diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8687 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1235,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004048582995951417,
+      "grad_norm": 7.275977905097055,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.9044,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 7.4069087669556275,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.9067,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 7.413854985763196,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.921,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 7.455238009528975,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.9191,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.90024333733292,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.8783,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 6.734850283711219,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.8872,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 5.213107211910938,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.8181,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 5.054364149981319,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.8327,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.6845487620585784,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.7937,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.377247410846431,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.7594,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.1061926653329985,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.735,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 4.012655232554972,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.7397,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.612731944213642,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.7632,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 4.490419989670422,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.7381,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 4.36505515598906,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.7281,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 3.419223676234191,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.685,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 3.2190958670983356,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.6924,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.5616360819173773,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.6755,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 2.100567806427278,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.676,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.5441602727124746,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.6615,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.400319506753679,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.6327,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.8887193660519197,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.6297,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.7898727101112968,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.6324,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.220984186811487,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.5963,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 1.0397401730312463,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.5807,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.2926326283430147,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.6022,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 1.071586544318273,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.5993,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.8501104444265077,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.5658,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.9785592791636112,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.5803,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.9535141559603885,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.5804,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.774650500178834,
+      "learning_rate": 1e-05,
+      "loss": 0.56,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.7960520663059848,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.5868,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.9793286674642282,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.5673,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.6414741041284113,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.5509,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.6797625657766085,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.5721,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.8635216076323639,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.5357,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6654748111644249,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.5207,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.9151994757130825,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.5545,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.7030040002639778,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.5457,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.7044633439970485,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.5343,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.6389132878574936,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.5336,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.6687432053481495,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.5482,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.6340272442512994,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.5118,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.5563746202342167,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.5197,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.698362631815219,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.5083,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.665981305903771,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.5356,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.6501957133166935,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.5124,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.7099848269468725,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.5227,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.6135105159696573,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.4981,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.7861725127788906,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.4895,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.5725905623252423,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.4832,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.6264603543100075,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.493,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.5987440942276143,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.5273,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.6803235503468169,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.5034,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.6833329349556572,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.4959,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.7055141096547679,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.5135,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.6930118290029021,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.5135,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.8087244990120844,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.5196,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.6504992103684728,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.4834,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.8437856167212222,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.4956,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.6180170324442382,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.5098,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 1.0126615549416882,
+      "learning_rate": 2e-05,
+      "loss": 0.4956,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.720685050916055,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.5071,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.7676172391998656,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.4809,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.5974293149587432,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.4776,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.8036328777309153,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.4919,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.5803961640525475,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.5127,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.8236600346577386,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.5062,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.6172584588089784,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.4964,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.8601944053731871,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.5034,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.5634820467292033,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.4978,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.9561191336336693,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.5117,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.7847460992431499,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.5091,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.848645567663745,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.4828,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.8353196514085017,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.4959,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.6061226664683573,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.487,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.7865076219591718,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.514,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.8892603786999965,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.5175,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.921229034064134,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.4721,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.9812079355311403,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.5026,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.8857580586539445,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.4881,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.7148543976130458,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.4994,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.8069902027153574,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.4922,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 1.162602579699741,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.5011,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.7686908548418727,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.5039,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.8540815819077442,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.4671,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.8325847265584174,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.4883,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.661400982978596,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.4756,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.6808497185116672,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.4777,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.9171847535219868,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.4978,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.8796503309275424,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.5047,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.9297601490333285,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.4894,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 1.322623055144344,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.464,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.8814403208300101,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.4976,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 1.0048755069394946,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.4618,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.7536760660716029,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.507,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.7348035763070585,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.5008,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.6656188995889362,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.5048,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.793616444527924,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.4821,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 0.6761129529516502,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.471,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.7505686462385478,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.4645,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.811899507829449,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.4629,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.7372258025233704,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.4857,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.7110004506291467,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.4971,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.7903042167827937,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.4911,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 0.8283629833316797,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.4947,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 0.5722675857512747,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.466,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.8848413684574592,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.4879,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.7400393591031535,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.4848,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 0.8467056012405709,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.4876,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.8708618863042098,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.4684,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.9217048039426967,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.4772,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.1358985918238493,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.4663,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 1.0223035302042636,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.4933,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 1.0055828792890165,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.5001,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 0.9191513465872055,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.4729,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.7158968627364597,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.4857,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.6641189861182478,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.4765,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.660341473080701,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.4967,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 1.0036066928202687,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.4807,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.9556044396040129,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.4761,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.7222037318314434,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.485,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.7453125611027462,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.4906,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.85801548242796,
+      "learning_rate": 4e-05,
+      "loss": 0.4783,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.7154701104616293,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.4764,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 0.7022312959075391,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.4843,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.7464402101772355,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.4772,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.5946787287838982,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.4665,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 0.8038275372234152,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.4621,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.6051557694809152,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.4893,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.8051295171293328,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.4802,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.8761701284584422,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.5049,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 1.2611146759444987,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.4698,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.5983708407053087,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.4773,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.9911821958976875,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.4518,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.9314308126761588,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.4719,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.8463329261319257,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.487,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.6304233399589101,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.4796,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.8860823110964662,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.4712,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.8097221599299347,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.4851,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.5922046131462505,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.4699,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.6095844207880212,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.458,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.6086401536026065,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.4684,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.5719512461358531,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.4934,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.5280730222218594,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.4722,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.5642625974766016,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.466,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.573265982698988,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.4714,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.7057300980107618,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.4745,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.6958665780999796,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.4991,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.6493075289091848,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.476,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.6210660449783713,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.4743,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.6376941105083098,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.4855,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.5181295445372727,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.4432,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.6493631450515644,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.4775,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.5820374538328563,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.4745,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.6834183970599889,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.4546,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.5360173911745983,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.4751,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.7092158132226708,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.4575,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.5652026577100375,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.4619,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.6540860048895255,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.4791,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.4947581216680809,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.4711,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.675309443955505,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.4483,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.6895403676028385,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.4602,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.6542382931248798,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.4573,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.48509354083316536,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.4549,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.6776631125680258,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.4705,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.5844415088193936,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.4624,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.606362736737412,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.458,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.603862751170747,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.4695,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.6620977726059517,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.4766,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.6552938198642249,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.4847,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.7876679273654372,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.4723,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.6593408871592059,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.4723,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.7263185398785821,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.4666,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.6978115742714149,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.4704,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.7244411592352447,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.4837,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.6379688215900179,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.4585,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.7433363705530642,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.4564,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.6547684368841704,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.4671,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.6031862471292264,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.459,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.5805161956165513,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.4623,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.47353697904070924,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.4695,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.6050095481767488,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.4507,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.6873697974965376,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.4554,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.5499782431995968,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.4847,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.4808087290896966,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.4778,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.7700295377187546,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.4434,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.7703941346028077,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.468,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.6695977359235089,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.4597,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.6407308249704561,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.4711,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.8349736576533989,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.4698,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.63455294632798,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.4526,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.5670338674120905,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.4401,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.6049595806415177,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.4576,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.5764528922575268,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.4632,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.5645275725891955,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.4568,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.6083801729401134,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.4603,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.6900198767999102,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.48,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.6201004649418598,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.4465,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.642590399237489,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.4259,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.7613887441832784,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.464,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.5836706401358343,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.4587,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.7153979720290682,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.457,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.5770625234834065,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.4577,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.6184061864298169,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.4524,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.5418425982720202,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.4703,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.6422770075762518,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.4815,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.4964732932756472,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.4506,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.5611584890023693,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.4527,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.48987900298630765,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.4724,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.5254696901380811,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.4748,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.5200191865856123,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.4526,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.5601833795017078,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.4581,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.6719959435731362,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.4727,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.48169132346359544,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.4539,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.5964911700034093,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.4653,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.5820450614524169,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.4568,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.5670090274455973,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.4645,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.4769260322964696,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.4524,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.538372126154825,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.4541,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.4846827818696297,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.4581,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.5851389956761461,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.4448,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.41573958544792466,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.4504,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.5469483760347043,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.4625,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.4366040491313625,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.4404,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.49570994724403206,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.462,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.550457788227688,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.4365,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.41361255604283753,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.4482,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.527972558261362,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.4712,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.5208138309939608,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.4576,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.4559624332095123,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.4348,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.5182969850705083,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.4416,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.4574130341920026,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.4761,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.42778258504466926,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.414,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.48374540619821876,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.4435,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.4857595594016277,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.464,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.5163753998593823,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.4508,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.43691496225652665,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.4803,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.5734170954967478,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.4496,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.44641806789534555,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.4553,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.510131043522479,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.4516,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.4473589435454056,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.4678,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.5430751784902083,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.4512,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.46206671402709215,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.4446,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.4810648403973155,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.4351,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.4768014503863076,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.4603,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4845441058937541,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.4524,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.47033365624799234,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.3787,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.42784660796158713,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.3887,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.5482773273309751,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.411,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.6457010444136537,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.3971,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.6002873098116498,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.4121,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.5131888030132924,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.3926,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.5428517586913714,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.4022,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.48336628861040426,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.3674,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.5080760767002823,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.3845,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.5874411218210889,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.4008,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.4897000609151336,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.405,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.6408621553578083,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.4011,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.4493199388194718,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.3779,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.5955234477343943,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.3973,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.4698745556748999,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.385,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.6601166522026815,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.3802,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.5192671605789758,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.414,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.6735352194227974,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.4161,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.5163797035694302,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.3828,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.6156843872309451,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.3864,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.517637478911438,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.3887,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.48762631940181905,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.3777,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.5466855303401632,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.4082,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.4736282370809333,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.4074,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.5681688205695992,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.4369,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.4832385091585762,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.3764,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.49331275645629713,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.4052,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.45500362095576435,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.3914,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.5257779005542343,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.3955,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.4072692071017041,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.381,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.5357178783807889,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.3849,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.4492116309154521,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.3968,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.5152641468776802,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.3911,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.5259048105885105,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.415,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.4381159043916622,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.3816,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.4870123675456168,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.3966,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.512666969945794,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.3967,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.5860974452234223,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.4035,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.5015364676435564,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.3957,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.5584588276016821,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.3919,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.5823158866891774,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.3987,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.5748525623118078,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.383,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.4651577451258453,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.3814,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.5610284852929731,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.3786,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.39815687336551214,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.395,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.5197977926850802,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.3862,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.5724967442505265,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.4141,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.46488623037900795,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.376,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.5470017064707835,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.3946,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.6178519926347964,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.3864,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.5078405589909004,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.3783,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.5180113987840773,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.3916,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.5000583334398896,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.4029,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.4451683316897013,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.4024,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.46389545130623483,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.4068,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.458274404366199,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.3694,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.4709884554200486,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.4056,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.4836476848956885,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.3948,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.4223786317501595,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.3701,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.5090095805330819,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.412,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.45605208978580486,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.4036,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.512534890988147,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.4031,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.4349774311200878,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.4099,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.4342752350830698,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.3906,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.4217990736555159,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.3509,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.5016658461847243,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.4016,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.5126540400231234,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.4056,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.5217858416138768,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.4263,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.5334782220103823,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.3966,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.4815120745786221,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.3989,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.6123478712669504,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.3933,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.46187383197184056,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.3859,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.5782130385522367,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.4014,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.4116818428874057,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.401,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.5013694172159613,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.3879,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.4566705530383082,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.4055,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.5311299000390501,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.3793,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.41898842773361167,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.3961,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.42097488430550106,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.4168,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.42182726397351183,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.4087,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.40450436515815524,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.3607,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.4335794458880387,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.3674,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.493264823004969,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.3874,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.43991423167783833,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.412,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.5305939322148485,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.3867,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.3908896588661339,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.3824,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.48938343647023314,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.3961,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.4398898412291056,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.3913,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.4330554436542949,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.3916,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.4382157122827953,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.4013,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.42336162817778605,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.4051,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.3950424610015226,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.3825,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.4657060682324172,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.389,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.4727471119517503,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.3951,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.43984334018205734,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.3865,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.5017145919057676,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.3918,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.4408933289347658,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.3883,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.44802264355007626,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.3765,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.4123628123400515,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.411,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.44320761247042756,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.3965,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.3664205109990966,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.3831,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.5349508991778925,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.3995,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.4337077392407409,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.4085,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.4425756734206006,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.3871,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.5208653965861593,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.3915,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.4003502891221946,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.4116,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.4108576555975618,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.4033,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.3657054242208904,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.3957,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.39259023138158416,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.3809,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.38794486275440515,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.4029,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.3763509015125008,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.4056,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.4485531289536458,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.3962,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.39650811449124657,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.3926,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.34991161412145927,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.3775,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.474307758179994,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.3788,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.4118789691975226,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.4078,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.46936985688438276,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.384,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.39640852578938895,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.3888,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.45586074976112084,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.3969,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.4109963783244454,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.3949,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.4294679517681899,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.3838,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.38935514163208634,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.3856,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.40376736407629754,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.3869,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.48295443048657033,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.3903,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.4796989499897645,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.398,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.4310677242541532,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.3901,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.5047942857198991,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.4007,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.3940509093076501,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.3762,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.42243157533437375,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.3881,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.3901678977445397,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.3927,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.44213441729243597,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.4107,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.3678939969424658,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.3814,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.4298406948916599,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.4112,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.4417186840673196,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.3872,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.40370206913818846,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.3516,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 1.7347168827996853,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.3967,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.49196987030233275,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.4164,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.3965204825098348,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.3792,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.4472981252170048,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.3928,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.44147244216472487,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.4045,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.42663437301879525,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.3843,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.43151947082248243,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.3859,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.48461439523610084,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.4004,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.37853015854732347,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.4038,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.5672638976776629,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.4062,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.39229237123445015,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.4047,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.46902548328598287,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.384,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.39908071465297607,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.3756,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.5328111039877291,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.3744,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.4621657183143115,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.4107,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.47327306076253867,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.4006,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.5440139635234672,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.3892,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.4373768204947701,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.3843,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.4160451040051793,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.3924,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.4785566352587425,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.3744,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.4083395790077776,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.4006,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.44434569293968695,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.3694,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.4292381960772523,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.4059,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.47692831576893635,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.3778,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.47384964825555315,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.391,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.41391825280858263,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.3917,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.5102410662760768,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.3945,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.38648037903976645,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.3945,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.4513499868044546,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.4234,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.3790520775634508,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.3989,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.4809085245533959,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.3784,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.3762800339617832,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.3877,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.4404814270999145,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.4253,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.39090928841467504,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.4013,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.409802938046921,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.3953,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.3882596952352414,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.3861,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.41080888564265083,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.3989,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.37563526400327835,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.3662,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.38680869298874043,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.3882,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.3983771819875863,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.386,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.34533124249662306,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.4041,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.43938526634173936,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.3935,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.3506922756230256,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.4091,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.4254458617823834,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.3957,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.3840494576754243,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.3933,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.38419578312007957,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.3876,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.41601594900939953,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.3934,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.3802307693750005,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.3912,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.4091550907292291,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.4036,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.39944449713347147,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.4181,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.3738099875866948,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.3867,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.3898625985802638,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.3862,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.3294359162409121,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.3783,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.4249046909301452,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.403,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.3805599174322321,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.3953,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.4625917801757574,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.4051,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.39749478348566447,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.3715,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.42523291413902525,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.3791,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.42043517844854283,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.3867,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.416673479066539,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.4005,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.40096617114804806,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.3919,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.38607248932754284,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.3765,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.4190907086426814,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.397,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.41243856122216505,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.4084,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.3787134506988698,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.3878,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.41163720909070345,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.3866,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.40023389915742796,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.3935,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.4611102875157549,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.3937,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.4221252787896728,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.3784,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.4195103498054905,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.3934,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.4378438795109396,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.4076,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.4360912891345114,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.3849,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.4411916069324773,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.4004,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.4797176208212669,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.4026,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.5189760194424248,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.4016,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.42708755147264993,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.4108,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.5298709527539708,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.3896,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.3996264800207,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.3949,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.45830122207497703,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.3922,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.46239861020108675,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.4159,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.39192653581618697,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.3951,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.5126161403321441,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.4069,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.3782904147261098,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.4046,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.4663684830029706,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.379,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.3876980070399386,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.4086,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.43476009554466827,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.3901,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.4529582125234779,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.3953,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.42480789930745705,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.3622,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.45209069305941596,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.3792,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.4531675055187749,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.3856,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.4856709692404634,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.3733,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.4549257392675174,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.384,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.4006400312985046,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.3908,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.4564517703113749,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.3929,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.42453536620451343,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.4116,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.427053380181687,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.3865,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.4118109413965145,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.4,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.3778073124167814,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.3807,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.39333936818159365,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.3942,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.47387494938512625,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.3817,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.39777260529190944,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.4191,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.4678900244823684,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.3784,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.42292163843916236,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.3948,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.4769154023779224,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.422,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.3848130923820043,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.4045,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.44511090867696745,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.3714,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.4166675353845981,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.4103,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.46355210184080275,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.3963,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.43755213694407163,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.3805,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.4095029408283316,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.3892,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.4268879023610746,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.398,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.4056911600222893,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.3736,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.5117706793187681,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.317,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.7009215924219138,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.3212,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.5444842002927579,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.2975,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.4770060235183278,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.303,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.47545538904020884,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.2993,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.4241166238805624,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.2893,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.4709413696058917,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.3077,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.4386460469209481,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.3203,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.4173879488335685,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.3103,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.4566195453259995,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.2993,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.4111563293413763,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.3047,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.4266261800453164,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.3075,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.3939760366182995,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.3101,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.42616135964343715,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.3079,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.4116939630620017,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.2997,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.388645312838121,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.2999,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.4574410237439001,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.3132,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.3581827145570827,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.3173,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.4266326093821135,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.3117,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.38639756402811687,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.2804,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.3605963091068024,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.3077,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.4246067789558628,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.304,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.33645799130729703,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.3158,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.46018213783226886,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.3284,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.35251144011752295,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.3209,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.39030535302972097,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.2902,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.3804130220551314,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.3169,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.37555177220246455,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.3065,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.3946718050988652,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.3199,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.33744497381339134,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.2952,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.37198342314538596,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.3006,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.3825918573113383,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.3096,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.37405175103606964,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.292,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.4133602395812363,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.271,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.3653623037741298,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.2996,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.4202368110246856,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.3023,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.4619314445940727,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.3126,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.42388427373731474,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.3115,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.5079606116505753,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.3103,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.42737694869540416,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.3054,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.516561643967645,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.316,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.45846546528447324,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.2977,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.4368219071707384,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.3046,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.46672221896374244,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.3128,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.4640965018365767,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.3151,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.4738670810551601,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.3037,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.4925726572408984,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.3094,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.45993746880985453,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.3097,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.48919466658444344,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.2964,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.40587050764133015,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.3032,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.5127902505000423,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.3214,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.39671971760655483,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.3004,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.48715002186924705,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.314,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.48623061744189905,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.2987,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.41252086769215446,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.3174,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.48057403152440803,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.3056,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.4677256407226484,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.3057,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.43064145634276635,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.3112,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.493859780603153,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.3097,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.5217074723438467,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.3174,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.4795918237334772,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.3147,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.55411370691473,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.3085,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.39863487202277803,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.3046,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.530342655881451,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.3159,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.3628531091228931,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.3061,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.4291808065765857,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.3041,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.4029934902154258,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.3314,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.39105284623943437,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.306,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.4391611465794769,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.3046,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.4390331397170878,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.3205,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.3806690477255541,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.3228,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.4002589764631278,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.3111,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.34386255012620714,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.2933,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.41395963617286413,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.3322,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.3554036383818479,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.3013,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.3953924771166274,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.2942,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.3705577613775223,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.2892,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.3861204501176122,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.3018,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.39903852407709633,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.2944,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.4040943274148699,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.3024,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.43150778489102315,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.3092,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.36844651598364075,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.2968,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.36180931407404954,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.3024,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.3776589279215786,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.3102,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.3571384123030856,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.308,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.3902857844756805,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.3024,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.39290805595247985,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.3293,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.40593553411624406,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.3253,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.39212239647099856,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.2894,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.34130288061847947,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.321,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.38495381560304465,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.318,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.38403040976480485,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.3124,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.3670408040137574,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.3009,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.36750878126704106,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.3023,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.3210555040760778,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.319,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.3481657777043042,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.319,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.3487722399912304,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.3242,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3326629584803632,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.3012,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.3768640505138563,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.2986,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.3549753150586871,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.3543,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.35251582611372234,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.3266,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.3541830109139311,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.3242,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.3658359126794346,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.2987,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.3718283355753688,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.3244,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.3944074923560604,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.2965,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.37475664481004733,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.3144,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.3364415986422043,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.3167,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.37443617528502415,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.3107,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.3897678096406138,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.3017,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.3635955967235287,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.3145,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.3908734310177627,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.3046,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.37253728967191185,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.3048,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.3665273986156465,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.3111,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.4349994915370694,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.3192,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.33766625856379356,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.3286,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.34584119180215167,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.3005,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.38682533046115675,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.3122,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.3630831156129833,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.3194,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.3850572496209002,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.3016,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.3695157942175873,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.3018,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.4126778264162705,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.2975,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.3611046668736723,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.3132,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.42760557392593634,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.3094,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.31476229414251095,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.3048,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.3722090007492709,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.3057,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.34255260432676277,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.2965,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.3466791889894555,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.3182,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.3488662161451842,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.3144,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.31573303874031733,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.3051,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.3199709635228436,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.303,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.35679461286209835,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.2981,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.3723556293910416,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.3192,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.36606489479858406,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.3191,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.38885349723622303,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.3065,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.3129888483526819,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.3071,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.3698636218708573,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.3021,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.41448600503831035,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.3112,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3547967669475575,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.3192,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.3841501365579278,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.3191,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.3336058733603756,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.312,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.35667060951978685,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.3086,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.33061881850738084,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.3226,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.32015992686307815,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.3225,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.3438551354748823,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.2995,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.32367209491456495,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.3048,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.3365701405230737,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.2963,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.3621225329511636,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.3155,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.3644383799492137,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.303,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.3487422555057057,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.2887,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.41554877407665697,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.3112,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.3448303238700485,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.3289,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.3772242593345138,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.3198,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.34937452989095286,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.3145,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.37223504241035393,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.2948,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.38448804583699764,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.3339,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.3366366811874251,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.3143,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.35922990553425166,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.3103,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.3377262201372578,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.3165,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.3391970939250087,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.3013,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.35003943417855243,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.3134,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.3788705669782935,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.3004,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.34842113776912087,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.308,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.3965714382978196,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.3183,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.3267015194795653,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.294,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.372641250405026,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.3134,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.3699120989631689,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.3208,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.36247280998007103,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.3132,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.31520192908265204,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.3133,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.38809934741291396,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.3247,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.35355023009981545,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.3149,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.3754854336655211,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.3088,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.3776412496366882,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.3135,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.3421778849574661,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.3116,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.34652971448145864,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.3141,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.34986899831200285,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.3187,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.36333058442590954,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.3011,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.3444808556775036,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.319,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.35205014357580733,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.3181,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.4375312838002039,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.3214,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.3248515211840277,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.3165,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.3441365770294008,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.315,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.3194170105892405,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.3203,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.35048753958077394,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.2948,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.3621729233099919,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.328,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.3348122721797593,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.3023,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.37498409669068367,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.3069,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.3527954037003584,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.303,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.3831199859810749,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.3178,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.3916776914439734,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.3127,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.36989773375826646,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.3083,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.33514119700938666,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.2983,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.33048000078103523,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.2964,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.32259503200814504,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.3024,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.33142275166278573,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.3079,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.31036138352383974,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.3081,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.29638612971149936,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.3037,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.3677325644773342,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.2925,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.3215510974339207,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.3187,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.3072622082460064,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.298,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.3547414056024784,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.3003,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.33708934058172246,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.3082,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.32098542390686435,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.3023,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.3623344674373611,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.2942,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.33980462655943894,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.3128,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.35910626461774514,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.2976,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.3286701662160132,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.3063,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.3583341302398548,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.2864,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.3444204485158368,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.2795,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.3934187343448578,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.3017,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.3399001212051576,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.3194,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.3773448960499891,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.2886,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.32046140620512437,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.2956,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.35289057955265457,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.2974,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.3156621209787073,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.3215,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.33958863862519645,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.2995,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.3226703977840549,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.3095,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.33417199008868337,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.3149,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.3284166890216407,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.3006,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.32736569782757713,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.3031,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.30449840579501897,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.3132,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.3214985316109573,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.3178,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.3182583983090685,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.321,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.31884309187955817,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.3192,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.3162632832724032,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.3038,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.3295641883440261,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.2673,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.3424546052679951,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.3016,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.3444968029521446,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.3029,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.34284707202883374,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.3092,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.3498889711381514,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.3063,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.33590552238112387,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.2921,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.34377160213562186,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.3067,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.3283170582817543,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.2757,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.3276827532943503,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.31,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.349360224968901,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.3075,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.3096786144418433,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.3281,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.3394305790258687,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.2945,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.31561283194211653,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.3081,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.31333878652932246,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.3163,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.3055842599476079,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.2942,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.29641273924629263,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.2943,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.3153285423578078,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.316,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.30636326929865487,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.2846,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.3090360063058654,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.2964,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.339487930098107,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.3116,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.29943601511232515,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.303,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.32100649683507665,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.3117,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.31105843315119813,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.2979,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.5875511661668197,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2357,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.3593142364022501,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.2424,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.6794731878597904,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.2372,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.3618840435748989,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.2194,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 1.2176850873519072,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.2275,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.5090275011372124,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.2225,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.3644969608806717,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.1932,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.3951763265997531,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.2133,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.4324960807569846,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.2106,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.35937560896472925,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.2335,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.3747770632711186,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.224,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.35664157572501587,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.2253,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.3718542871302768,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.2265,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.3335949578132554,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.2348,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.3403410219484337,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.2237,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.35168845687265926,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2275,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.33990644011760746,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.2462,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.33517615847707294,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.2134,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.35164398126372376,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.2372,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.35681003969779856,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.2138,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.3210808476061114,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2121,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.37808638160989994,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.2162,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.3287230318437619,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.2136,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.39131657659583424,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.2045,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.3680751147636442,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.2106,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.3229481347155049,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.1886,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.3542279806914883,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.2297,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.35682506964050276,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2406,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.33329304153634415,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.2234,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.3220981628855482,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.2365,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.34549913442735736,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.217,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.3179328462350928,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.2315,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.3184834152300724,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.213,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.3190231924268406,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.2368,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.34424487828043815,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.2238,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.30291137984991723,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.2215,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.3338132873012882,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.2258,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.34032924381539464,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.1949,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.31374361175406246,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.2107,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.314925366485589,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2252,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.3599913027177405,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.2215,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.3101212603650611,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.2153,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.33800769256671787,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.2308,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.33626421883374413,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.2273,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.332603528897986,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.224,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.31369978345457844,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.2205,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.328135955386713,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.1939,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.3471698543246054,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.2138,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.2996804143740229,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.2313,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.3647129075701921,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.2134,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.33897557042392257,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.2151,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.3597498237648801,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.2095,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.3668167452208177,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2324,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.32342764641903843,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.2083,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.33654413621676876,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.2307,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.341977637170702,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.2435,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.34689269365412384,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.2309,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.3368990491766814,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2273,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.3343925345285812,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.2114,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.3283321149191047,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2145,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.31536642845244545,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.2382,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.3161092139947328,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.2279,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.32677216370670475,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.2132,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.32881381210067584,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.2227,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.3288830125967932,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.2484,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.3158668231723258,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.2223,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.36310317281957544,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2257,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.30775465100864097,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.2306,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.3173834651476592,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.2247,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.33531801259909705,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.2172,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.3225154000389329,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2292,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.3212212331763009,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.2424,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.36125429196617836,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2534,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.338199211033486,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.2343,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.31426759441372676,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.2216,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.3382634417273579,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.2382,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.336975801008616,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2359,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.33523588368376106,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.2207,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.3261585203369108,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.2265,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.3295645248729031,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2108,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.31777729054712367,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2296,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.29179689413808824,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.2192,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.3474043826303824,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.2389,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.30921442119996334,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.2369,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.3069593874620272,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.2196,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.3192409821176439,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.2131,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.3211239822182001,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2382,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.32395923237821433,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.2186,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.3090839863007279,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.2078,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.32976358414362017,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.2035,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.3294459590372628,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.233,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.3253886616801198,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.2238,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.3292560309235938,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2144,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.3283752166741241,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.2102,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.3236840111469993,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.2214,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.3334662969092617,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.2311,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.3015610574561273,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.223,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.31309222977831164,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.2319,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.31459077421601084,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.2241,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.32183199836627907,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.223,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.3267688867931578,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.2326,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.33235618888512924,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.2213,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.3260377809333252,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.2354,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.307711660771582,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.2363,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.3086925631196487,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.229,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.30456950272729616,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.2261,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.34079705450150033,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.1973,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.3077658194398993,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.2254,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.3290913199016462,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.2083,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.3292362388786449,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.2442,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.31033414775288565,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.1959,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.3175840915133588,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.209,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.3090146416445703,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2022,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.3206710022165329,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.2271,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.3103914657481469,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.2245,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.3129146808895674,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.2247,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.2824729277010796,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.2227,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.3400096507712892,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.2173,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.29802469211503313,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.224,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.324460715165944,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.2189,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.3372108475351135,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2347,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.3020353829605262,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.2152,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.3146376330572872,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.2134,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.30904432019288947,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2221,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.3180957660096019,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.2212,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.2965510039982041,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2257,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.28470159212604124,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.2282,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.34975378829454434,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.2315,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.31486280727107374,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.2131,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.2983148764898499,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.197,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.3339188930210267,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.2265,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.3052007638989247,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.2239,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.29394516758459016,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.2372,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.3214270134043459,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2193,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.3268052590999433,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.2312,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.3106721447470057,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.219,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.29672863147759443,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2398,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.31695618143903237,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.2156,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.33631134467983,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.2219,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.32994858649859765,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.228,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.2782703447709327,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.2212,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.3007550997761748,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.2214,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.30626453576739104,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.2275,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.3019485297811183,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.2223,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.30659926401868676,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.2205,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.32705920137919536,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.2164,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.3048049944674185,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.2078,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.29549455564879407,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.2141,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.31624630418499583,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.2244,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.3435694096264519,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.2281,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.30556944307257944,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.2204,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.3119334060656432,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.2184,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.3195950950514628,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2213,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.30172474781086894,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.2248,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.2863371556235529,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.2263,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.2972720273392032,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.2223,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.3131557849253587,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.2353,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.3103434584894928,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.2274,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.3070857409454537,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.2309,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.3023540805270718,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.2141,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.30917320333900417,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.212,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.35327114487989714,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.2186,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.32879834087824966,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.2192,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.3387634777718721,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.2317,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.31329254702916515,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2139,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.3243763935062458,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.2217,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.326870814133882,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2133,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.32702916912210345,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.1979,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.3053275166035286,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.2327,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.31421266286501326,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.2119,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.33543484445083405,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2177,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.32470705671408034,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2245,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.2891136975998334,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.2208,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.29709605594157756,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.214,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.28915836630792285,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.2085,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.3222252554896928,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.2154,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.3005178912633589,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.2184,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.291062393096376,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.2143,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.31919607642054343,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2297,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.3129108780343155,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.2201,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.29177371257130497,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.2124,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.3098641813725754,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.2145,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.284693964710422,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.22,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.2952734330218009,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.2235,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.28577238782284087,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.2273,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.30694511871105373,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.2095,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.28104155461916874,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.2048,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.29112105813267836,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.2137,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.30347742555071633,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.23,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.29457205959687394,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.2236,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.3002202753775487,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.2127,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.30199496026155626,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.229,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.3154325616592229,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.2118,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.29826297436633825,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.2116,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.3040151968561937,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.203,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.2892363842001969,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.228,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.29824216878724225,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.2327,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.32358677024315685,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.214,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.32478634883313423,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.2196,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.28245601300170203,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.2296,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.30892087399529483,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.2063,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.3018440093199603,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.2152,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.3114096463254954,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.2374,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.3210261752216518,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.2038,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.29493927175171647,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.2204,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.291331303345397,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2207,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.2859251277956734,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.2005,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.30858234432472625,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.229,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.28108709231041745,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.2281,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.28829568243658993,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.2282,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.2865381688853123,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.2111,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.28207404693445065,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.213,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.29701714263577356,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.2258,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.27394426562689783,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2309,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.28233745589108755,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.2265,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.2923741357298128,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.2055,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.29093988292449485,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.2074,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.3021770185186226,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.2096,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.3049687997747111,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.2265,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.2842341808651841,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.2182,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.30424472739689923,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.203,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.30314054027916776,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.2247,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.3039067075975134,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.2108,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.29502722760247646,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.218,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.2788705623344762,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.218,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.28533619095264606,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.2143,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.30090289517223157,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.2019,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.3140992688713735,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.2056,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.29755359080496874,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.2308,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.28663909242189234,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.2207,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.29298368490754206,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.2291,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.2969814651709595,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2151,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.3147324517956136,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.1985,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.30406172938139814,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.2094,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.280900526916418,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.22,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.2949901442059806,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.2203,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.30118535507689814,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2095,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.3017269299328402,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2309,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.28913927424247865,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.2026,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.30363208298683025,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.2207,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.3040363732552506,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2293,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.2807485514848752,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.2396,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.30148367811997223,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.2029,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.28325704967155035,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2222,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.29275158467402507,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.2092,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.302234150099369,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.2231,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.28159639410760373,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.2143,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.44684568311696155,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.1543,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.3565358195387298,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.1608,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.29842568776966194,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.1575,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.3162559090710253,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.1769,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.4363872979774328,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.1586,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.38500483180061584,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.1563,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.33011672409355,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.1643,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.32637924272498287,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.162,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.3505910138837611,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.1548,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.4048929445070551,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.1577,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.3570347334734474,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.157,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.30751189045756544,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.1629,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.30346108043857634,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.1638,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.3669470608290662,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.1661,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.34088976047432745,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.1778,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.31158179251727275,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.165,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.29723664923826393,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.1599,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.29146953323527564,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.1531,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.3057914616266899,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.1671,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.28571516084987814,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.1715,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.28568472631369546,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.155,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.29059537164579746,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.168,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.2924712742220355,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.1525,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.2956485425465924,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.1667,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.28319719079324207,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.162,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.2865912081386556,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.1775,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.3044907727475658,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.1468,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.29812642579035753,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.162,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.291233928511169,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.1405,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.27603486186825393,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.1562,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.2936562684062661,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.1559,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.3001730873818616,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.1634,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.29576759131905994,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.1382,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.3013257959888076,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.1589,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.2639022394337357,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.1459,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.2770321704522631,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.1605,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.26809204515631535,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.1736,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.29855210259785053,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.1459,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.258418929400964,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.1603,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.2730930379538582,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.1656,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.28997078583312136,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.1407,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.28231433414977813,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.168,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.28263432394203397,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.1708,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.2746643641397258,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.1599,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.2918353413096726,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.1556,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.26174626708607135,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.1674,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.2582514698778396,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.1584,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.27666799898806904,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.1474,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.26503274476711675,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.1669,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.27065375085133836,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.1681,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.28302701834628347,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1689,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.29080723332113567,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.1632,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.27185495280062694,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.1622,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.2634843867189924,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.1718,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.26925888761044703,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.1487,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.2581167231864253,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.1706,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.26951021267938297,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.1589,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.27626757518479467,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.18,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.27723464799617675,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.1603,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.27917508029849486,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.1587,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.2869668138208736,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.1546,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.27179383540157387,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.1585,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.2768337583005842,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.1784,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.2764488488733009,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.1662,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.25681196739789514,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.1637,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.2894624557798769,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.1626,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.2806508144467033,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.1605,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.2732392108426814,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.1603,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.2834055999730972,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.1634,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.26266014407059796,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.1723,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.2894141087206624,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.1597,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.28266541045277666,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.163,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.28676421982020656,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.1538,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.29018780840144814,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.1713,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.2868441118280559,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.1463,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.27898137401994244,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.1413,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.2786801103615278,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.1766,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.28324015982534645,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.1577,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.2834414220994184,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.1761,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.2723250629589045,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.1603,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.27117587086051,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.1658,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.2781022932369818,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.1606,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.2874486385366433,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.1644,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.26897346562825414,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.1601,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.28120099195646375,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.1598,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.28913142644213635,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.1597,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.2731105879400601,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.1487,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.2690532475678929,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.1511,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.27995032907040307,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.1596,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.27282817047486657,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.1625,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.2855233368790597,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.1654,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.2800592153169081,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.1546,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.2821718989788899,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.163,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.26909130038382173,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.1571,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.2864491641886432,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.1523,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.28117742119275674,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.1674,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.26887911053891445,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.1555,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.2667682869281734,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.1555,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.2657821433342385,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.1581,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.2714701100862805,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.14,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.27617966855598963,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.1587,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.27194105838491445,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.1656,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.2630690718372924,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.1622,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.26024186600673926,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.1502,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.2838580147181834,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.16,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.2776664257931295,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.1699,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.284863953398789,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.1604,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.27623045433062016,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.1626,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.28480389444501025,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.1675,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.28225106363212005,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.1459,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.2697434813477739,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.1592,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.2728929637317581,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.169,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.2670782125818066,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.164,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.2639276350897984,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.1582,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.28027795430046376,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.1538,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.2696207032880184,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.1637,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.2555237928146565,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.1782,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.26222021030173,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.1652,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.26886065423337774,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.1737,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.2793403734589966,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.1627,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.27805167205411285,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.1541,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.2792105887612544,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.1653,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.27440851054082915,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.1532,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.2814109499447084,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.1759,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.25880254903897326,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.1518,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.27608730836284706,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.162,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.27648920127333715,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.162,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.2725370677773201,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.1672,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.28271312889915307,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.1583,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.2627797884621987,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.1621,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.28535526489606833,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.1622,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.2614075510354235,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.1619,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.28702318631027307,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.1469,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.2762126421625312,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.1794,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.27232025747023386,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.1534,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.28072286460364726,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.161,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.27125308298476186,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.146,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.2654694520115554,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.1699,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.2631779062157047,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.1575,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.2620992509273041,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.1778,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.2763575230255207,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.152,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.2677694539178189,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.1647,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.2697477002692669,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.1755,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.2797102948791675,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.1499,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.26269274875255294,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.1713,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.26711689999008525,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.1608,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.2778427216969002,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.1467,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.27150332600982374,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.1667,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.2782159990788663,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.1518,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.2829675597003286,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.1584,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.2771783738440957,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.1539,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.2734729457015925,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.1568,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.2673465291512631,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.16,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.26735970106528817,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.1647,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.28197714747476366,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.1452,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.29370540892714647,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.1542,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.2666732404436997,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.1656,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.37001342154714834,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.1756,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.2863169433730486,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.1441,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.2755833034376334,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.1475,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.27166984817820816,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.1487,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.2830355392569271,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.163,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.272374982473099,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.1597,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.25919066039095123,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.1607,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.26649636848632063,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.1582,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.2648709566661642,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.1589,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.25730324995062714,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.1602,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.2655708055776232,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.1793,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.2707104242951893,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.1583,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.25910699275650034,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.155,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.28867730446693196,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.1548,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.26117919953971325,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.1565,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.26975740323979597,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.1499,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.2707285506721632,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1608,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.2701418204798252,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.1541,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.2699230382564218,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.1693,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.2713027652253941,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.1587,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.24830022390436599,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.1702,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.2580089341006499,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.1556,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.26704112493637255,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.1733,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.2713643208727579,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.143,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.2573400425484023,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.1763,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.2985491171243398,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.1586,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.2592576688098297,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.1554,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.2732863909548894,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.1754,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.26697201671950793,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.1424,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.2766245157465264,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.1572,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.26294126871367496,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.1625,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.27661812983360295,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.1424,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.2627559538811197,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.1722,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.25984250831631384,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.1669,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.2626949465357649,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.1576,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.2616618953394594,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.1742,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.2641889089953724,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.1573,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.2663078604042808,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.1645,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.265664994916671,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.1583,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.25346593492009767,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.1687,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.2747569919263603,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.1644,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.2690133628708084,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.1617,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.2619399813589476,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.1705,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.26230104168542545,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.1542,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.25402060243020286,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.1574,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.2705100260122791,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.1539,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.26640995478003293,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.1651,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.2599382143004445,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.1647,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.26213802336485326,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.1616,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.2745455052561242,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.1446,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.2671420912976016,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.164,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.2750349609360415,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.1511,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.26570974080613324,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.1484,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.2638738969288533,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.1619,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.258106426875614,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.1524,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.25862539692945147,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.1727,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.2675693902538549,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.1751,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.2731735582402155,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.1557,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.2638703209719976,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.1492,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.270296059870039,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.1594,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.26254549633593344,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.1621,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.27098922958451754,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.1639,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.2565726837493968,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.1565,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.2789378985515267,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.1474,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.266442188508462,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.1622,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.26066170670503536,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.155,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.27521573347475575,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.1668,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.27535339805157444,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.155,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.2849705603801833,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.1539,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.2685269596205029,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.1719,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.2637493765755841,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.1681,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.27152461292804464,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.1479,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.283801434306346,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.1483,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.27490333973051395,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.1551,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.25611380760640484,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.1648,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.26995914079702954,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.1757,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.26425306269323734,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.166,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.26669282205158495,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.1701,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.27685406702107274,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.1449,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.27565026725322117,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.1587,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.2613652188560177,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.1578,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.26555819878166825,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.1486,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.266510774746655,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.1512,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.2690162664521917,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.1566,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.2679627757064271,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.1632,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.2703111359053883,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.1562,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.2665174061778603,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.1645,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.2737934624109558,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.1566,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.25957417860194426,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.1623,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.25543261462937544,
+      "learning_rate": 0.0,
+      "loss": 0.1462,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 1443291817443328.0,
+      "train_loss": 0.31840668920321985,
+      "train_runtime": 24763.6609,
+      "train_samples_per_second": 6.38,
+      "train_steps_per_second": 0.05
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1235,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1443291817443328.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}