937e-05,
+      "loss": 0.2832,
+      "step": 2107
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5054344534873962,
+      "learning_rate": 6.06293037063726e-05,
+      "loss": 0.1407,
+      "step": 2108
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4316577613353729,
+      "learning_rate": 6.032673319988008e-05,
+      "loss": 0.1835,
+      "step": 2109
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8070529103279114,
+      "learning_rate": 6.002487109467347e-05,
+      "loss": 0.2378,
+      "step": 2110
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8368496894836426,
+      "learning_rate": 5.972371787711262e-05,
+      "loss": 0.2302,
+      "step": 2111
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8410806059837341,
+      "learning_rate": 5.94232740324156e-05,
+      "loss": 0.3104,
+      "step": 2112
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5159313678741455,
+      "learning_rate": 5.912354004465709e-05,
+      "loss": 0.1375,
+      "step": 2113
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7453793287277222,
+      "learning_rate": 5.882451639676856e-05,
+      "loss": 0.3513,
+      "step": 2114
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4245200455188751,
+      "learning_rate": 5.852620357053651e-05,
+      "loss": 0.1594,
+      "step": 2115
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4825901985168457,
+      "learning_rate": 5.822860204660252e-05,
+      "loss": 0.2567,
+      "step": 2116
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5375970602035522,
+      "learning_rate": 5.793171230446198e-05,
+      "loss": 0.2063,
+      "step": 2117
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3189495801925659,
+      "learning_rate": 5.76355348224632e-05,
+      "loss": 0.1273,
+      "step": 2118
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4021792411804199,
+      "learning_rate": 5.7340070077807506e-05,
+      "loss": 0.1487,
+      "step": 2119
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4046955704689026,
+      "learning_rate": 5.7045318546547206e-05,
+      "loss": 0.1817,
+      "step": 2120
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.42547404766082764,
+      "learning_rate": 5.6751280703585985e-05,
+      "loss": 0.1607,
+      "step": 2121
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4201669991016388,
+      "learning_rate": 5.645795702267731e-05,
+      "loss": 0.1621,
+      "step": 2122
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3623116910457611,
+      "learning_rate": 5.61653479764242e-05,
+      "loss": 0.1593,
+      "step": 2123
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.35421204566955566,
+      "learning_rate": 5.587345403627803e-05,
+      "loss": 0.1387,
+      "step": 2124
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.366296648979187,
+      "learning_rate": 5.5582275672538315e-05,
+      "loss": 0.1083,
+      "step": 2125
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4885554909706116,
+      "learning_rate": 5.529181335435124e-05,
+      "loss": 0.2163,
+      "step": 2126
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6899632811546326,
+      "learning_rate": 5.5002067549709654e-05,
+      "loss": 0.2344,
+      "step": 2127
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8063470125198364,
+      "learning_rate": 5.471303872545175e-05,
+      "loss": 0.5257,
+      "step": 2128
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6802717447280884,
+      "learning_rate": 5.442472734726062e-05,
+      "loss": 0.1328,
+      "step": 2129
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6845070123672485,
+      "learning_rate": 5.413713387966329e-05,
+      "loss": 0.4564,
+      "step": 2130
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4023562967777252,
+      "learning_rate": 5.3850258786030394e-05,
+      "loss": 0.1358,
+      "step": 2131
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4080947935581207,
+      "learning_rate": 5.3564102528574576e-05,
+      "loss": 0.133,
+      "step": 2132
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.0173286199569702,
+      "learning_rate": 5.327866556835087e-05,
+      "loss": 0.3058,
+      "step": 2133
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5023651719093323,
+      "learning_rate": 5.299394836525506e-05,
+      "loss": 0.1837,
+      "step": 2134
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4773707687854767,
+      "learning_rate": 5.270995137802315e-05,
+      "loss": 0.1941,
+      "step": 2135
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4644302427768707,
+      "learning_rate": 5.242667506423121e-05,
+      "loss": 0.1207,
+      "step": 2136
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3452518582344055,
+      "learning_rate": 5.214411988029355e-05,
+      "loss": 0.1432,
+      "step": 2137
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6030331254005432,
+      "learning_rate": 5.186228628146317e-05,
+      "loss": 0.2071,
+      "step": 2138
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.48162102699279785,
+      "learning_rate": 5.1581174721829994e-05,
+      "loss": 0.2011,
+      "step": 2139
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5776282548904419,
+      "learning_rate": 5.130078565432089e-05,
+      "loss": 0.1709,
+      "step": 2140
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3757902681827545,
+      "learning_rate": 5.1021119530698434e-05,
+      "loss": 0.1188,
+      "step": 2141
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4376067519187927,
+      "learning_rate": 5.0742176801560625e-05,
+      "loss": 0.1617,
+      "step": 2142
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4627598822116852,
+      "learning_rate": 5.0463957916339676e-05,
+      "loss": 0.1832,
+      "step": 2143
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7415582537651062,
+      "learning_rate": 5.01864633233019e-05,
+      "loss": 0.2478,
+      "step": 2144
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7800180912017822,
+      "learning_rate": 4.99096934695461e-05,
+      "loss": 0.2867,
+      "step": 2145
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.479362815618515,
+      "learning_rate": 4.9633648801004015e-05,
+      "loss": 0.124,
+      "step": 2146
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7260376214981079,
+      "learning_rate": 4.9358329762438314e-05,
+      "loss": 0.4363,
+      "step": 2147
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.2685643136501312,
+      "learning_rate": 4.908373679744316e-05,
+      "loss": 0.0968,
+      "step": 2148
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.9176412224769592,
+      "learning_rate": 4.880987034844231e-05,
+      "loss": 0.4028,
+      "step": 2149
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8368576765060425,
+      "learning_rate": 4.853673085668947e-05,
+      "loss": 0.1951,
+      "step": 2150
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.33170264959335327,
+      "learning_rate": 4.82643187622665e-05,
+      "loss": 0.1476,
+      "step": 2151
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4928680956363678,
+      "learning_rate": 4.799263450408387e-05,
+      "loss": 0.2186,
+      "step": 2152
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3743893802165985,
+      "learning_rate": 4.7721678519878906e-05,
+      "loss": 0.1144,
+      "step": 2153
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.29859933257102966,
+      "learning_rate": 4.745145124621586e-05,
+      "loss": 0.1595,
+      "step": 2154
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4638236463069916,
+      "learning_rate": 4.7181953118484556e-05,
+      "loss": 0.1801,
+      "step": 2155
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5337496995925903,
+      "learning_rate": 4.6913184570900436e-05,
+      "loss": 0.1769,
+      "step": 2156
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4615919589996338,
+      "learning_rate": 4.664514603650305e-05,
+      "loss": 0.2222,
+      "step": 2157
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5072957277297974,
+      "learning_rate": 4.637783794715589e-05,
+      "loss": 0.143,
+      "step": 2158
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.34347933530807495,
+      "learning_rate": 4.6111260733545715e-05,
+      "loss": 0.1216,
+      "step": 2159
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6206907629966736,
+      "learning_rate": 4.58454148251814e-05,
+      "loss": 0.1809,
+      "step": 2160
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5323420166969299,
+      "learning_rate": 4.558030065039387e-05,
+      "loss": 0.2531,
+      "step": 2161
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6821646690368652,
+      "learning_rate": 4.5315918636334776e-05,
+      "loss": 0.1666,
+      "step": 2162
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 17.155078887939453,
+      "learning_rate": 4.505226920897637e-05,
+      "loss": 0.1554,
+      "step": 2163
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8804238438606262,
+      "learning_rate": 4.4789352793110305e-05,
+      "loss": 0.3038,
+      "step": 2164
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6869809031486511,
+      "learning_rate": 4.452716981234745e-05,
+      "loss": 0.3591,
+      "step": 2165
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3857569694519043,
+      "learning_rate": 4.426572068911677e-05,
+      "loss": 0.1542,
+      "step": 2166
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5485772490501404,
+      "learning_rate": 4.4005005844665046e-05,
+      "loss": 0.1443,
+      "step": 2167
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6751975417137146,
+      "learning_rate": 4.37450256990557e-05,
+      "loss": 0.3121,
+      "step": 2168
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5036139488220215,
+      "learning_rate": 4.348578067116882e-05,
+      "loss": 0.1378,
+      "step": 2169
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3228605091571808,
+      "learning_rate": 4.322727117869951e-05,
+      "loss": 0.1331,
+      "step": 2170
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7899322509765625,
+      "learning_rate": 4.2969497638158384e-05,
+      "loss": 0.194,
+      "step": 2171
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4198910892009735,
+      "learning_rate": 4.271246046486993e-05,
+      "loss": 0.1081,
+      "step": 2172
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5952415466308594,
+      "learning_rate": 4.24561600729721e-05,
+      "loss": 0.2887,
+      "step": 2173
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4206009805202484,
+      "learning_rate": 4.220059687541617e-05,
+      "loss": 0.1278,
+      "step": 2174
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7451596856117249,
+      "learning_rate": 4.194577128396521e-05,
+      "loss": 0.1805,
+      "step": 2175
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3754965364933014,
+      "learning_rate": 4.1691683709194184e-05,
+      "loss": 0.0756,
+      "step": 2176
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.45300057530403137,
+      "learning_rate": 4.143833456048868e-05,
+      "loss": 0.1907,
+      "step": 2177
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.44321155548095703,
+      "learning_rate": 4.1185724246044884e-05,
+      "loss": 0.172,
+      "step": 2178
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5555132031440735,
+      "learning_rate": 4.0933853172868185e-05,
+      "loss": 0.176,
+      "step": 2179
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6069960594177246,
+      "learning_rate": 4.0682721746773344e-05,
+      "loss": 0.2285,
+      "step": 2180
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4582599103450775,
+      "learning_rate": 4.043233037238281e-05,
+      "loss": 0.0671,
+      "step": 2181
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.49211809039115906,
+      "learning_rate": 4.0182679453127316e-05,
+      "loss": 0.1965,
+      "step": 2182
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.42519688606262207,
+      "learning_rate": 3.993376939124399e-05,
+      "loss": 0.1974,
+      "step": 2183
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4760355055332184,
+      "learning_rate": 3.9685600587776814e-05,
+      "loss": 0.1635,
+      "step": 2184
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8243657350540161,
+      "learning_rate": 3.9438173442575e-05,
+      "loss": 0.3051,
+      "step": 2185
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6314058303833008,
+      "learning_rate": 3.9191488354293146e-05,
+      "loss": 0.1886,
+      "step": 2186
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8862356543540955,
+      "learning_rate": 3.894554572038999e-05,
+      "loss": 0.1478,
+      "step": 2187
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8624824285507202,
+      "learning_rate": 3.870034593712835e-05,
+      "loss": 0.2178,
+      "step": 2188
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.468068391084671,
+      "learning_rate": 3.845588939957373e-05,
+      "loss": 0.1341,
+      "step": 2189
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5962413549423218,
+      "learning_rate": 3.821217650159453e-05,
+      "loss": 0.1545,
+      "step": 2190
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5785366296768188,
+      "learning_rate": 3.7969207635860594e-05,
+      "loss": 0.1613,
+      "step": 2191
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.47664937376976013,
+      "learning_rate": 3.772698319384349e-05,
+      "loss": 0.1817,
+      "step": 2192
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5389711260795593,
+      "learning_rate": 3.748550356581482e-05,
+      "loss": 0.1529,
+      "step": 2193
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6163721680641174,
+      "learning_rate": 3.724476914084657e-05,
+      "loss": 0.1763,
+      "step": 2194
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6463589668273926,
+      "learning_rate": 3.700478030680987e-05,
+      "loss": 0.2533,
+      "step": 2195
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6780967712402344,
+      "learning_rate": 3.676553745037447e-05,
+      "loss": 0.3076,
+      "step": 2196
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4204101264476776,
+      "learning_rate": 3.652704095700849e-05,
+      "loss": 0.1204,
+      "step": 2197
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.24303801357746124,
+      "learning_rate": 3.628929121097707e-05,
+      "loss": 0.1052,
+      "step": 2198
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7799328565597534,
+      "learning_rate": 3.6052288595342705e-05,
+      "loss": 0.2575,
+      "step": 2199
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.820830225944519,
+      "learning_rate": 3.5816033491963716e-05,
+      "loss": 0.2906,
+      "step": 2200
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.515025794506073,
+      "learning_rate": 3.5580526281494216e-05,
+      "loss": 0.1646,
+      "step": 2201
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.41199442744255066,
+      "learning_rate": 3.534576734338324e-05,
+      "loss": 0.1084,
+      "step": 2202
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.35481488704681396,
+      "learning_rate": 3.5111757055874326e-05,
+      "loss": 0.1134,
+      "step": 2203
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4869052767753601,
+      "learning_rate": 3.487849579600455e-05,
+      "loss": 0.1652,
+      "step": 2204
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.44682395458221436,
+      "learning_rate": 3.46459839396045e-05,
+      "loss": 0.1637,
+      "step": 2205
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.49349167943000793,
+      "learning_rate": 3.441422186129689e-05,
+      "loss": 0.173,
+      "step": 2206
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4319952428340912,
+      "learning_rate": 3.418320993449692e-05,
+      "loss": 0.1473,
+      "step": 2207
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7080026268959045,
+      "learning_rate": 3.3952948531410564e-05,
+      "loss": 0.3402,
+      "step": 2208
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6393918991088867,
+      "learning_rate": 3.372343802303507e-05,
+      "loss": 0.1397,
+      "step": 2209
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.379871666431427,
+      "learning_rate": 3.349467877915746e-05,
+      "loss": 0.1576,
+      "step": 2210
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3997807800769806,
+      "learning_rate": 3.3266671168354634e-05,
+      "loss": 0.134,
+      "step": 2211
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3560653626918793,
+      "learning_rate": 3.3039415557992226e-05,
+      "loss": 0.1433,
+      "step": 2212
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.37161263823509216,
+      "learning_rate": 3.2812912314224285e-05,
+      "loss": 0.1445,
+      "step": 2213
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.32512399554252625,
+      "learning_rate": 3.258716180199278e-05,
+      "loss": 0.1491,
+      "step": 2214
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.45097672939300537,
+      "learning_rate": 3.23621643850267e-05,
+      "loss": 0.1338,
+      "step": 2215
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3282792568206787,
+      "learning_rate": 3.21379204258419e-05,
+      "loss": 0.1231,
+      "step": 2216
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7666659951210022,
+      "learning_rate": 3.191443028573993e-05,
+      "loss": 0.2667,
+      "step": 2217
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.39555269479751587,
+      "learning_rate": 3.169169432480806e-05,
+      "loss": 0.1244,
+      "step": 2218
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4426737129688263,
+      "learning_rate": 3.1469712901918244e-05,
+      "loss": 0.2311,
+      "step": 2219
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.356372594833374,
+      "learning_rate": 3.124848637472688e-05,
+      "loss": 0.1687,
+      "step": 2220
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.520386278629303,
+      "learning_rate": 3.1028015099673954e-05,
+      "loss": 0.192,
+      "step": 2221
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5919626355171204,
+      "learning_rate": 3.080829943198277e-05,
+      "loss": 0.0939,
+      "step": 2222
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.37351474165916443,
+      "learning_rate": 3.058933972565897e-05,
+      "loss": 0.1308,
+      "step": 2223
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3607011139392853,
+      "learning_rate": 3.0371136333490312e-05,
+      "loss": 0.153,
+      "step": 2224
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.1393941640853882,
+      "learning_rate": 3.0153689607045842e-05,
+      "loss": 0.4392,
+      "step": 2225
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.31766223907470703,
+      "learning_rate": 2.9936999896675755e-05,
+      "loss": 0.1114,
+      "step": 2226
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.3002568483352661,
+      "learning_rate": 2.9721067551510273e-05,
+      "loss": 0.4896,
+      "step": 2227
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4739823639392853,
+      "learning_rate": 2.9505892919459543e-05,
+      "loss": 0.1524,
+      "step": 2228
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.89152592420578,
+      "learning_rate": 2.9291476347212686e-05,
+      "loss": 0.2135,
+      "step": 2229
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.49700748920440674,
+      "learning_rate": 2.9077818180237692e-05,
+      "loss": 0.1727,
+      "step": 2230
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.9929296374320984,
+      "learning_rate": 2.8864918762780302e-05,
+      "loss": 0.3351,
+      "step": 2231
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.2762332558631897,
+      "learning_rate": 2.8652778437864013e-05,
+      "loss": 0.0778,
+      "step": 2232
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5412799715995789,
+      "learning_rate": 2.844139754728914e-05,
+      "loss": 0.168,
+      "step": 2233
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4540776312351227,
+      "learning_rate": 2.823077643163252e-05,
+      "loss": 0.1809,
+      "step": 2234
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.47652360796928406,
+      "learning_rate": 2.802091543024671e-05,
+      "loss": 0.2425,
+      "step": 2235
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.40920040011405945,
+      "learning_rate": 2.78118148812595e-05,
+      "loss": 0.1246,
+      "step": 2236
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3881155252456665,
+      "learning_rate": 2.760347512157374e-05,
+      "loss": 0.127,
+      "step": 2237
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.42091360688209534,
+      "learning_rate": 2.739589648686619e-05,
+      "loss": 0.1981,
+      "step": 2238
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4713582694530487,
+      "learning_rate": 2.7189079311587595e-05,
+      "loss": 0.3037,
+      "step": 2239
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6591949462890625,
+      "learning_rate": 2.6983023928961405e-05,
+      "loss": 0.1954,
+      "step": 2240
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.34767112135887146,
+      "learning_rate": 2.6777730670984103e-05,
+      "loss": 0.0982,
+      "step": 2241
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5306887626647949,
+      "learning_rate": 2.6573199868423937e-05,
+      "loss": 0.1856,
+      "step": 2242
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.35791003704071045,
+      "learning_rate": 2.636943185082097e-05,
+      "loss": 0.0849,
+      "step": 2243
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.35932284593582153,
+      "learning_rate": 2.616642694648591e-05,
+      "loss": 0.1499,
+      "step": 2244
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3278028964996338,
+      "learning_rate": 2.596418548250029e-05,
+      "loss": 0.1292,
+      "step": 2245
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3723858594894409,
+      "learning_rate": 2.576270778471529e-05,
+      "loss": 0.1014,
+      "step": 2246
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5411638021469116,
+      "learning_rate": 2.5561994177751735e-05,
+      "loss": 0.3914,
+      "step": 2247
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5179588794708252,
+      "learning_rate": 2.536204498499922e-05,
+      "loss": 0.1782,
+      "step": 2248
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3990313708782196,
+      "learning_rate": 2.5162860528615827e-05,
+      "loss": 0.117,
+      "step": 2249
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3464663624763489,
+      "learning_rate": 2.4964441129527336e-05,
+      "loss": 0.0986,
+      "step": 2250
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4069399833679199,
+      "learning_rate": 2.476678710742697e-05,
+      "loss": 0.1375,
+      "step": 2251
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7908629179000854,
+      "learning_rate": 2.4569898780774813e-05,
+      "loss": 0.2084,
+      "step": 2252
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6421664953231812,
+      "learning_rate": 2.4373776466797172e-05,
+      "loss": 0.1788,
+      "step": 2253
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.47193995118141174,
+      "learning_rate": 2.417842048148622e-05,
+      "loss": 0.1399,
+      "step": 2254
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7262789011001587,
+      "learning_rate": 2.3983831139599287e-05,
+      "loss": 0.337,
+      "step": 2255
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.548237144947052,
+      "learning_rate": 2.379000875465881e-05,
+      "loss": 0.123,
+      "step": 2256
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6953282356262207,
+      "learning_rate": 2.359695363895109e-05,
+      "loss": 0.215,
+      "step": 2257
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.40476202964782715,
+      "learning_rate": 2.340466610352654e-05,
+      "loss": 0.1506,
+      "step": 2258
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.0758031606674194,
+      "learning_rate": 2.321314645819855e-05,
+      "loss": 0.1984,
+      "step": 2259
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5065340995788574,
+      "learning_rate": 2.3022395011543685e-05,
+      "loss": 0.1649,
+      "step": 2260
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4201124906539917,
+      "learning_rate": 2.283241207090031e-05,
+      "loss": 0.1678,
+      "step": 2261
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5411754250526428,
+      "learning_rate": 2.264319794236902e-05,
+      "loss": 0.166,
+      "step": 2262
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.49341070652008057,
+      "learning_rate": 2.2454752930811396e-05,
+      "loss": 0.1396,
+      "step": 2263
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3542145788669586,
+      "learning_rate": 2.226707733984995e-05,
+      "loss": 0.1023,
+      "step": 2264
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3271152973175049,
+      "learning_rate": 2.208017147186736e-05,
+      "loss": 0.1617,
+      "step": 2265
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4652055501937866,
+      "learning_rate": 2.1894035628006514e-05,
+      "loss": 0.107,
+      "step": 2266
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6131516695022583,
+      "learning_rate": 2.170867010816907e-05,
+      "loss": 0.2299,
+      "step": 2267
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5888462066650391,
+      "learning_rate": 2.1524075211016013e-05,
+      "loss": 0.1955,
+      "step": 2268
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.0021755695343018,
+      "learning_rate": 2.134025123396638e-05,
+      "loss": 0.5078,
+      "step": 2269
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.49626588821411133,
+      "learning_rate": 2.1157198473197415e-05,
+      "loss": 0.1646,
+      "step": 2270
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6627007722854614,
+      "learning_rate": 2.0974917223643418e-05,
+      "loss": 0.1834,
+      "step": 2271
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.37689921259880066,
+      "learning_rate": 2.079340777899602e-05,
+      "loss": 0.0955,
+      "step": 2272
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3804769217967987,
+      "learning_rate": 2.0612670431703062e-05,
+      "loss": 0.1269,
+      "step": 2273
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5390046238899231,
+      "learning_rate": 2.0432705472968326e-05,
+      "loss": 0.1642,
+      "step": 2274
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4527387022972107,
+      "learning_rate": 2.025351319275137e-05,
+      "loss": 0.1366,
+      "step": 2275
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3216545283794403,
+      "learning_rate": 2.007509387976658e-05,
+      "loss": 0.1264,
+      "step": 2276
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.47823092341423035,
+      "learning_rate": 1.9897447821483115e-05,
+      "loss": 0.1351,
+      "step": 2277
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.48046648502349854,
+      "learning_rate": 1.9720575304124134e-05,
+      "loss": 0.0849,
+      "step": 2278
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.49452316761016846,
+      "learning_rate": 1.9544476612666673e-05,
+      "loss": 0.1521,
+      "step": 2279
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5453348159790039,
+      "learning_rate": 1.9369152030840554e-05,
+      "loss": 0.1759,
+      "step": 2280
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.0472928285598755,
+      "learning_rate": 1.919460184112892e-05,
+      "loss": 0.4168,
+      "step": 2281
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.36070898175239563,
+      "learning_rate": 1.9020826324766703e-05,
+      "loss": 0.1048,
+      "step": 2282
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.24006277322769165,
+      "learning_rate": 1.88478257617411e-05,
+      "loss": 0.0508,
+      "step": 2283
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7325834035873413,
+      "learning_rate": 1.8675600430790306e-05,
+      "loss": 0.2544,
+      "step": 2284
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.1334954500198364,
+      "learning_rate": 1.850415060940386e-05,
+      "loss": 0.3253,
+      "step": 2285
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.38791072368621826,
+      "learning_rate": 1.8333476573821394e-05,
+      "loss": 0.1274,
+      "step": 2286
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.43326082825660706,
+      "learning_rate": 1.8163578599033003e-05,
+      "loss": 0.1844,
+      "step": 2287
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8000028133392334,
+      "learning_rate": 1.7994456958778048e-05,
+      "loss": 0.2442,
+      "step": 2288
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4378860890865326,
+      "learning_rate": 1.782611192554534e-05,
+      "loss": 0.1339,
+      "step": 2289
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.29711124300956726,
+      "learning_rate": 1.765854377057219e-05,
+      "loss": 0.0871,
+      "step": 2290
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.49163389205932617,
+      "learning_rate": 1.7491752763844293e-05,
+      "loss": 0.1452,
+      "step": 2291
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.35742995142936707,
+      "learning_rate": 1.7325739174095302e-05,
+      "loss": 0.0725,
+      "step": 2292
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5720276832580566,
+      "learning_rate": 1.7160503268806083e-05,
+      "loss": 0.2338,
+      "step": 2293
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5191448330879211,
+      "learning_rate": 1.6996045314204735e-05,
+      "loss": 0.1811,
+      "step": 2294
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4045849144458771,
+      "learning_rate": 1.683236557526574e-05,
+      "loss": 0.1241,
+      "step": 2295
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.42636770009994507,
+      "learning_rate": 1.666946431570987e-05,
+      "loss": 0.1014,
+      "step": 2296
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.43911635875701904,
+      "learning_rate": 1.6507341798003396e-05,
+      "loss": 0.1814,
+      "step": 2297
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6515835523605347,
+      "learning_rate": 1.6345998283358142e-05,
+      "loss": 0.1598,
+      "step": 2298
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4561358690261841,
+      "learning_rate": 1.6185434031730615e-05,
+      "loss": 0.153,
+      "step": 2299
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5683574080467224,
+      "learning_rate": 1.6025649301821876e-05,
+      "loss": 0.2343,
+      "step": 2300
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.27942177653312683,
+      "learning_rate": 1.5866644351076876e-05,
+      "loss": 0.0742,
+      "step": 2301
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8072366714477539,
+      "learning_rate": 1.5708419435684463e-05,
+      "loss": 0.4447,
+      "step": 2302
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7639555335044861,
+      "learning_rate": 1.5550974810576323e-05,
+      "loss": 0.2481,
+      "step": 2303
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6173632144927979,
+      "learning_rate": 1.5394310729427264e-05,
+      "loss": 0.1728,
+      "step": 2304
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.32854118943214417,
+      "learning_rate": 1.5238427444654367e-05,
+      "loss": 0.221,
+      "step": 2305
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.33887967467308044,
+      "learning_rate": 1.5083325207416565e-05,
+      "loss": 0.1707,
+      "step": 2306
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.2839271128177643,
+      "learning_rate": 1.4929004267614622e-05,
+      "loss": 0.0916,
+      "step": 2307
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.753061056137085,
+      "learning_rate": 1.4775464873890254e-05,
+      "loss": 0.2666,
+      "step": 2308
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.33757948875427246,
+      "learning_rate": 1.4622707273625968e-05,
+      "loss": 0.0579,
+      "step": 2309
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4285932183265686,
+      "learning_rate": 1.4470731712944884e-05,
+      "loss": 0.1669,
+      "step": 2310
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3342825472354889,
+      "learning_rate": 1.4319538436709745e-05,
+      "loss": 0.1877,
+      "step": 2311
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5861527323722839,
+      "learning_rate": 1.4169127688523186e-05,
+      "loss": 0.1866,
+      "step": 2312
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.49546265602111816,
+      "learning_rate": 1.4019499710726914e-05,
+      "loss": 0.1389,
+      "step": 2313
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6112160086631775,
+      "learning_rate": 1.3870654744401357e-05,
+      "loss": 0.2626,
+      "step": 2314
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5993901491165161,
+      "learning_rate": 1.372259302936546e-05,
+      "loss": 0.137,
+      "step": 2315
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7596952319145203,
+      "learning_rate": 1.3575314804176175e-05,
+      "loss": 0.2502,
+      "step": 2316
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.38936638832092285,
+      "learning_rate": 1.3428820306128075e-05,
+      "loss": 0.127,
+      "step": 2317
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6624545454978943,
+      "learning_rate": 1.3283109771252966e-05,
+      "loss": 0.1938,
+      "step": 2318
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.58150714635849,
+      "learning_rate": 1.3138183434319661e-05,
+      "loss": 0.1757,
+      "step": 2319
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7801070809364319,
+      "learning_rate": 1.2994041528833267e-05,
+      "loss": 0.3382,
+      "step": 2320
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5321682691574097,
+      "learning_rate": 1.285068428703523e-05,
+      "loss": 0.1785,
+      "step": 2321
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3542149066925049,
+      "learning_rate": 1.2708111939902568e-05,
+      "loss": 0.1054,
+      "step": 2322
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5901962518692017,
+      "learning_rate": 1.2566324717147803e-05,
+      "loss": 0.2445,
+      "step": 2323
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6147281527519226,
+      "learning_rate": 1.2425322847218367e-05,
+      "loss": 0.1645,
+      "step": 2324
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7888690829277039,
+      "learning_rate": 1.2285106557296478e-05,
+      "loss": 0.2833,
+      "step": 2325
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.40575289726257324,
+      "learning_rate": 1.2145676073298473e-05,
+      "loss": 0.1676,
+      "step": 2326
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.39446914196014404,
+      "learning_rate": 1.2007031619874654e-05,
+      "loss": 0.1422,
+      "step": 2327
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6484162211418152,
+      "learning_rate": 1.1869173420408885e-05,
+      "loss": 0.1607,
+      "step": 2328
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.428232878446579,
+      "learning_rate": 1.173210169701816e-05,
+      "loss": 0.1006,
+      "step": 2329
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4206710457801819,
+      "learning_rate": 1.1595816670552429e-05,
+      "loss": 0.1357,
+      "step": 2330
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.40357252955436707,
+      "learning_rate": 1.1460318560593986e-05,
+      "loss": 0.2266,
+      "step": 2331
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4601038098335266,
+      "learning_rate": 1.1325607585457365e-05,
+      "loss": 0.1449,
+      "step": 2332
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4225379228591919,
+      "learning_rate": 1.1191683962188725e-05,
+      "loss": 0.1328,
+      "step": 2333
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5930483937263489,
+      "learning_rate": 1.1058547906565741e-05,
+      "loss": 0.1819,
+      "step": 2334
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5578705668449402,
+      "learning_rate": 1.0926199633097156e-05,
+      "loss": 0.2003,
+      "step": 2335
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.0705053806304932,
+      "learning_rate": 1.0794639355022506e-05,
+      "loss": 0.4297,
+      "step": 2336
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4729509651660919,
+      "learning_rate": 1.0663867284311457e-05,
+      "loss": 0.1532,
+      "step": 2337
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6401896476745605,
+      "learning_rate": 1.0533883631663966e-05,
+      "loss": 0.1334,
+      "step": 2338
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5156150460243225,
+      "learning_rate": 1.0404688606509617e-05,
+      "loss": 0.2067,
+      "step": 2339
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4541298747062683,
+      "learning_rate": 1.0276282417007399e-05,
+      "loss": 0.2222,
+      "step": 2340
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7148359417915344,
+      "learning_rate": 1.0148665270045209e-05,
+      "loss": 0.1902,
+      "step": 2341
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.9044702053070068,
+      "learning_rate": 1.0021837371239739e-05,
+      "loss": 0.2728,
+      "step": 2342
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.802169919013977,
+      "learning_rate": 9.895798924936028e-06,
+      "loss": 0.1493,
+      "step": 2343
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5715659856796265,
+      "learning_rate": 9.770550134207135e-06,
+      "loss": 0.1511,
+      "step": 2344
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6106379628181458,
+      "learning_rate": 9.646091200853802e-06,
+      "loss": 0.2314,
+      "step": 2345
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3333510756492615,
+      "learning_rate": 9.522422325404235e-06,
+      "loss": 0.1453,
+      "step": 2346
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7438256740570068,
+      "learning_rate": 9.3995437071136e-06,
+      "loss": 0.2547,
+      "step": 2347
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6261045932769775,
+      "learning_rate": 9.277455543963809e-06,
+      "loss": 0.1194,
+      "step": 2348
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.38642680644989014,
+      "learning_rate": 9.156158032663397e-06,
+      "loss": 0.174,
+      "step": 2349
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7166641354560852,
+      "learning_rate": 9.035651368646646e-06,
+      "loss": 0.2225,
+      "step": 2350
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.41237127780914307,
+      "learning_rate": 8.915935746073967e-06,
+      "loss": 0.1258,
+      "step": 2351
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.39462098479270935,
+      "learning_rate": 8.797011357830953e-06,
+      "loss": 0.1271,
+      "step": 2352
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3374442458152771,
+      "learning_rate": 8.678878395528666e-06,
+      "loss": 0.0932,
+      "step": 2353
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.41088926792144775,
+      "learning_rate": 8.561537049502688e-06,
+      "loss": 0.1186,
+      "step": 2354
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.36306336522102356,
+      "learning_rate": 8.444987508813451e-06,
+      "loss": 0.0945,
+      "step": 2355
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.46291017532348633,
+      "learning_rate": 8.329229961245354e-06,
+      "loss": 0.229,
+      "step": 2356
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.44555771350860596,
+      "learning_rate": 8.214264593307098e-06,
+      "loss": 0.2109,
+      "step": 2357
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3475627601146698,
+      "learning_rate": 8.100091590230619e-06,
+      "loss": 0.1154,
+      "step": 2358
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8154802322387695,
+      "learning_rate": 7.986711135971491e-06,
+      "loss": 0.2949,
+      "step": 2359
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.42342695593833923,
+      "learning_rate": 7.874123413208145e-06,
+      "loss": 0.1155,
+      "step": 2360
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.32322409749031067,
+      "learning_rate": 7.762328603341973e-06,
+      "loss": 0.1139,
+      "step": 2361
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6089414358139038,
+      "learning_rate": 7.651326886496612e-06,
+      "loss": 0.2278,
+      "step": 2362
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8325931429862976,
+      "learning_rate": 7.541118441517947e-06,
+      "loss": 0.1786,
+      "step": 2363
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6140848398208618,
+      "learning_rate": 7.4317034459737694e-06,
+      "loss": 0.2853,
+      "step": 2364
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3261142373085022,
+      "learning_rate": 7.323082076153509e-06,
+      "loss": 0.1329,
+      "step": 2365
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5719247460365295,
+      "learning_rate": 7.215254507067781e-06,
+      "loss": 0.2611,
+      "step": 2366
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5409030318260193,
+      "learning_rate": 7.1082209124482825e-06,
+      "loss": 0.1242,
+      "step": 2367
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.40979480743408203,
+      "learning_rate": 7.001981464747565e-06,
+      "loss": 0.1111,
+      "step": 2368
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4101642370223999,
+      "learning_rate": 6.896536335138426e-06,
+      "loss": 0.1219,
+      "step": 2369
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8164629936218262,
+      "learning_rate": 6.791885693514133e-06,
+      "loss": 0.2295,
+      "step": 2370
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3487977981567383,
+      "learning_rate": 6.688029708487586e-06,
+      "loss": 0.1394,
+      "step": 2371
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5112009644508362,
+      "learning_rate": 6.584968547391656e-06,
+      "loss": 0.1904,
+      "step": 2372
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.44279658794403076,
+      "learning_rate": 6.4827023762782375e-06,
+      "loss": 0.1473,
+      "step": 2373
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.963850736618042,
+      "learning_rate": 6.381231359918638e-06,
+      "loss": 0.1812,
+      "step": 2374
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3383929133415222,
+      "learning_rate": 6.2805556618028556e-06,
+      "loss": 0.0942,
+      "step": 2375
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8859298229217529,
+      "learning_rate": 6.180675444139528e-06,
+      "loss": 0.3527,
+      "step": 2376
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5733866691589355,
+      "learning_rate": 6.081590867855535e-06,
+      "loss": 0.3098,
+      "step": 2377
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7813094854354858,
+      "learning_rate": 5.983302092595955e-06,
+      "loss": 0.2109,
+      "step": 2378
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.0429630279541016,
+      "learning_rate": 5.885809276723608e-06,
+      "loss": 0.2854,
+      "step": 2379
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.40367573499679565,
+      "learning_rate": 5.789112577318789e-06,
+      "loss": 0.1157,
+      "step": 2380
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.41928422451019287,
+      "learning_rate": 5.693212150179205e-06,
+      "loss": 0.1522,
+      "step": 2381
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4992380440235138,
+      "learning_rate": 5.598108149819536e-06,
+      "loss": 0.1022,
+      "step": 2382
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5325288772583008,
+      "learning_rate": 5.503800729471376e-06,
+      "loss": 0.2389,
+      "step": 2383
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5733557939529419,
+      "learning_rate": 5.410290041082622e-06,
+      "loss": 0.199,
+      "step": 2384
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6977936625480652,
+      "learning_rate": 5.317576235317756e-06,
+      "loss": 0.2491,
+      "step": 2385
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5766323804855347,
+      "learning_rate": 5.225659461557175e-06,
+      "loss": 0.1344,
+      "step": 2386
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.46796679496765137,
+      "learning_rate": 5.134539867897081e-06,
+      "loss": 0.1577,
+      "step": 2387
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.2981155514717102,
+      "learning_rate": 5.04421760114937e-06,
+      "loss": 0.0962,
+      "step": 2388
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.44071149826049805,
+      "learning_rate": 4.954692806841188e-06,
+      "loss": 0.1026,
+      "step": 2389
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5823246836662292,
+      "learning_rate": 4.865965629214819e-06,
+      "loss": 0.2667,
+      "step": 2390
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5454072952270508,
+      "learning_rate": 4.778036211227466e-06,
+      "loss": 0.1883,
+      "step": 2391
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4172871708869934,
+      "learning_rate": 4.6909046945509125e-06,
+      "loss": 0.2257,
+      "step": 2392
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3752857446670532,
+      "learning_rate": 4.604571219571474e-06,
+      "loss": 0.1187,
+      "step": 2393
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.667158842086792,
+      "learning_rate": 4.519035925389492e-06,
+      "loss": 0.1834,
+      "step": 2394
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5379573106765747,
+      "learning_rate": 4.434298949819449e-06,
+      "loss": 0.1988,
+      "step": 2395
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.33744725584983826,
+      "learning_rate": 4.350360429389411e-06,
+      "loss": 0.1138,
+      "step": 2396
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.610379695892334,
+      "learning_rate": 4.267220499341195e-06,
+      "loss": 0.1548,
+      "step": 2397
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6929248571395874,
+      "learning_rate": 4.184879293629707e-06,
+      "loss": 0.2944,
+      "step": 2398
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4586219787597656,
+      "learning_rate": 4.103336944923153e-06,
+      "loss": 0.1222,
+      "step": 2399
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6248422861099243,
+      "learning_rate": 4.02259358460233e-06,
+      "loss": 0.1373,
+      "step": 2400
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4469952881336212,
+      "learning_rate": 3.942649342761117e-06,
+      "loss": 0.1707,
+      "step": 2401
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5199698209762573,
+      "learning_rate": 3.863504348205426e-06,
+      "loss": 0.1622,
+      "step": 2402
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4371327757835388,
+      "learning_rate": 3.7851587284537526e-06,
+      "loss": 0.1913,
+      "step": 2403
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.2656461000442505,
+      "learning_rate": 3.707612609736399e-06,
+      "loss": 0.0633,
+      "step": 2404
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7166005969047546,
+      "learning_rate": 3.630866116995757e-06,
+      "loss": 0.1747,
+      "step": 2405
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6910379528999329,
+      "learning_rate": 3.5549193738856343e-06,
+      "loss": 0.2199,
+      "step": 2406
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.5567171573638916,
+      "learning_rate": 3.4797725027713723e-06,
+      "loss": 0.3095,
+      "step": 2407
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5317336916923523,
+      "learning_rate": 3.4054256247296188e-06,
+      "loss": 0.1363,
+      "step": 2408
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.48229438066482544,
+      "learning_rate": 3.3318788595479434e-06,
+      "loss": 0.158,
+      "step": 2409
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3785305619239807,
+      "learning_rate": 3.2591323257248896e-06,
+      "loss": 0.1057,
+      "step": 2410
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4483920633792877,
+      "learning_rate": 3.1871861404696445e-06,
+      "loss": 0.1992,
+      "step": 2411
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4894324839115143,
+      "learning_rate": 3.1160404197018156e-06,
+      "loss": 0.1317,
+      "step": 2412
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.840140163898468,
+      "learning_rate": 3.0456952780513746e-06,
+      "loss": 0.4115,
+      "step": 2413
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.24554389715194702,
+      "learning_rate": 2.976150828858326e-06,
+      "loss": 0.0586,
+      "step": 2414
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5406511425971985,
+      "learning_rate": 2.9074071841727055e-06,
+      "loss": 0.1746,
+      "step": 2415
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4391881227493286,
+      "learning_rate": 2.8394644547541373e-06,
+      "loss": 0.1778,
+      "step": 2416
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.39811643958091736,
+      "learning_rate": 2.7723227500719985e-06,
+      "loss": 0.1229,
+      "step": 2417
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6342200040817261,
+      "learning_rate": 2.705982178304922e-06,
+      "loss": 0.2461,
+      "step": 2418
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6434400677680969,
+      "learning_rate": 2.640442846340796e-06,
+      "loss": 0.3197,
+      "step": 2419
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.358191579580307,
+      "learning_rate": 2.5757048597765396e-06,
+      "loss": 0.2248,
+      "step": 2420
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.374965101480484,
+      "learning_rate": 2.51176832291794e-06,
+      "loss": 0.183,
+      "step": 2421
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.714931309223175,
+      "learning_rate": 2.448633338779593e-06,
+      "loss": 0.2366,
+      "step": 2422
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4851800501346588,
+      "learning_rate": 2.386300009084408e-06,
+      "loss": 0.1991,
+      "step": 2423
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.49642789363861084,
+      "learning_rate": 2.3247684342639353e-06,
+      "loss": 0.1683,
+      "step": 2424
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.414558082818985,
+      "learning_rate": 2.2640387134577057e-06,
+      "loss": 0.1265,
+      "step": 2425
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.3093879222869873,
+      "learning_rate": 2.204110944513449e-06,
+      "loss": 0.1367,
+      "step": 2426
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3708260953426361,
+      "learning_rate": 2.144985223986817e-06,
+      "loss": 0.0874,
+      "step": 2427
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.518834114074707,
+      "learning_rate": 2.0866616471409973e-06,
+      "loss": 0.284,
+      "step": 2428
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4590609669685364,
+      "learning_rate": 2.029140307946986e-06,
+      "loss": 0.0934,
+      "step": 2429
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4364655911922455,
+      "learning_rate": 1.9724212990830937e-06,
+      "loss": 0.1725,
+      "step": 2430
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4944010078907013,
+      "learning_rate": 1.9165047119349965e-06,
+      "loss": 0.2614,
+      "step": 2431
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.373772531747818,
+      "learning_rate": 1.8613906365954614e-06,
+      "loss": 0.1082,
+      "step": 2432
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3929751217365265,
+      "learning_rate": 1.8070791618641779e-06,
+      "loss": 0.2013,
+      "step": 2433
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6776648163795471,
+      "learning_rate": 1.753570375247815e-06,
+      "loss": 0.201,
+      "step": 2434
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3880409896373749,
+      "learning_rate": 1.7008643629596864e-06,
+      "loss": 0.1275,
+      "step": 2435
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.0755947828292847,
+      "learning_rate": 1.6489612099197526e-06,
+      "loss": 0.4629,
+      "step": 2436
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3853505849838257,
+      "learning_rate": 1.5978609997542304e-06,
+      "loss": 0.1067,
+      "step": 2437
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3436007797718048,
+      "learning_rate": 1.5475638147957604e-06,
+      "loss": 0.1003,
+      "step": 2438
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5678531527519226,
+      "learning_rate": 1.4980697360831852e-06,
+      "loss": 0.2014,
+      "step": 2439
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8699470162391663,
+      "learning_rate": 1.4493788433612708e-06,
+      "loss": 0.2102,
+      "step": 2440
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5635812282562256,
+      "learning_rate": 1.4014912150808745e-06,
+      "loss": 0.1321,
+      "step": 2441
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4257364273071289,
+      "learning_rate": 1.354406928398333e-06,
+      "loss": 0.211,
+      "step": 2442
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4384666979312897,
+      "learning_rate": 1.308126059176018e-06,
+      "loss": 0.1086,
+      "step": 2443
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6304940581321716,
+      "learning_rate": 1.2626486819814487e-06,
+      "loss": 0.2481,
+      "step": 2444
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3732280731201172,
+      "learning_rate": 1.2179748700879012e-06,
+      "loss": 0.1213,
+      "step": 2445
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7894262075424194,
+      "learning_rate": 1.1741046954736878e-06,
+      "loss": 0.1665,
+      "step": 2446
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.38154077529907227,
+      "learning_rate": 1.131038228822434e-06,
+      "loss": 0.174,
+      "step": 2447
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5728100538253784,
+      "learning_rate": 1.0887755395228016e-06,
+      "loss": 0.1732,
+      "step": 2448
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7279780507087708,
+      "learning_rate": 1.047316695668432e-06,
+      "loss": 0.2935,
+      "step": 2449
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.48812609910964966,
+      "learning_rate": 1.006661764057837e-06,
+      "loss": 0.2119,
+      "step": 2450
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.33381742238998413,
+      "learning_rate": 9.668108101940631e-07,
+      "loss": 0.1395,
+      "step": 2451
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5023744106292725,
+      "learning_rate": 9.277638982850834e-07,
+      "loss": 0.198,
+      "step": 2452
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3011619448661804,
+      "learning_rate": 8.895210912431839e-07,
+      "loss": 0.0884,
+      "step": 2453
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5444861054420471,
+      "learning_rate": 8.520824506851876e-07,
+      "loss": 0.291,
+      "step": 2454
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.37627458572387695,
+      "learning_rate": 8.15448036932176e-07,
+      "loss": 0.1373,
+      "step": 2455
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.37127017974853516,
+      "learning_rate": 7.796179090094891e-07,
+      "loss": 0.0967,
+      "step": 2456
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.49025216698646545,
+      "learning_rate": 7.445921246466702e-07,
+      "loss": 0.2074,
+      "step": 2457
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6639369130134583,
+      "learning_rate": 7.103707402771886e-07,
+      "loss": 0.2283,
+      "step": 2458
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.33157289028167725,
+      "learning_rate": 6.769538110384943e-07,
+      "loss": 0.1517,
+      "step": 2459
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5349112749099731,
+      "learning_rate": 6.443413907720186e-07,
+      "loss": 0.2761,
+      "step": 2460
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4786204695701599,
+      "learning_rate": 6.125335320227299e-07,
+      "loss": 0.1533,
+      "step": 2461
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6830154657363892,
+      "learning_rate": 5.815302860395777e-07,
+      "loss": 0.2346,
+      "step": 2462
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5201008319854736,
+      "learning_rate": 5.513317027748821e-07,
+      "loss": 0.1663,
+      "step": 2463
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7330418229103088,
+      "learning_rate": 5.219378308845557e-07,
+      "loss": 0.3421,
+      "step": 2464
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.40466931462287903,
+      "learning_rate": 4.933487177280482e-07,
+      "loss": 0.0992,
+      "step": 2465
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3118881285190582,
+      "learning_rate": 4.655644093681244e-07,
+      "loss": 0.1092,
+      "step": 2466
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4302273094654083,
+      "learning_rate": 4.3858495057080837e-07,
+      "loss": 0.0732,
+      "step": 2467
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.45233699679374695,
+      "learning_rate": 4.124103848054395e-07,
+      "loss": 0.2131,
+      "step": 2468
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3744388818740845,
+      "learning_rate": 3.8704075424439435e-07,
+      "loss": 0.1228,
+      "step": 2469
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.28190115094184875,
+      "learning_rate": 3.6247609976319816e-07,
+      "loss": 0.1065,
+      "step": 2470
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8877017498016357,
+      "learning_rate": 3.3871646094052466e-07,
+      "loss": 0.2447,
+      "step": 2471
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5464051365852356,
+      "learning_rate": 3.1576187605775186e-07,
+      "loss": 0.1845,
+      "step": 2472
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.31595584750175476,
+      "learning_rate": 2.9361238209935083e-07,
+      "loss": 0.1672,
+      "step": 2473
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.48857152462005615,
+      "learning_rate": 2.722680147525525e-07,
+      "loss": 0.181,
+      "step": 2474
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.48503577709198,
+      "learning_rate": 2.517288084074587e-07,
+      "loss": 0.2191,
+      "step": 2475
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7234102487564087,
+      "learning_rate": 2.3199479615670926e-07,
+      "loss": 0.2045,
+      "step": 2476
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.37276914715766907,
+      "learning_rate": 2.1306600979581481e-07,
+      "loss": 0.1307,
+      "step": 2477
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5118582248687744,
+      "learning_rate": 1.9494247982282386e-07,
+      "loss": 0.1446,
+      "step": 2478
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6104289293289185,
+      "learning_rate": 1.7762423543832286e-07,
+      "loss": 0.1743,
+      "step": 2479
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6961856484413147,
+      "learning_rate": 1.61111304545436e-07,
+      "loss": 0.1857,
+      "step": 2480
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8121128678321838,
+      "learning_rate": 1.4540371374988093e-07,
+      "loss": 0.22,
+      "step": 2481
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.45882028341293335,
+      "learning_rate": 1.3050148835958008e-07,
+      "loss": 0.1321,
+      "step": 2482
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.638444185256958,
+      "learning_rate": 1.1640465238516029e-07,
+      "loss": 0.1822,
+      "step": 2483
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5327040553092957,
+      "learning_rate": 1.0311322853928662e-07,
+      "loss": 0.1234,
+      "step": 2484
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.42821553349494934,
+      "learning_rate": 9.06272382371065e-08,
+      "loss": 0.4178,
+      "step": 2485
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5191227197647095,
+      "learning_rate": 7.89467015961387e-08,
+      "loss": 0.231,
+      "step": 2486
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.949320912361145,
+      "learning_rate": 6.807163743594025e-08,
+      "loss": 0.236,
+      "step": 2487
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.9863948225975037,
+      "learning_rate": 5.800206327855051e-08,
+      "loss": 0.2471,
+      "step": 2488
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5361063480377197,
+      "learning_rate": 4.873799534788059e-08,
+      "loss": 0.3381,
+      "step": 2489
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5105975270271301,
+      "learning_rate": 4.027944857032395e-08,
+      "loss": 0.1752,
+      "step": 2490
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.8464347124099731,
+      "learning_rate": 3.262643657425679e-08,
+      "loss": 0.2586,
+      "step": 2491
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7352217435836792,
+      "learning_rate": 2.57789716902046e-08,
+      "loss": 0.2027,
+      "step": 2492
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.678185224533081,
+      "learning_rate": 1.9737064950786642e-08,
+      "loss": 0.2029,
+      "step": 2493
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4828554391860962,
+      "learning_rate": 1.4500726090715954e-08,
+      "loss": 0.176,
+      "step": 2494
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6859709024429321,
+      "learning_rate": 1.0069963546743833e-08,
+      "loss": 0.2641,
+      "step": 2495
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7563552856445312,
+      "learning_rate": 6.444784457770858e-09,
+      "loss": 0.1031,
+      "step": 2496
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5268996357917786,
+      "learning_rate": 3.625194664735876e-09,
+      "loss": 0.1467,
+      "step": 2497
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5631054639816284,
+      "learning_rate": 1.6111987103939462e-09,
+      "loss": 0.1588,
+      "step": 2498
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5639719367027283,
+      "learning_rate": 4.027998398714594e-10,
+      "loss": 0.1766,
+      "step": 2499
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.45474207401275635,
+      "learning_rate": 0.0,
+      "loss": 0.1373,
+      "step": 2500
+    },
+    {
+      "epoch": 1.0,
+      "step": 2500,
+      "total_flos": 0.0,
+      "train_loss": 0.35517678638547656,
+      "train_runtime": 41836.0017,
+      "train_samples_per_second": 0.478,
+      "train_steps_per_second": 0.06
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2500,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}