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": 6.86441110073457,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.8532,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.6045158436343865,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.8211,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.691342523632329,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.8504,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.2214903644378055,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.791,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.57343181795257,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.8401,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 6.010640116292601,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.7843,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 5.029625924582106,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.7761,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.8987397401684,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.7868,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.7767148560221595,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.7112,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.4814643831179017,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.6921,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.471646565125194,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.7174,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.618681186862345,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.6748,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 3.8473779514254316,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.6787,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.724985128038088,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.6733,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.4565127166253564,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.6475,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 2.9070901844668047,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.6031,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.8181653852827995,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.6011,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.3085753794741013,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.5951,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.6730555395031828,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.5558,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.4443273397488952,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.5628,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.493466757307261,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.5563,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.66888702161661,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.5444,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.5287984412597866,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.5239,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.0975285019189867,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.5241,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.7707774769949128,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.5286,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.0973689299690406,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.5356,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 1.0623063274460032,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.5074,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.9213098637363795,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.5156,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.7279846642167532,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.5095,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.8480523540918409,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.5206,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.8247049068761533,
+      "learning_rate": 1e-05,
+      "loss": 0.4899,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.7141212225261305,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.4744,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.7522946522317299,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.4901,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.6999081531605006,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.4779,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.700401011718557,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.5006,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.6118116546046325,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.4676,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6257061757816362,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.4788,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.6280389937252292,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.4726,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.5392826690649501,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.451,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.6173240264494104,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.4618,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.6369491831755606,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.4562,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.5856771993549854,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.4758,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.5668366169897744,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.4538,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.6437120007069501,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.479,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.627743494612135,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.4717,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.5512481802175274,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.4722,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.5563107063190943,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.4655,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.6567151328079639,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.4526,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.5352201319611875,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.4526,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.5127917780532257,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.4413,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.6918387003982951,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.4707,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.5279918639890843,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.4475,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.5189691620415579,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.4475,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.5549784969268997,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.4526,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.5875409511925007,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.4524,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 1.2420234149584681,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.463,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.6002420271862579,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.4518,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.565463506877763,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.4705,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.585962908866909,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.4418,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.5737797839906102,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.4698,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.6038007868582123,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.4406,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.5274973631375717,
+      "learning_rate": 2e-05,
+      "loss": 0.4442,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.5506757525717034,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.4359,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.6833300549677357,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.4478,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.546938217042063,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.434,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.6730178871942498,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.4518,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.5376574049904592,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.4259,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.6126690461256278,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.459,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.681740221261121,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.436,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.5817418265676915,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.4357,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.6104557162236337,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.4434,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.5776840889997493,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.4405,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.6407925511195628,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.4572,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.6063400363195796,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.4327,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.6364540326250308,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.4494,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.6369876051967273,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.4295,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.5119598661028096,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.4496,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.581702879715988,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.4464,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.6483409368419777,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.4304,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.7484008100440136,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.4442,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.8684556918760044,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.4399,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.755186188730311,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.421,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.6538236985116527,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.4315,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.7176532198893885,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.4223,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.9782342536095716,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.4399,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.7366475349252818,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.4143,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.8514564371143069,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.4294,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.766600206488167,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.4501,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.9156567165794338,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.4367,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.5906086782366816,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.4291,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.0155877680564096,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.4357,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.9128724878283175,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.4532,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.685339649412695,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.4366,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.6535576400369847,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.43,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.611152454356326,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.4478,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.6882390249973058,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.432,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.5715008957432033,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.429,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.6542593281403334,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.4318,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.5833689707089386,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.4404,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.369845042618769,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.4673,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.6846827343138276,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.4346,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.7530309803928045,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.4331,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.8513018532447264,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.4284,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.785095602429644,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.4256,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.5890626228063734,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.4611,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 0.7719627831588908,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.4321,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 0.5575798912602299,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.4402,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.7098455279243973,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.4362,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.7586668800159,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.4621,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 0.7926665305290603,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.4388,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.7745980781159333,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.4044,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.854884416224998,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.4286,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 0.8223582429081083,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.4381,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.6982805481012927,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.4187,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.9640412550978397,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.4211,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 0.6296555438026555,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.4271,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.8982451918558461,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.4231,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.6567519812896049,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.4333,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.6629820013629012,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.4243,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.7488191123711658,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.4179,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.6317703334557587,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.4364,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.8021947028626214,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.4115,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.7374611332389471,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.4126,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.8551867929597495,
+      "learning_rate": 4e-05,
+      "loss": 0.4444,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.740272420434549,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.4312,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 1.048198781357459,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.4305,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.7358645139825101,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.4353,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.101878545091989,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.4502,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 0.7227496021913535,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.4321,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.8388473027968623,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.4456,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.7778354304354507,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.4273,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.7980819893320266,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.4262,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.6596263752434344,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.4221,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.8385525692503705,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.4185,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.7513176796954192,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.4279,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.8197980204065312,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.4224,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.5974738852328039,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.4177,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.8289659303979477,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.4124,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.6068372354135484,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.4501,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.7887981789775002,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.4194,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.6259788448116591,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.4315,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.6707799716089261,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.443,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.615918082879032,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.4142,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.7711166901875551,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.42,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.7714824521778909,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.4252,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.48855712357735515,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.4278,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.6970239552110373,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.4244,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.46117866276771546,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.4122,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.6069282509726415,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.4229,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.4266450368693798,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.4201,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.589528710566419,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.4291,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.6059685030114478,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.4304,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.5256631108781732,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.425,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.6864743710941952,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.4273,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.5229171426063617,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.4209,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.7840701324605094,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.4409,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.44172941408665817,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.414,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.5831644916701094,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.4116,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.47952591632800196,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.4209,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.5203677061183267,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.4169,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.5764781242858122,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.4146,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.4430208374755826,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.4386,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.5538789366089516,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.414,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.47637149103029347,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.4098,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.5790697111086321,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.4269,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.45193717405487455,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.4369,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.7330827083889814,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.4395,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.5075141142579204,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.4156,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.5304419330123171,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.4086,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.5006082345698895,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.4171,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.4533087944791989,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.4228,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.5344080678937222,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.4162,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.42590383510132723,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.4298,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.5276474561672112,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.4087,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.4240056369922797,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.4039,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.5542885870106194,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.4446,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.4434889701302054,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.4164,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.4666152171588739,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.4045,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.5285752875253302,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.4153,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.437358927674123,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.4111,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.607685524818305,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.4145,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.5300023504104183,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.4186,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.517975865963516,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.4203,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.7215295862054601,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.4311,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.4135556991606804,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.4108,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.6241577605714629,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.4385,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.49992879162207887,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.4305,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.5088574566732762,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.4102,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.4814710568906568,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.4104,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.4660897546346199,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.4197,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.5597595825925853,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.4228,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.46515420607196817,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.4075,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.563933907412093,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.4068,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.43472364060734286,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.4282,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.5434244661639891,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.4227,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.47096456146245924,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.4125,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.5338576576923281,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.4215,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.4945214324795181,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.41,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.5619966441549231,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.4319,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.44733444232705577,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.4102,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.5704223532869496,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.4088,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.44576882239613685,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.4275,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.48592491795637355,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.4165,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.44907460987846765,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.4205,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.4959381349169951,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.4207,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.4188889694480753,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.4036,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.5034104648843911,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.421,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.5563330483975186,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.4151,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.4816807365147703,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.3953,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.45121218762847876,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.4162,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.47729083483174856,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.4171,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.42114850113531227,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.4064,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.42978394328651404,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.4114,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.4691187627365153,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.422,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.5133158714053301,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.4185,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.4999376527603344,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.4102,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.4716274336555487,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.3938,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.5134208777288632,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.4306,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.522603276356573,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.4024,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.5266705205794929,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.4081,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.5288539477446305,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.4314,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.498790017083251,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.4232,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.48961517611191574,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.4212,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.4634259998930869,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.386,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.45769298583285156,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.4125,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.3874802649502647,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.4099,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.4485211571781878,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.4005,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.44270643326936143,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.4205,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.42896957321414064,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.3868,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.4469299941902741,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.4131,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.4676858887354166,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.3941,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.42775768394211466,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.4298,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.46396634608646525,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.411,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.47813763967065226,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.3962,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.482839212342319,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.4099,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.4089160809189028,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.3987,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.45591066574083183,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.4103,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.4449979736852856,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.422,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.40727479497685815,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.4225,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.45665102782522093,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.4188,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.45002026954102614,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.436,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.46403946610003427,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.414,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.4474429086026941,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.4028,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.461198428801036,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.3945,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.39513372169894545,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.4084,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.4418082042580754,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.3976,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4212870541488876,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.432,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.5085860017993891,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.3687,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.4316145227236286,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.3501,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.46243390108290083,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.3269,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.5919003187057721,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.355,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.496152679194222,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.3337,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.5076798251857062,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.3452,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.5355890355707907,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.3288,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.598900863725646,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.3424,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.5021597461871536,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.3439,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 3.339894124098262,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.3359,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.7212216852097976,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.3382,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.4419458175022745,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.3246,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.5497838867848197,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.3392,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.5136206796337063,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.3411,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.3977575935233999,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.3385,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.5593695513391831,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.3556,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.4200427061857459,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.3256,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.47000441911540175,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.3387,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.47364239127526675,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.3455,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.4890821157886735,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.3377,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.46104906105554044,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.336,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.42682599614089517,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.3503,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.4367843199607088,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.3474,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.4308261119590789,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.3106,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.5001209055441397,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.3433,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.47302504190540484,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.3426,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.4398871675574398,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.3441,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.4401512479408716,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.3228,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.49056660367468713,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.3227,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.43902126375380623,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.3509,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.5619486188896424,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.3462,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.44064598567393115,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.3395,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.48109965699624635,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.3475,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.45154914715175126,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.3424,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.42297506559141995,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.3424,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.5011405255050914,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.3484,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.4314504126327535,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.3448,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.466661960239932,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.3114,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.47246265834742174,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.3391,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.48856123988454425,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.3508,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.43766160778077434,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.3286,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.431825010442556,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.3239,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.4535054325242467,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.3533,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.5266178086437343,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.3305,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.5212748809353211,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.3471,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.45086940719762597,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.3288,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.607989522466672,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.3439,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.4748962314676769,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.3477,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.5851603093439823,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.3496,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.5015864648704115,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.3462,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.5335746308362832,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.3197,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.4529777003405075,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.3454,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.4495659653570179,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.3474,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.5076434092610291,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.317,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.39966028802989095,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.3465,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.5772148912274491,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.343,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.4602056098144374,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.3207,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.47705533951115886,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.3436,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.4427702885884884,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.3335,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.43938279989110823,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.323,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.49582591180425534,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.3383,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.4939985733934902,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.344,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.4165610354063779,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.3294,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.5073929121441291,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.3218,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.43747913186396387,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.3306,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.45891067119023793,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.3328,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.41148877565739306,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.3407,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.4203754276930739,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.3439,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.36879736934312407,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.3365,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.4006556663713527,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.3482,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.38603945537570605,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.3252,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.46493426553959916,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.333,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.46210685983659044,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.3418,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.4439699838562416,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.3659,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.4013169364451426,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.3455,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.44015385941249396,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.3299,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.38068151771339437,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.3434,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.48017152930186774,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.3445,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.37227537585029863,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.3446,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.4072418398511352,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.3417,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.39758711510938105,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.3527,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.4008809773508571,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.3485,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.40427812247082173,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.3513,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.37042309048319766,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.3228,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.36819676401529416,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.3396,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.3623043110279907,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.3395,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.3623162003934327,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.3445,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.38953450864175765,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.3453,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.34836535700015986,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.341,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.39490569503330414,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.3456,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.3695773863024579,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.32,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.42057571688478507,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.316,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.3892039168917071,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.3156,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.39994984681165985,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.3334,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.39559939216981826,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.3457,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.37062176285045956,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.3405,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.4305417139021015,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.3443,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.3588368335065246,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.335,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.4690801595404091,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.3545,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.428737081900067,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.3519,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.8103859245372029,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.3487,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.3699816856920258,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.3408,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.4096080931325833,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.3522,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.4013281176942817,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.3429,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.4154225643998519,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.3468,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.45092054483600313,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.347,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.4397558206208973,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.3346,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.42704597667636884,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.3406,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.44141033434369326,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.3395,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.46606008776176894,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.3304,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.3450993806321395,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.339,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.4174476566576018,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.3628,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.40010419987998164,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.3533,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.4019367790822593,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.3508,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.36637354955969603,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.3541,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.4058285123437725,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.3343,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.4227474520813658,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.3377,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.38867807314573616,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.3476,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.42591540426541924,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.3442,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.4058630157189682,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.3587,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.375296661643467,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.3494,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.40680315941391076,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.3432,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.4063018919659273,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.349,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.401725502964154,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.3257,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.3801838214525121,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.335,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.4390072152078119,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.3445,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.4258102642928917,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.3513,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.39609940187997666,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.3333,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.37078116909082764,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.3244,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.3823039422255809,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.3358,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.402407169408178,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.3361,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.41910510741030865,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.3433,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.3537683899424441,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.3321,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.3526684774678277,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.3244,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.42176184222174834,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.3248,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.4154584967435749,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.3289,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.4180945899111746,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.3302,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.39409225290109323,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.3357,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.4370006067435742,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.3518,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.35282579532296243,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.3302,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.4329864907813953,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.3393,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.3866587570463195,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.3404,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.39653666855120473,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.3233,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.4001385551240655,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.3389,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.43411482967744186,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.3449,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.4606756084652878,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.3372,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.4203461651902823,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.3304,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.42574979318312034,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.3359,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.4395154066600487,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.3587,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.4091491479894734,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.3373,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.45125371830570643,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.3426,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.40672515285108807,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.3484,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.413064728726335,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.3292,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.4193844256740911,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.353,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.40690079373298194,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.3447,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.4125014340119729,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.3407,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.4092487202912703,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.3314,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.3607996825836964,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.3544,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.37154504837440394,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.3455,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.3642849619810481,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.3368,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.41280650920048695,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.3423,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.3678957944419821,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.3285,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.4395918679781465,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.3567,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.37360525670487227,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.348,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.402051785150735,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.3331,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.36059464734855845,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.3321,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.3868791890986445,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.3438,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.37119146049017754,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.3629,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.40413213697713213,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.3435,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.45867826520003124,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.3282,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.4121808528080243,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.3201,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.4367875930390528,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.3558,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.4271139878240527,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.3438,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.37786962424476694,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.3283,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.40452780389895315,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.3312,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.371207137020947,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.3406,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.43896912873659644,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.3371,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.3935720022914307,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.3435,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.4134583003571375,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.3484,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.4008069013999872,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.35,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.42344331472414165,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.3402,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.3576696114711609,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.3414,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.44418214117708277,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.3486,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.3647832290048461,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.3519,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.4244781407475609,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.3391,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.41814601225146514,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.3337,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.4230731642482515,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.3569,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.4190415356209953,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.3287,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.3525072701591149,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.3535,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.3950407015633332,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.3315,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.37174214028559416,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.3295,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.4278023980114727,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.3532,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.3966673910299886,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.3313,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.40722060590033843,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.3592,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.42141992715312715,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.334,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.36479488930706505,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.3345,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.3782513686195105,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.3469,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.3262438497036014,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.3321,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.3859641363407318,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.3266,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.37817179904624676,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.3347,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.4421574208779731,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.3367,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.42785695208889074,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.3364,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.36827746748167034,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.3222,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.38775562109904743,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.339,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.34671442598912766,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.3202,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.41273007305488985,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.3604,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.4113369543156994,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.3397,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.4107409398046558,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.3325,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.40986583807789634,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.34,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.4283534119960764,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.3308,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.37197027299476554,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.3467,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.4025194909473333,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.3371,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.4218018760788875,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.3304,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.41374427768183486,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.3295,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.5078620286509733,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.3482,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.4325985381565834,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.3505,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.45360983271984867,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.3486,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.4224628355366537,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.3252,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.4737663839901841,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.3391,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.3876940477636295,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.3383,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.44198560639867857,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.3289,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.39436021880243594,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.3455,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.4453950713698233,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.3369,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.3858802800782535,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.3189,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.4589681046849382,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.3412,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.4012896243767029,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.3221,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.45651398289521145,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.3337,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.4023448513140741,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.3362,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.3839282859837464,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.3293,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.3549269659672028,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.3501,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.39708620284040275,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.3415,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.3376313107153828,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.3261,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.3555837134120192,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.3445,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.357774993885783,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.3317,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.3672638296255847,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.3385,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.40826183429184904,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.345,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.37607031771880034,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.3177,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.3827393316910095,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.3613,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.3528015110848137,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.3431,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.33246081666585836,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.3377,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.3571095643927091,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.3459,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.35017707108040746,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.3384,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.39105692205823944,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.3228,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.37874425069918005,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.3326,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.3732976008558461,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.3303,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.35343233949482733,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.3333,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.36754646577576083,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.325,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.485684513014217,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.2687,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.46321296386169564,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.2748,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.5053852594785839,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.2362,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.4461699775823914,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.251,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.4434268348149064,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.2503,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.4039748006809421,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.2463,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.38423584754995355,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.2621,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.42611899299200473,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.2425,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.4050854302189589,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.2376,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.3910157494810868,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.2492,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.40110140338357175,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.2507,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.3869590901190326,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.2544,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.37774563068369094,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.2668,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.3881133761704007,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.2587,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.3937600398724546,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.236,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.41124436721714286,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.2622,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.4019527994230186,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.2536,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.366182983142453,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.2238,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.39510213903024893,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.251,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.34473631198316135,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.2622,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.38962153438344815,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.2437,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.35513971936483313,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.2495,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.3623834881705085,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.2231,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.35732507196906765,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.2338,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.37704611330279936,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.2493,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.3500270375451894,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.2568,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.39334475949259073,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.2548,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.3553762565263293,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.2452,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.38284184632537566,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.2531,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.3784541163115441,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.2612,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.34049490531194465,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.2383,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.3448245047169509,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.2303,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.3516783146904345,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.2568,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.3742810631044441,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.2422,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.35940548781407705,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.2335,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.3786055341277785,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.2574,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.35959747304811024,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.2478,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.3633751841803164,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.2361,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.34702423086718753,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.2579,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.3570572419001988,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.2504,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.39355590030857657,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.2654,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.3334520858336479,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.2664,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.3659491644172636,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.2593,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.3864919818358154,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.2587,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.30476483356275585,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.2391,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.36375926752755355,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.2593,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.3480044514100528,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.2391,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.3441530377026589,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.2528,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.31861799860555534,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.2462,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.3630936336234069,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.228,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.31146097840802633,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.2539,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.3354173080870005,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.238,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.33716218598037495,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.2333,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.32807444871367764,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.246,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.339553202277947,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.2517,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.3533796721856939,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.276,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.35646194699281164,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.2575,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.3264102279399703,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.2462,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.36514533195095933,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.2552,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.36103786711140784,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.2668,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.35666624560183413,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.2492,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.355282533923525,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.2241,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.3626728874347034,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.254,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.4020628484128989,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.2559,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.35670117606455515,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.2638,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.3855172111155011,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.2424,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.4011090485179891,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.2701,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.36592956504073915,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.2582,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.40216182925996863,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.2611,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.3694979391572729,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.2472,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.3659504636130495,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.2592,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.3336193003865009,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.2628,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.3601798256672282,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.273,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.37361929565158875,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.2443,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.3318485240276134,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.2428,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.40087072217365116,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.2478,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.36318254810382605,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.2551,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.4400093128431157,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.2669,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.3261305251966751,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.2421,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.4101709352421423,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.2749,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.32962802377854383,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.2486,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.356820398111218,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.25,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.3851732544055837,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.2525,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.3764586246703676,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.2436,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.374018729084165,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.2516,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.36777033444114343,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.2548,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.39231927625445556,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.2743,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.32863409704532526,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.2508,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.3571343224090065,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.2308,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.3247546129094774,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.2343,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.3465312471129698,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.2356,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.3452600556576025,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.2482,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.349672794717827,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.2594,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.36031176125052283,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.2416,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.3562910278359719,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.2566,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.366676455150237,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.2428,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.36041977007297804,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.2316,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3692220269119681,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.257,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.31402003062126743,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.2415,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.3850702961781947,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.2405,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.3348320694202487,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.253,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.36206380242236663,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.2392,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.3030608936121492,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.2463,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.35281727329248963,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.2388,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.3424731426542641,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.2558,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.33571501296873935,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.2298,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.3789220131260508,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.2538,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.33030066151160054,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.2813,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.38714638312002636,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.2572,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.34530755971209565,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.249,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.3342787283673605,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.2558,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.35442889789377624,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.2465,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.3283143541353557,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.2441,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.37163793846021914,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.2807,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.3356062700409527,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.2578,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.3291900449863413,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.2495,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.354575189215529,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.2514,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.3487408084269113,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.2334,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.34436676021967894,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.2441,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.31355242180719206,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.2436,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.35091743157391164,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.255,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.3076626604410713,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.2425,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.3141078779344448,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.2454,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.3450003781272326,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.2477,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.3215041942459992,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.2501,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.3094622272708945,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.2469,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.30945640736295804,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.2537,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.3025026934986793,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.2546,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.3155107651029845,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.2522,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.3037402713452186,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.2574,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.32205719103578356,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.234,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.3038589003650163,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.2632,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.33693831456615836,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.2546,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.3560586151367552,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.2526,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.31303605766895815,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.2473,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.3778864583700518,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.2439,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.3447870342462345,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.2408,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3433514449096688,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.2483,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.3330702951677056,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.2448,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.3326856717067384,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.2558,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.34393700951344985,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.2253,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.3510992162351781,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.2566,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.3587233505345799,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.2464,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.3523242758339907,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.2555,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.38006856864825395,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.2629,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.32427128895828,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.2534,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.3570189819323133,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.2862,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.33314938600601157,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.2575,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.3276745785369723,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.2307,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.31400134497842236,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.2378,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.3330902814889232,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.2593,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.3326670006995015,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.2633,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.3280690701558275,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.236,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.3278095091311887,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.2703,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.3512551562463348,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.2463,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.3104318252453822,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.241,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.3401207938122007,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.2545,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.3336810016257656,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.2429,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.28762206596555123,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.2441,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.3115718230595019,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.2504,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.31111117530283644,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.247,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.32776983300763457,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.241,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.32523824801644197,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.2474,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.3066441888378739,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.2429,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.3487954318713217,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.246,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.3139585777524444,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.2374,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.3233572383210003,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.2275,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.3318718292445782,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.2534,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.327412726016868,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.2261,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.35332023999420237,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.2581,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.32874819891166723,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.2397,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.32652809250499387,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.2514,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.3125902814792465,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.2581,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.31469273103577694,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.2529,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.32219735978666564,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.257,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.2879933038949073,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.248,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.3343106228483908,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.2481,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.3323017883255548,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.241,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.29255377939986427,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.2549,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.32825663144100264,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.2432,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.3319260526790599,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.2343,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.3176455685724195,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.243,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.33030437329367945,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.252,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.3405455330941157,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.2355,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.32277839722391527,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.2353,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.34489438779661474,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.2452,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.33049019779086536,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.2401,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.31551598234960704,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.2449,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.3491516361413624,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.2371,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.3327723352359509,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.2449,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.3038901196749291,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.2493,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.3539862905316079,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.253,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.3279259345703185,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.2557,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.314123669287423,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.226,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.3112061402395981,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.2418,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.3335793608835712,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.2462,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.2913734099747909,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.2458,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.3009821335355907,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.2532,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.9492238902469251,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.284,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.3263920097329214,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.2391,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.29533390583782154,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.2611,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.30384397873768954,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.2291,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.6875562683821207,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.2521,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.3085806650226585,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.268,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.3064413738201948,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.2634,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.3171532554300705,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.2387,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.29244834757296123,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.2285,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.3159638163740333,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.239,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.3181132192687635,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.237,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.34914614928036664,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.2441,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.31954147038498176,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.2594,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.30873847469028737,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.2465,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.34101024973870026,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.2435,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.3160488491952463,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.2362,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.31769043489753773,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.2375,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.344678309990249,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.23,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.329367157609335,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.246,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.34406797285806523,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.2515,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.30457833925476197,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.2536,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.3463502346213935,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.2515,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.33193256239205404,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.2503,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.3218812465974451,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.2503,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.3981379785296001,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.2229,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.3301920596244831,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.2491,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.33068742298123,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.23,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.34783363021020886,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.2492,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.3224035159195022,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.2506,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.3713875943949281,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.2298,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.3269626548120076,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.2338,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.3089329623580857,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.2339,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.33128591834367027,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.2419,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.3156291696008085,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.2506,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.3236160230510039,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.2526,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.3148436442766608,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.2645,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.3332698858714502,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.2206,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.3143575868962332,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.2339,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.33068600823639005,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.2471,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.2994988805276603,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.2581,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.32727522392071545,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.2313,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.3371076574809368,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.2433,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.32847048920503735,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.243,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.3094513600430449,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.2582,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.3143346908510896,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.2582,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.30448720420196346,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.2539,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.3233849180776509,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.2695,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.335089338840107,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.258,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.3084761651077523,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.2296,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.42748249345640127,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.1947,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.2981724306892771,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.1856,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.45004301561948656,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.1708,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.44762399407753783,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.1685,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.3367719404619726,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.1775,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.38780063236498297,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.1713,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.36842944371447967,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.1722,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.3165425973925215,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.1809,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.3313149656560062,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.1539,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.3695282249347558,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.1864,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.359038142922943,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.1679,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.29325725083481047,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.179,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.35181864612103037,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.1963,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.31933096316893517,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.1764,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.3208086228576384,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.1993,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.3331229045888001,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2018,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.30488442160739315,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.1858,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.3119480910731893,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.1804,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.3014067954306663,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.177,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.3018777993288663,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.1815,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.30318810409235497,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.1749,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.28421077473428674,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.1673,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.2896580901094542,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.1736,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.3147141909560688,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.1766,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.29583229622596674,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.1695,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.29449669292867653,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.1592,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.3017641211962018,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.1592,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.2808197062230855,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.1886,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.3177228670629668,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.1728,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.3212082470141559,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.1675,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.29149272371548657,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.1959,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.3138660474242436,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.1807,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.32041434351295567,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.1695,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.3034587216816694,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.1681,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.29012921741803865,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.1663,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.32592141279463976,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.1774,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.3085131622199101,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.195,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.29435823658119076,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.1663,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.2675635744184346,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.1742,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.28016612077238046,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.1837,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.303000911085431,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.1797,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.2739113344659076,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.1727,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.30166339552268984,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.18,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.2774537580520377,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.1935,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.2941657896727021,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.1832,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.31142157094295025,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.166,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.2933804250449896,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.1774,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.29618416599053593,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.1741,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.30108254238526416,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.1824,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.2782118168440768,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.1639,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.28717260233287123,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.1632,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.2842604064798886,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.1715,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.30786210065174807,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.1642,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.2791489774350654,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.1817,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.2873268462848564,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.1811,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.2949187820105075,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.1822,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.2773804508320204,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.1729,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.27648853906376014,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.1752,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.2778105766371568,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.1731,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.29019990866300865,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.1801,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.2710466101765393,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.1764,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.2704665879085231,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.1917,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.2760458584557888,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.1969,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.2857688698181733,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.1695,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.27175652722532945,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.1711,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.2751023337292547,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.1824,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.2800535070147004,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.1556,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.2781679976882393,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.168,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.29160490722221755,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.1657,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.2729385685205224,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.1762,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.262201532087736,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.1913,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.2791822630644507,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.182,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.2787308698400942,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.19,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.26836122945263186,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.1646,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.28373097602123326,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.1872,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.2622831677900864,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.1503,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.27418146951794026,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2022,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.30196072131482876,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.1717,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.2701569768696176,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.1895,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.2686675460551628,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.1689,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.31269473600320014,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.1862,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.28869024260143944,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.1713,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.2572495130226004,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.18,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.3020136153919306,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.1532,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.28013413413712973,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.1706,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.2618856780464729,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.1686,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.2946977212678809,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.1798,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.27041972586194707,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.1847,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.281780379777813,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.1735,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.2739803459818425,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.1945,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.27643966405929127,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.1738,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.27535124524312404,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.1867,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.27626346089720494,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.1833,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.26779400744924226,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.1832,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.2721010101422728,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.1681,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.27132427743036663,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.1746,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.2893771228328681,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.1694,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.28239051040290913,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.1605,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.28218253812743493,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.1622,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.2676965653999051,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.171,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.2636962057085972,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.1485,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.28927878244970623,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.1753,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.2666533829400996,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.1603,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.273833171759711,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.1706,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.27601454743971887,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.1679,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.2670251201471592,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.1517,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.2782962513604988,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.18,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.281437174398041,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.1682,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.27611511378184983,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.1721,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.2759096706233652,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.1669,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.27397282626809205,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.168,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.2936341232164949,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.1881,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.27956497548922055,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.1869,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.266320087415046,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.1793,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.2830907036065366,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.1674,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.2807687143484723,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.1755,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.27474446654085344,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.1707,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.26747771753478056,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.1935,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.2815504383288791,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.1619,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.2730504797122469,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.1493,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.25732307138469523,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.1696,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.2900162113256968,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.158,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.2784367492098622,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.184,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.2546972360558842,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.1884,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.277318769034133,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.1832,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.29036091637810196,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.1605,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.2714913329184137,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.1775,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.290709378929528,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.1511,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.26636303220317975,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.1694,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.26481401521227815,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.1691,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.27519210923847887,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.1713,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.2739232282524256,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.162,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.27832301544671084,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.1688,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.2740626585944513,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.1731,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.2617991291434224,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.1703,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.25444100632598243,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.1785,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.6278058024665621,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2101,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.26912766278705863,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.1684,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.2677460882214703,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.1619,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.28308044330695054,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.1527,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.26200863114673345,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.1659,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.2772197305792563,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.1699,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.2642751458790247,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.1754,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.2561716306311563,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.1799,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.26059749129926824,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.1719,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.26853688175355106,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.1649,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.27137349010804285,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.1483,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.2532277994792483,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.1829,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.2707064256642492,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.181,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.26164979358638735,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.1669,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.26263007835833235,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.1796,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.27380976445761346,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.1821,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.27055180937079554,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.181,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.2640975262984008,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.1761,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.2618245645944789,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.1665,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.27240776168694153,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.1703,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.28478728975560486,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.1676,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.2720018957484135,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.168,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.2645139282492166,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.1919,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.26770828360414617,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.1724,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.281632321763113,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.1677,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.2560566703763865,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.172,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.2710569492638631,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.1739,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.2837858473896614,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.1789,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.2640463427014568,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.1706,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.25820858404644015,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.1803,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.25442455925849966,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.1575,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.25808752140750035,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.1789,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.2665605409660662,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.1658,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.26437288110970236,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.1579,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.25881012106467854,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.1786,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.2592420890853956,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.178,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.25422560569005037,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.1571,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.2562657239138758,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.1652,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.25467827756921474,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.1751,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.2830668414722815,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.1709,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.2565638341910584,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.1627,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.28210954732512816,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.1681,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.25430979549753496,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.1647,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.2707548614216691,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.1837,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.2709350821429233,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.1586,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.25602672024704476,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.1642,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.2516470407836315,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.1632,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.294369293127213,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.1833,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.2659677262583918,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.1757,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.26377217473320425,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.182,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.27427840080150745,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.1636,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.26095444575618276,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.1897,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.26622159790610234,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.1654,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.26216903964243626,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.1778,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.2557059537753074,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.163,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.2630327252817714,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.1871,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.26316824577984127,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.1556,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.26449090046403756,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.161,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.2576098566451076,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.1678,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.25950637702964313,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.1691,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.26408894002703837,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.1457,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.26367733004998667,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.1596,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.258695451476985,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.1918,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.26736961463275993,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.1455,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.2461484105231951,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.1774,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.25866684174470866,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.1615,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.25959471045580046,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.1686,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.2591567401269358,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.1804,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.2538223625920029,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.1745,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.2503222363063209,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.1576,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.2565259648375137,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.1562,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.26138018414128744,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.1585,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.24966612765646545,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.1615,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.27495868086194386,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.1831,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.268775537529647,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.1633,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.2607904245717732,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.1533,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.24520555098911176,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.1668,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.2700747445560125,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.1849,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.2544941878484937,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.1526,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.255261448299792,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.162,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.25916353487304733,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.1597,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.27428903552942047,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.1561,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.2689226770898772,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.1694,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.2586779417754685,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.1741,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.2587949996168294,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.1614,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.25769308278909253,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.157,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.25134784192387916,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.1882,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.2495973557616066,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.1614,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.2602026759913978,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.1735,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.25374124792932007,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.1799,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.25537397960011465,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.1903,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.2588564589598406,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.1968,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.2554485781085126,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.192,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.24725570613343695,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.1507,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.25501932525912857,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.1702,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.2558570453888187,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.1913,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.27022864972636645,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.1645,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.2574966171908723,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.1585,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.26294445396187016,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.1666,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.25233608142235425,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.1676,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.2954040549744632,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.1627,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.25116420081459895,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.1885,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.26367710946096057,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.1576,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.2627137947787128,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.1983,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.2698873094378224,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.1637,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.24867914272663086,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.172,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.2582698446509852,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.1621,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.2566825896698051,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.1587,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.2654339287075346,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.1824,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.26027315778015964,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.1744,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.25545450223866156,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.1756,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.3519832479864061,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.1123,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.2615474667990096,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.1326,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.21782773260355798,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.134,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.23896096594111538,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.1242,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.3075279856110031,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.1442,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.29973899638492907,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.1385,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.28094999037796875,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.1264,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.24753683781355484,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.1268,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.2543940902539816,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.1526,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.2618564402487778,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.1212,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.26101424001687124,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.1387,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.2506727414983693,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.1365,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.21022918145660138,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.137,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.22269123311372105,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.1316,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.23742385068487493,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.1154,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.2510200626218,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.1292,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.23878127675484,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.1391,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.21795244179022658,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.1438,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.2152690122247501,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.1252,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.21010105684027744,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.1356,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.2119851983447927,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.1174,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.20514676080870292,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.1315,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.22274746153597513,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.1286,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.221628107743366,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.13,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.2136995051334424,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.1368,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.21608505251895566,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.1232,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.21625578636803175,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.1399,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.22008461042144908,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.1601,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.21678054998646568,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.1407,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.20840747552843847,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.122,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.21210639867974712,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.1352,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.21215795123980663,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.1215,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.21922102354065262,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.1259,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.2138058289490979,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.1159,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.2197581951252457,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.1502,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.2154099146911876,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.1268,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.20928382168456872,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.1276,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.20476815962909906,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.1281,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.2183728228727097,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.138,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.21303861891136272,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.1098,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.2017394324549134,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.1156,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.2136703146306283,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.1179,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.20175853279015593,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.1103,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.2132244721346289,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.1231,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.2048620764758079,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.136,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.20321920854961256,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.1461,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.2025576883930301,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.1289,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.2038022297949977,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.1139,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.20606899791704858,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.1206,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.22881704420890692,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.1223,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.21369549300924234,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1217,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.20764189247714174,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.1638,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.2095533921481287,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.122,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.2234393622028264,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.111,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.2010836261283663,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.1338,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.210033409773503,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.1449,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.20942519825781591,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.1157,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.21314101537397007,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.1284,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.20084681232784835,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.1292,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.20034169958233805,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.114,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.20766360499747688,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.1242,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.22618157763775557,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.1319,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.2098911100426065,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.1089,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.21352767973438003,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.1177,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.22087953180143022,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.1311,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.2042428213007183,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.133,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.2066973136133399,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.1389,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.20428323370613546,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.1359,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.2142762808237647,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.1317,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.19712775670673632,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.1055,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.20858501150189937,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.1397,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.20484558679734305,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.1281,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.20809098833904247,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.1303,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.21796118454591132,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.1501,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.20651347979424414,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.134,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.20434743089554372,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.1241,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.20573238925949888,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.1487,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.205415809504104,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.1143,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.20602655881547235,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.1263,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.19951209508678963,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.1477,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.21406392057866652,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.1216,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.20398262565644806,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.1253,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.20770663777023782,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.1367,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.19486846264554433,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.1152,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.2109323153672861,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.1308,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.21162667861137197,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.1236,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.2075145790235573,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.1281,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.2111610631687707,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.1325,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.21602240468185785,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.1446,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.2107725379405047,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.1486,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.20606422813137074,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.1353,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.21901621835943527,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.142,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.20601337653638047,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.1268,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.21205825546118004,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.146,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.2114296036727296,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.1435,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.2127340884847858,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.1429,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.21071780479772512,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.1185,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.210606020289066,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.1501,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.20771832162004333,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.1442,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.19441026767048875,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.1288,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.20907768853512532,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.1226,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.21518686118584654,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.1475,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.21050740249065716,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.1229,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.20326101997147122,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.1092,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.21063567553728574,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.1355,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.21198502174630798,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.1374,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.20817647151131405,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.1246,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.20407521196842146,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.1282,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.22612799341755213,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.1429,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.20142234755345648,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.1599,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.20593950413172393,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.1084,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.21393360934932093,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.0993,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.20373084412572404,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.1272,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.19634422466006765,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.1525,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.21743067777602848,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.126,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.20831678492621025,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.1165,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.21047454695546147,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.126,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.20563176727093085,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.1286,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.21102047433156648,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.1162,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.2085341862986006,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.1206,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.2035487351320283,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.1471,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.20228374511627717,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.1247,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.21387137926173913,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.1292,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.20741021481602387,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.137,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.2054474645386455,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.1242,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.2130091181217475,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.0934,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.2068533205453476,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.1244,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.20588061142217937,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.1334,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.20867676607275865,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.1179,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.20498355770916415,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.1231,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.2006058827160042,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.1277,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.21397863454505922,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.1261,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.2146832224179299,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.109,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.20268647340560347,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.1272,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 1.0546130925244834,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.1426,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.20957664691112557,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.1428,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.20249884991277234,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.1413,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.20436487580570883,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.1251,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.2206215251644577,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.0999,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.19847781517675678,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.1132,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.2091638292188479,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.1357,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.20546196573559472,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.1154,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.2051562467638848,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.1377,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.19929733706206712,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.1362,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.2064621215537241,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.1394,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.19434414806022515,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.1182,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.21723146789056935,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.1469,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.2085435536798722,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.1263,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.22548264993392456,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.1291,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.21281628843123437,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.125,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.20759398920856506,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.1198,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.20577652235296703,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.1399,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.20944808352992036,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.1289,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.20146371848762498,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.1388,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.2043633494451667,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.1463,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.20810396644613346,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.1355,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.1989753791186097,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.1138,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.21236041507710499,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.1278,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.2073971026030745,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.1218,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.2133463611033241,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.1519,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.2039970147991027,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.1055,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.1964649918356051,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.1127,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.19413173662048971,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.1453,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.1951638123841024,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.1132,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.20107709872485335,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.1339,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.19965559624978857,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.1414,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.20557985827306438,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.1216,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.19876874472864417,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.1255,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.20503315006786274,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.1315,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.21395411566568687,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.1125,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.20408992373543774,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.1375,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.2070489045679755,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.13,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.20774239532828587,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.1113,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.20704584784693586,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1402,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.21247502828938178,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.1352,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.21770975307719415,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.115,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.21508748432961122,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.1098,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.20698649285625384,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.1203,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.21353840295318885,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.133,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.22217036955611003,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.1333,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.20494702181694346,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.1335,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.19829866283521913,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.1401,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.21403271596917894,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.1533,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.20354254388278414,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.109,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.19923480215494943,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.1315,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.20448106186675266,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.1198,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.2012330756277137,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.1183,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.2097509094649665,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.1441,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.20050976077839622,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.1153,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.19144205617999782,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.1237,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.20138289801564987,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.1274,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.19700298505142155,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.1024,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.20658250907759465,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.128,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.20636647486124102,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.1254,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.2062947665063621,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.1269,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.2121750272182928,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.1195,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.20829129541645525,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.136,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.21212092106523975,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.1405,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.21133706358970292,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.1192,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.19889858919675027,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.1284,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.20870529158327947,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.1373,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.20330249161134575,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.1545,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.2070648315213439,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.1324,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.1974930695319357,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.1248,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.20131951276982035,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.1026,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.21158550057449568,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.1241,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.19794654760836503,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.156,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.20079684670134942,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.1319,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.20761604897538596,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.1291,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.20688687806585274,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.1396,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.21861049483398243,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.1086,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.2119088785967145,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.1297,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.20422245553347862,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.1177,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.22038582549215022,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.1363,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.2078360726476976,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.1359,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.20494312206921161,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.1333,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.20070231686293555,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.1168,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.20093375647654815,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.1283,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.20234990821782226,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.1405,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.20732184612480647,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.1397,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.2050933876038907,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.119,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.19616786050670085,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.1402,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.19283576571349195,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.1191,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.19518288490622543,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.1258,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.2023234997317281,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.149,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.20947522611433175,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.122,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.20559274677192121,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.131,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.19735945267073243,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.1346,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.21418903699332717,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.1173,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.19368783828177,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.1409,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.2010785483151224,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.1167,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.20722105853760556,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.143,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.20802366852703078,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.1446,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.20842212531545204,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.1128,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.21065805943426633,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.112,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.22599574260358699,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.1302,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.2052743269608484,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.1378,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.21366234100456424,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.1206,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.20500499638270533,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.1226,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.21114121173620037,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.1195,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.19251314761028024,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.1206,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.20722430664173502,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.1187,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.20678773512871657,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.1292,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.20149737185514885,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.148,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.21165938729004324,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.1171,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.20766409025062108,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.1348,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.19921581832785998,
+      "learning_rate": 0.0,
+      "loss": 0.1466,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 1.9354956165367726e+18,
+      "train_loss": 0.2693426305707167,
+      "train_runtime": 23659.8967,
+      "train_samples_per_second": 6.678,
+      "train_steps_per_second": 0.052
+    }
+  ],
+  "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": 1.9354956165367726e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}