diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8687 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1235,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004048582995951417,
+      "grad_norm": 7.454889426657729,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.1726,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 7.556283493630047,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.1681,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 7.728127459862402,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.1491,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 7.880099643923719,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.1988,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 7.031924338238101,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.1578,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 6.921223576459166,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.1129,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 5.980295425937195,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 1.1213,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 5.226030246671039,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 1.083,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 3.109196154146172,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 1.0261,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.7023285948939644,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.9854,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.4695380147947033,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.9944,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 4.615188222830743,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 1.0198,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 5.0336705242336395,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 1.0363,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 4.724329160721173,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 1.0069,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 4.189598553416865,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.9379,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 3.273540184795265,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.9415,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 3.0554613773607917,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.9182,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.7691329121057513,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.9074,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 2.5200502093826205,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.8974,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 2.2768616719391015,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.8861,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.689403867491459,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.865,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.5890840246246187,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.8329,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.959838121831129,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.8676,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.8008186359744591,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.8395,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 1.7730827989807325,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.8227,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.155023935449256,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.8575,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 1.2027392495075266,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.8097,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 1.2558558754778868,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.8128,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 1.0144215140662438,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.7983,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 1.0755373257648224,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.8138,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.9720998889433852,
+      "learning_rate": 1e-05,
+      "loss": 0.7892,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 1.1242200979124373,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.7838,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.9664098138931704,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.8126,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.8118095166493328,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.8019,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.9051789090115774,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.772,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.9010023184803124,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.7847,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.8125265390937744,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.7682,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.846608520782126,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.7655,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.9125784847580609,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.7919,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.8209404479056757,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.7712,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.7366421853413032,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.7819,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.8540489231141889,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.7603,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.7645911719304255,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.7644,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.7520404240864489,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.7517,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.6896560549970475,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.7397,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.7874733576929183,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.7276,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.7244725593026731,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.7871,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.7444486657832518,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.7503,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.7069269870025613,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.763,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.7805375211821368,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.7425,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.7927368945530565,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.7764,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.7388617033893469,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.747,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.7092078336012967,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.7553,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.7414807449688536,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.7514,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.8360147289904136,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.7531,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.8324159858953343,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.737,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.8954905773756376,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.7573,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.8664051278039926,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.777,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.766517393709195,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.7301,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.81724036255086,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.7708,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.8574568445274805,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.7624,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.8252372894131362,
+      "learning_rate": 2e-05,
+      "loss": 0.7208,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.90152380966248,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.738,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.8133690163848929,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.7527,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.9380110618172063,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.7252,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.8931861955645777,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.7657,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.9013666743886262,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.7579,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.9150596587340517,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.7652,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.7764579219799037,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.7317,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 1.0447573520767575,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.7416,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.8469595274495791,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.7417,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.8884848504290066,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.7805,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.7603430223057974,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.7203,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.9699443015351574,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.7435,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.835571258709978,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.7353,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 1.1003222411154148,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.7646,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.8176647254351551,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.7238,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.7628077483634728,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.7366,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.9277530538493254,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.7242,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.9771440791492284,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.756,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.7851524612001493,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.742,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 1.0813948221336724,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.7494,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.7241118914411758,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.7098,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.780254040984099,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.7367,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.8104647142531276,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.7482,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 1.0953283016912472,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.7573,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 5.934431546473419,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.726,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 1.5773414564992818,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.7593,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 1.0855110562760921,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.7351,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.9662942814734907,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.7293,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.0569022751071495,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.7386,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.8366216575065158,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.7443,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 1.4546321174795265,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.7267,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.9087947847684602,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.7395,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 1.7250212514540073,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.7529,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 1.0199920168444183,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.7324,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 1.6126057263963438,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.7313,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 7.350569146620389,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.7199,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 1.0988473891701043,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.7482,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.2732763430290455,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.7497,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 1.1676600080231005,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.7692,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 1.2063493968380168,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.734,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.9836247533473498,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.7375,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 1.235874136940937,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.7415,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.9101201658002491,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.7131,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.1363770708993106,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.7538,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 0.8448942065719263,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.7412,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 1.072478438420134,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.7235,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 1.018751465783303,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.707,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 0.7819249644348771,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.7274,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.9435488963720449,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.7175,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.7476341768554403,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.7277,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.002240418526346,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.7254,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.8207575979798073,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.7369,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.9076358887185025,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.7639,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 1.0800339134366699,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.7432,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.9544961687134588,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.7515,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.9407299639265377,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.7265,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.8991726472256936,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.7407,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.8446048737499414,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.7423,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.7937743930667347,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.7133,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.7897764470459405,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.7432,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.7147082867450952,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.7052,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.8172238096291465,
+      "learning_rate": 4e-05,
+      "loss": 0.7718,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.7452857595332925,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.7307,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 0.7559032027913444,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.748,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.8426407329754019,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.7325,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.88563320794796,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.7271,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 0.9914547878140324,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.7215,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.7133539069646969,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.7319,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.8831579942147648,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.7501,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.7355998841038047,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.7617,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.8330369429094817,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.7717,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 1.291936203433456,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.7392,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.7804457487099835,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.7511,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.969902422039863,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.7589,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.7322496859543882,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.7464,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.7878502410800878,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.7113,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.7169032247645498,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.7549,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.9153077053659708,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.7319,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.8767183962414844,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.746,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.9396989438800243,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.7422,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 1.2338041246929363,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.7292,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.8707376903651884,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.7042,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 1.13611305942649,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.7592,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.6985162420955117,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.6956,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.9092864934050781,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.7074,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 1.0546559158395037,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.7327,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.8506871897074162,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.7765,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 1.0000369687940913,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.7122,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.9239854756014538,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.7299,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.8667881417167732,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.7196,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.8525587939619286,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.7227,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.8517627569567406,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.7444,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.700062437995987,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.7361,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.8339808201683738,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.7457,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.7023040151709463,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.7442,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.8687016255172384,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.7562,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.6280392963026954,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.7125,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.6848353084409962,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.7193,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.7741374306132609,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.707,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.6477869181505791,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.7032,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.7340055547855625,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.7235,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.6334963903833289,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.7119,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.6543345728725187,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.7292,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.8174323961385286,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.7288,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.7161757424066894,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.7541,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.7504704347158987,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.7189,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.832151869043479,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.7351,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.7592803710217869,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.7358,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.7751613452723001,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.7216,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.8538798057106199,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.7362,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.7989205846324064,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.7019,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.8823164395902494,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.7461,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.6330356851325206,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.7447,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.6776620291345613,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.7532,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.5868139441217585,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.701,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.7361065283281699,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.741,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.6812204722088507,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.7031,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.7957505567907356,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.7321,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.8497955698353346,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.7238,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.6543335234324337,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.739,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.8915431446757472,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.7346,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.5921168030571476,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.7579,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.6572553012735893,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.709,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.5924217288119579,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.7334,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.6849427295060206,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.715,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.695020383476883,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.6983,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.7028068756365717,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.7042,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.7030949537683134,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.7188,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.6681695421397799,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.7286,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.688057101566295,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.7356,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.6986143583677252,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.7126,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.6960446455872176,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.7313,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.7201095419481844,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.6939,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.6828606313677814,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.7133,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.6065916121944235,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.7292,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.7021064149271137,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.7013,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.6653640757947389,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.7269,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.6636570891304056,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.691,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.654406293197285,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.728,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.6210160147859848,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.7088,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.657968463958554,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.7334,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.646898688243669,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.716,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.6538580665161694,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.7324,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.633929638490613,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.7437,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.6956831320188602,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.694,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.6286772992439593,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.6943,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.7075661962279893,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.6954,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.629887930203558,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.702,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.8156371436838773,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.7346,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.6542998238894917,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.6992,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.6727347312113469,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.7134,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.6387223803571399,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.7221,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.5488767202030257,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.7016,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.5741726586147848,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.7312,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.6282822123927814,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.7053,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.6193824362884464,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.7264,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.6024049084850582,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.6801,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.6090667446799868,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.709,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.6427346933225297,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.7297,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.6924166446532752,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.7126,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.6242579501415373,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.7185,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.6876315414916022,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.7299,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.584318072736566,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.7206,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.639730242943339,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.6889,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.6943358139481162,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.7201,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.759645117818188,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.6992,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.6654381842072482,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.7118,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.7007991633379679,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.7222,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.7102669713101889,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.711,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.7118872256990117,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.7089,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.6192319455598642,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.7198,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.6700757873163334,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.7365,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.6597045975005748,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.7228,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.605934171165389,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.7442,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.7592861337328721,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.7186,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.5559326371588483,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.7102,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.7610618523470297,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.731,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.5137934255306319,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.7095,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.7348092966278277,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.7205,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.6825745078879735,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.698,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.6313810103530532,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.7184,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.7141549310228709,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.7122,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.6737845585697636,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.7084,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.6797004387395865,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.6796,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6527337461189842,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.7043,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.7415363250766693,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.5837,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.6015614727172367,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.5822,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.7381134298326952,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.5498,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.7080527831977786,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.5615,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.7463246211106564,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.523,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.6833673302809224,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.5368,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.7523130415655831,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.5534,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.5752496665776489,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.5221,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.6497994137225057,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.5643,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.671401713759617,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.5659,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.56580914894785,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.546,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.6305421890158724,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.5554,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.6915897771965782,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.5538,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.5969844587727262,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.5628,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.6866004192203462,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.5334,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.5976761548525404,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.5291,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.6390226013113499,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.5376,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.5975490635700509,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.5605,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.5929777326644718,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.5344,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.5894191279877744,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.5406,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.592263195014907,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.5487,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.6215720814929531,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.5185,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.628374616311505,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.5543,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.6622845352874919,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.5309,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.594918078954156,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.5629,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.604312054799509,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.5656,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.7149776053459727,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.5568,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.6541061796334416,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.5222,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.6606599272809313,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.539,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.8079773314281606,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.5589,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.6528081608400045,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.5426,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.7022777125648515,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.5519,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.7068180511252317,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.5336,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.5714261507026394,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.5573,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.6812207886441315,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.5345,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.7441550143447562,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.5544,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.5813723665150397,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.5443,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.7032426029436176,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.5539,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.618904629891256,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.5481,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.6901284996414192,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.5799,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.7578981019399347,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.5474,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.7028765869868537,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.5569,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.7155025933637756,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.5414,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.6744992720304196,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.5265,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.7070593177281985,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.558,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.6617291964607351,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.5592,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.7650409889217411,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.5586,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.6703048503731371,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.5434,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.6408074837336767,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.5312,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.7983262842455874,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.5453,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.560305203395298,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.5173,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.8765557912525573,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.571,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.5908345069415392,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.5605,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.7200225474257682,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.53,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.6404155234278113,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.5548,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.5583860959244044,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.5439,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.6221825313528179,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.5494,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.630933904429803,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.5452,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.5552650628406777,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.5508,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.7455644457067707,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.547,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.6502285483236555,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.5412,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.6432084962546095,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.5681,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.6615152005840033,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.5624,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.6193973039725048,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.5379,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.7763923565525893,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.5342,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.6804376779952765,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.5546,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.7663239619056817,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.5535,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.736855351494969,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.5453,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.5906325887557353,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.5727,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.6050356540427285,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.558,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.602612951015401,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.5373,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.6308201535373565,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.5437,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.5759702207819418,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.5542,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.8162477651919275,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.5525,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.5701836053753849,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.5593,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.6870875920771838,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.5665,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.5086069648888691,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.5273,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.6954979582217966,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.5534,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.6395176815748319,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.5118,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.6151839078424179,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.5282,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.6104568827900563,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.5566,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.6678984353442663,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.5271,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.5693155252886369,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.5583,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.650003454286105,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.5405,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.595983598559985,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.5415,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.6092064324223806,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.5467,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.5811737990780306,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.5415,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.6086248986865653,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.5255,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.6105411483872506,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.5788,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.6352766250297284,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.5626,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.611860958866652,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.5691,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.6094109346511715,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.5304,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.6475212996424631,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.5419,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.6540716399044977,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.5422,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.6676573998077369,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.5554,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.6149968477852698,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.5742,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.6673083570812869,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.5435,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.613361733237636,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.5417,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.6616825194127279,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.532,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.5804294347361222,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.5475,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.671865352307447,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.5817,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.6341005267122721,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.5755,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.6335122956065911,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.5471,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.7135056345776168,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.5692,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.5534218531939525,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.5688,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.6334273692749232,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.5437,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.6389423770900483,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.5438,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.6473858536911002,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.5398,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.6191189579281257,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.5409,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.7948430643157435,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.5563,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.6421164798671769,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5399,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.7881480492295089,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.5773,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.6687097238790799,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.5639,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.6888290276138881,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.562,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.674020984184678,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.5654,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.6284356974793214,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.5711,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.6986533271558577,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.5721,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.5245733427016464,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.5447,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.5484076634880412,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.543,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.5654506433094877,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.536,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.6244421886846149,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.5602,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.5557446308276278,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.5419,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.6385397101208404,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.5308,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.6201039647380648,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.554,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.5591680680892864,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.5411,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.6802844634557895,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.5553,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.5918088722403265,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.5289,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.7038969223546906,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.554,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.6591269516536962,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.5366,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.6995200089926897,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.5551,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.6609807294359593,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.5402,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.7458746044308535,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.5479,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.7785865771789131,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.5686,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.6317175528004376,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.562,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.6860508263303973,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.5667,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.6181899665215203,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.5589,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.665751173130227,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.5258,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.6280535258093207,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.555,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.7326377786806418,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.5402,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.5327575538599719,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.5191,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.7505858015064305,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.5786,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.5370088379102237,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.539,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.6434374235488057,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.5629,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.5994270153707124,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.5607,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.6239236292405855,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.5632,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.6281552176224234,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.5495,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.6461524920968427,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.5456,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.6096863110043778,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.5462,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.695319842092044,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.5849,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.5970326707230009,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.556,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.6430524019880118,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.5112,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.6334688115263374,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.5333,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.6916624459392493,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.5282,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.6356764052267342,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.5408,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.6754056178868303,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.567,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.5907251252882877,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.5498,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.6333493041186574,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.5345,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.5865746089196598,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.5276,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.612662262378851,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.5517,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.57675393623253,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.5648,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.5524644269513597,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.5671,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.6473927967520497,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.5716,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.5133886139852795,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.5385,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.5584497177969927,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.558,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.5967006931176193,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.5474,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.5807703132012968,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.5469,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.5375015466059412,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.5443,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.6126014386966496,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.5781,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.5448944890453543,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.547,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.5800657401538091,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.551,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.555217779141729,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.5554,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.602611409581035,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.5705,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.5155528130185477,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.5547,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.6044468931429654,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.5697,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.6036590905360657,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.5533,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.6175978320906118,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.5422,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.6748933547829647,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.5395,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.535925085806829,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.5328,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.7091437847001897,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5506,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.6017979140387502,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.5569,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.628429295203149,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.5274,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.5708392049641764,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.5375,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.5667766805006015,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.5479,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.5782943436523474,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.5601,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.5946613855576903,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.5482,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.508120340112973,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.5368,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.5263037538947121,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.553,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.5234173825125371,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.5196,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.5202903696650285,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.5585,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.5604190610735366,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.5412,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.5113739498247271,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.5551,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.5147096851082532,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.5641,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.5046701760319894,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.5368,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.5147457112703564,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.565,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.5482086334204254,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.5496,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.5125434835063153,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.5317,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.5274069890234635,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.5349,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.512739503804595,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.5315,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.485887096359889,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.5436,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.5044623632723702,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.5768,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.5359968241055624,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.5251,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.5291607772139247,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.5576,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.5533988001387792,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.544,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.5893738485077585,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.5446,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.5273331876601901,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.5524,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.642264976716721,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.5606,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.482616383333094,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.5718,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.5597732760489146,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.5362,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.5468810055201337,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.5484,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.4974161391280859,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.5648,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.5145378784532924,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.5562,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.5516832050153421,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.5277,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.6039393517213616,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.5605,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.5618334669288353,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.5322,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.545011635342119,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.5518,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.6097786811957976,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.5517,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.5952044610696057,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.5393,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.6384592455481898,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.5487,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.554114653070345,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.5291,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.5754875353048134,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.5481,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.5467965985092087,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.562,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.5643181847703312,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.5405,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.545266390979697,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.5268,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.5476860641944866,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.5454,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.5942140181717757,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.554,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.610111222273267,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.5491,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.545130377852326,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.557,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.7409867023372009,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.5272,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.5884054134703255,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.566,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.8122084788225696,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.5605,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.538936265853653,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.5726,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.6711788144944449,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.5412,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.5257419873399986,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.57,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.5647053290976441,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.554,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.58357301001217,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.5236,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.5721947949838725,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.5667,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.5055338221683349,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.5474,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.5112860105110894,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.5403,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.5282927175611861,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.5642,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.5113576578604748,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.5347,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.5606479554836722,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.5354,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.6137526214583768,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.5666,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.5862608528239538,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.5688,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.5281844440717022,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.5461,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.6348421668159999,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.5567,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.512505902136275,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.5682,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.6707300149153278,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.5319,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.8528547679943367,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.352,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.6292530516307692,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.3506,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.9578247494617638,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.3457,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.6015965195621304,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.3457,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.7356104097240195,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.3564,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.6188919438843287,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.3414,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.6265442908200591,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.3637,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.6666496293392694,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.3488,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.6594330807288064,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.3148,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.6465254613411922,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.3781,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.6464683020565957,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.3529,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.6405012216908608,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.349,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.7185943243446985,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.3558,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.6053072372948999,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.3511,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.5805693685915706,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.3477,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.6091545016501174,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.3537,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.5013480618914894,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.3676,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.5765431268719341,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.3575,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.5375023135557007,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.3764,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.538609982453298,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.3574,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.5870387234777462,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.3648,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.5954218122527503,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.3326,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.615746902981407,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.3643,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.5669935541106053,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.373,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.5296140442305677,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.3331,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.5679961818591175,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.3364,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.5023247312850206,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.3599,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.5553984088974631,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.3524,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.539945092307488,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.3539,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.6278777418351896,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.3529,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.5031922714414606,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.3461,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.6228644612173944,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.3496,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.535025702872418,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.3504,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.5494837156224354,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.342,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.5035206696827287,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.3554,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.6099765097446729,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.3332,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.5268548874218437,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.3492,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.5309892311818502,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.3433,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.5512195375309215,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.3666,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.5044058688204048,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.3551,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.5461711153847015,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.3785,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.5182211305885436,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.3515,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.5722997692375141,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.3672,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.4787768623187289,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.3551,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.6262978186992101,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.3803,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.4970596055093539,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.3677,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.5926501216278955,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.3192,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.5213608254769921,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.3692,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.4852480001757321,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.3424,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.5659078731745072,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.3657,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.5016647825864979,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.3461,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.582131911020813,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.363,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.47396653075307604,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.3419,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.52232271092274,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.3408,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.47170281499422856,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.347,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.531275429799263,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.3399,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.517836019252842,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.3616,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.46996468890436716,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.3931,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.5305702570582271,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.3507,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.4899252122422905,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.3536,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.5015565388778443,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.3624,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.4867574817342976,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.3676,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.4710115246067571,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.3599,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.48182558748056065,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.3398,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.4878925780792303,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.3467,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.4817121645362814,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.3586,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.49786179214899634,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.3432,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.49147814181493577,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.3522,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.49720109865757917,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.3662,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.461775701702239,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.3458,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.45943516145333657,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.3495,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.49114117083492587,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.3508,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.48028551941358877,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.3666,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.4495212927693714,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.369,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.47590991857180787,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.3727,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.46623030743644345,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.3593,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.49249238504883047,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.3435,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.485765151149387,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.3483,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.49548726650072056,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.3579,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.47856812584138225,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.3438,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.542405738101963,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.353,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.502989392467007,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.3706,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.49036818517321235,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.352,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.5198462720500407,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.3596,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.5085259229341418,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.3336,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.500950113490336,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.3484,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.4875392002235684,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.3764,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.5202300020990951,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.3612,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.48255301951388896,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.376,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.5057829682672558,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.3524,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.5534508170152965,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.3431,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.4879780157898661,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.3864,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.4830366742506408,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.3498,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.5450423090124019,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.356,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.5279857702742214,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.3523,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.4809308707727922,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.3809,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.5289490487719453,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.3711,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.47040932449496164,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.3478,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.5384202421360518,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.3684,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.5021577243070288,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.3695,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.4872027809221477,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.364,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.49972216614311055,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.3956,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.4773020211555297,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.3563,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.48313353015531113,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.3498,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.5144326031232184,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.3629,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.482446293812447,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.351,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.4822406374778427,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.3587,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.4900932453229376,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.377,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.4703147400631094,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.3391,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.47005526361396904,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.3695,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.5205129061653525,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.3691,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.46561809121607656,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.3602,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.522159268199779,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.3659,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.4677862823374327,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.3613,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.4927825386806599,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.3841,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.4829561957990391,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.3486,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.47323815173278305,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.3613,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.47332438582293046,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.3488,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.5135970728885948,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.3523,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.46398318014800394,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.3589,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.48023455593567355,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.3387,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.5525229382369932,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.3838,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.49414407577221814,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.3422,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.5168638208443109,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.3405,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.5334466860764975,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.3621,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.5261243583957049,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.3379,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.5419932451257636,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.3731,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.46873874462802323,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.3739,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.5473946402972354,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.3651,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.5212509278513097,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.3746,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.5422918698549882,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.368,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.5721513495075348,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.3627,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.47153406265421105,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.3539,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.571332277276124,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.3477,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.4807680036124873,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.3689,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.4840981913293731,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.3641,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.45162170949205976,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.3402,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.4940150965434929,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.3587,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.477993945856564,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.3582,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.5077579643029546,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.3541,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.4712767662830117,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.3639,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.4797519319156723,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.3491,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.472124695419673,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.3752,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.500249426226829,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.3762,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.48760025964819564,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.3612,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.4841514861935699,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.3494,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.5284808305883554,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.3707,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.45397453017935163,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.3622,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.5378652974158598,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.3725,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.4892996892026256,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.3681,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.48846081481898224,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.3755,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.4905376196294419,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.3637,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.5227414927629768,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.3502,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.5171957324238273,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.3572,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.5232858663943076,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.3267,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.5209730034134405,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.3486,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.4844120187720201,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.376,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.4820992851083534,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.3387,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.4878712659230944,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.3474,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.5191195778310033,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.3644,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.4815289234684776,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.3543,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.49562861590538715,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.36,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.4860028424784276,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.3721,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.5003013984851483,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.3396,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.45965057943846527,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.3655,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.536253927873961,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.3415,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.492561033571633,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.3729,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.4315583603641344,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.3461,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.4980409332137073,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.3409,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.488488054199095,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.3535,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.44519853154655026,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.3857,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.5351969605060185,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.3508,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.45336505021360246,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.3565,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.4840273955698136,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.3387,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.5189087942152372,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.3735,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.4545126218901034,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.3569,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.5084771117484814,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.3578,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.5292531397347597,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.3515,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.4544711356799578,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.3863,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.5143913763218684,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.3806,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.46759127221439595,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.3468,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.46598680576951,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.3609,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.4986277511342936,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.3502,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.5001102363998018,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.37,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.45431831293444,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.3468,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.5166614910164914,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.3553,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.46627750329622375,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.348,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.4732059883811104,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.3621,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.5226708156768378,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.3759,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.48326255786777667,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.3522,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.4699320631001541,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.3704,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.44788825557561246,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.3533,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.4873447477612037,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.3645,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.4877395356060039,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.3428,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.4813269970614065,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.3537,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.47719870181361573,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.3602,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.467589856330923,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.3584,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.46434873231162166,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.3624,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.47050878936382384,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.362,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.49049347922271136,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.3303,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.48043837473259643,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.3679,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.48585721401573856,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.3566,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.47733054036952594,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.3789,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.5197472332732598,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.3498,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.49413755228887324,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.3441,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.5325131304069431,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.3455,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.45415503774857796,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.3384,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.46440814951121073,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.3269,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.48560851475239974,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.3662,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.49218926088474235,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.3341,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.47506195279276886,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.3455,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.4853725806774603,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.3576,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.5112054345703686,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.3504,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.4590568220501768,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.369,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.4908126138730769,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.3691,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.491420531285308,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.3578,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.47953331440073016,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.3508,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.4625408439015722,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.3578,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.4918669310125897,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.3607,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.45325847322156526,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.3412,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.46193233587314264,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.3394,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.4766920170226187,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.3911,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.4589904903640558,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.365,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.4668277735603302,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.3654,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.44715284113566617,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.3452,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.4581288215056413,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.384,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.4755444855607783,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.3512,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.4374459332731288,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.3686,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.47952566179594347,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.3509,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.4527178060978881,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.3714,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.45995844668336583,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.3306,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.47275965846353446,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.3625,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.43372428988166045,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.3458,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.45623419604851073,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.3368,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.46387502161583977,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.3442,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.4378564200401261,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.354,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.43943749071391147,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.3254,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.47365410563973315,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.3764,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.4754877646598367,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.3487,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.43276922834280673,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.3744,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.43909187160353763,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.3583,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.46538539840495663,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.3676,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.44558028022378904,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.376,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.47707336485108354,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.3549,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.4519930951121466,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.3517,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.4440960920060742,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.3414,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.4398775280357564,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.3571,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.640153366986614,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2311,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.4155476017674851,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.2359,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.47635762251455577,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.2186,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.8087172734239296,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.2491,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.5735508651672419,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.2163,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.4534956842766407,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.2391,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.5916100032790922,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.236,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.5299268112698897,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.2399,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.41152478057703296,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.2403,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.41472624117945883,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.2439,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.4860968256252654,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.227,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.4537833013677253,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.245,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.4377844983419034,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.199,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.3943018676726728,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.2198,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.41301804406807713,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.1971,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.4337920664861115,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2216,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.41561754414402735,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.2092,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.39155557433084826,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.2375,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.432343944914742,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.1964,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.44260304589823274,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.2278,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.42002804353045975,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2069,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.40825727169223,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.2204,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.39978922609245116,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.2095,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.38755105458603545,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.2458,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.405154892558861,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.2358,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.4256167964633526,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.2156,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.3783108690273355,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.2127,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.3909685687995559,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2132,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.3961458407232893,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.2138,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.3907096344666283,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.214,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.37660630985496535,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.2137,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.37691176267732723,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.2117,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.38435517774732286,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.2282,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.391305311841858,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.2104,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.3783526758125713,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.2164,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.376393754717016,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.1928,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.3987772914000561,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.2405,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.4676670967179273,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.2209,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.3754552807058376,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.2396,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.3718678388970287,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2193,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.38229733127171467,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.2253,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.3952134838548383,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.2147,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.40399364506324165,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.2082,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.3831837725301163,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.2196,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.3924643984103387,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.2486,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.3958169461489923,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.2272,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.38029218142862437,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.2315,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.3960745008782395,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.2012,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.39179011556224796,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.2259,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.391044089269105,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.2188,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.39255912012724725,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.2154,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.3885905783022072,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.2218,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.38370844741829935,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2161,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.37331753362148673,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.229,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.37327257570793015,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.2361,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.3923924042344419,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.2267,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.3717485064345846,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.2291,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.3762893792186839,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2203,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.4008977138418235,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.2456,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.39186988662775557,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2389,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.3920807697755107,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.2141,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.37948110524623974,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.2446,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.3972792243439334,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.2335,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.3852486802241308,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.2135,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.37994046609681165,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.2256,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.39696221194946674,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.2343,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.3821263972374047,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2227,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.38007643882346,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.2045,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.39139993263429496,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.2284,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.39196515641895757,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.1945,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.3845683733232929,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2042,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.3898342573980226,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.2118,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.3822508014213828,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2272,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.38281413809238796,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.2143,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.3866264270433031,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.2101,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.3699442673631117,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.1931,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.38594955518760876,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.243,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.3778163241637843,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.1944,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.3741720561701385,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.2163,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.37403514476216393,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2101,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.3783955713466114,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2321,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.36877656681187077,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.207,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.3496155574566921,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.2187,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.36844196531539114,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.23,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.39031999383557836,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.1973,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.363453638571451,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.2284,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.3777207607686767,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2132,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.3556556680282712,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.2257,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.3676580294267114,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.2112,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.3732226619688841,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.2037,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.35483856245991574,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.2333,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.3830204355689591,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.2237,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.36790958888319697,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2102,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.36494870071945323,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.2248,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.36034124727857125,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.221,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.36807799716272277,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.2068,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.38687886872322064,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.2356,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.39737673766737897,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.2157,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.37818164453653624,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.2121,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.37817544997050634,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.2017,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.3849783596118707,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.209,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.3932528395626269,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.2066,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.3603436059825685,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.2192,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.37407988955983884,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.2211,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.39702653951737615,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.216,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.38649975378728263,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.2228,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.37754863899470287,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.2299,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.38170421641527846,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.2292,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.37169031051090834,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.2144,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.3670705025269254,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.23,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.3742338145151368,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.1993,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.3730466617421573,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.2052,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.3607289491227044,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2085,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.3741314975117656,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.2168,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.3719585427573466,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.2015,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.36512758534506307,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.2038,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.37142617325655003,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.2294,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.38467953148011663,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.2545,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.38959186990138583,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.2164,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.395585044913964,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.2199,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.3573211971453291,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2486,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.3889277459365515,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.2128,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.37156756250319,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.2383,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.36882580957648,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2218,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.3706349669525345,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.2029,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.3723690314187093,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.234,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.3640938784681141,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.2393,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.3765734771902473,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.2135,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.3699991058087453,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.2319,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.36677672731385846,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.2002,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.36431621263870956,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.2153,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.37097344462722576,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.2334,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.3519264893797521,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.2303,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.36741328887662134,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2309,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.35557889901592793,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.2044,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.36193401706205147,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.2083,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.36606300785811363,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2225,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.36932266723394996,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.2375,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.3700848371608422,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.1924,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.37071035181982115,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.2335,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.37195092565854965,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.2086,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.3709000866833078,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.1908,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.36609546318530034,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.2115,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.3629347213805701,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.2028,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.3466949327797586,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.2275,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.36392916540195824,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.209,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.36265674573224327,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.2221,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.35408197556647814,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.1973,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.36239378845915277,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.2078,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.3508296150607566,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.2086,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.3577469457527409,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.1989,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.3569786719074932,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.2284,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.36488379125876685,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2316,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.3639474015829333,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.2134,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.36727057160936893,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.228,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.3632188318449293,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.2189,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.3887899553155703,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.2134,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.38306000166677623,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.2203,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.35681524179485347,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.24,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.354577537822667,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.1994,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.36203162470201733,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.2165,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.3937395066826415,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.2122,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.36840466820884443,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.2127,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.3561376224263651,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.2193,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.3703271578037372,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2141,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.37940733822197265,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.2024,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.35770925153882516,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2142,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.3596220831859992,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.2158,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.37384052152719555,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.2149,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.38140892596197246,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.2031,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.3644190034601311,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2072,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.3639081136038976,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2214,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.36666924380514826,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.2051,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.3694471848145104,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.2069,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.3631057020459373,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.1952,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.36477917892761935,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.2364,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.35326154079481725,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.2154,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.35922270994853006,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.2189,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.3573478448892426,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2086,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.3517224291782662,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.2217,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.3392795327672267,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.2414,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.3648032761543932,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.2184,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.3845180399776809,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2078,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.3729818966471594,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.1934,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.37922188912770566,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.2084,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 1.1747444991406821,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.2197,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.3668150526507194,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.2042,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.3613971158925735,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.2186,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.36707731789815756,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.2177,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.3718811106368119,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.2202,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.36411459145170616,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.2214,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.34751443453980985,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.2271,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.3661042717368965,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.1916,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.3564085178269922,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.2285,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.3569854942135001,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.2414,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.371595518286346,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.2099,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.371633954714102,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.1852,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.3668870443117562,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.208,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.3658624612706471,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.1987,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.3492925586419697,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.2539,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.3643551315481829,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.2312,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.3544430253751141,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.2264,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.3758386870742103,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.2009,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.3623877872748303,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.2004,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.3466986867265676,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.1881,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.35850740197910635,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2162,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.3794088478291877,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.2234,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.35593246510457854,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.2142,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.35938409440002606,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.2107,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.3696959845992932,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.1956,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.37338422456773485,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.2195,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.3557738537848673,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.2272,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.3494125932896799,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.2267,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.3596480438970406,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2213,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.3547055024682695,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.1854,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.36407562411288735,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.224,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.35396805328921493,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.2415,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.35978759429315355,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.2167,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.35438372405283336,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.2276,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.3660331417548499,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.2365,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.3401163996965438,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.2085,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.3357200864687539,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.2145,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.37060029657831245,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.1916,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.3545237242813778,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.2325,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.34060520547409406,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.2336,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.3275917227985635,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.2333,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.3457824104970346,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.23,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.3670285552400866,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.2413,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.3430553526995863,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.2266,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.3526363333195927,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.2171,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.36250669750430975,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.2258,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.34353628421402016,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2262,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.3368560665680922,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.2109,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.34052734278470226,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.2249,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.3788598267079223,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.2198,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.3476235067254503,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.2365,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.3559225544381174,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2128,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.35748857402719114,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2311,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.34525122350612236,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.2101,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.3690202737925482,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.2187,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.3494073000408654,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2242,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.356590868433465,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.2156,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.3684254890837076,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.2074,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.35558213018859025,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2246,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.3726359815987295,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.2233,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.34712909250840435,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.2417,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.3598192984935332,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.1894,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.4466419763582759,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.1608,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.41087598196033837,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.1762,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.3296120265404383,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.1824,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.2737042487665465,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.1632,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.2885192198038537,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.1478,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.39772087871374134,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.1828,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.36649462993743465,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.1538,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.42171417281356166,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.1612,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.35225819994587065,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.1729,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.3074908095723963,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.1522,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.2897624744413009,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.1453,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.31176996155048115,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.1381,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.299986899933267,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.1796,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.29460763480868,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.1585,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.29014362597173965,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.1634,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.28088592230585796,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.1584,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.2827729792173495,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.1276,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.27167981826796533,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.1397,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.28074671084313485,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.1614,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.3036652669397379,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.164,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.27835504816728085,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.1481,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.28542372630375423,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.1648,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.2706109653594635,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.1654,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.26529230158220257,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.1766,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.2756533527164116,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.1652,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.2781889638179517,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.165,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.2815334244255796,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.1701,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.268644790360771,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.1583,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.28665142285117456,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.1595,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.27767404313513566,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.1652,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.2639515093412289,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.13,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.2754027654047642,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.1439,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.28370467263126276,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.18,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.2677899703310951,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.142,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.26875660917312644,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.1463,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.2843620122556652,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.1734,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.2633508531786127,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.1472,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.25714110031896553,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.1664,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.2787539845042577,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.1515,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.2750643572086227,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.153,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.2515122114754006,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.1541,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.26905077688941603,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.1473,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.26622189012083813,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.1411,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.26143127637643804,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.1779,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.27649754949879257,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.1565,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.27709139232683055,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.1494,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.27021239246858997,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.1598,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.27611233113338307,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.1677,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.26558735497616204,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.1691,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.2682873526360369,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.1438,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.2796290881549062,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1408,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.2656698236375484,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.1529,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.23784001221992837,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.1354,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.27820437008034493,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.1362,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.27446581359165606,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.1425,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.281815708660148,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.1591,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.2738186784183145,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.1473,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.27157308949024456,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.1473,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.2580159619322298,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.1747,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.29084826646529255,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.1761,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.26033432125082434,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.1378,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.29638050507467034,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.1461,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.26397982597615455,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.1644,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.2672971458333459,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.1547,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.2731671605420955,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.1518,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.27740387406134576,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.1728,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.26038572801706517,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.1555,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.2650695863100288,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.1534,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.2566208707583657,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.1301,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.25313829322311304,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.145,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.25398863096054636,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.1262,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.27828769642431445,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.1411,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.2582020884981146,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.15,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.28289558687568467,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.1738,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.2708151262249744,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.1472,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.2833114374992537,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.1402,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.2674343454916604,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.1412,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.2590003976611068,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.1562,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.26356822942661695,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.1525,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.2589611433389551,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.1358,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.2646040136558168,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.1648,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.26588573019252165,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.1523,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.25585012961621006,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.1709,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.2727223818691643,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.1297,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.2519621754346925,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.1568,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.2680297388206725,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.133,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.27242063840376984,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.1483,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.26025252059955156,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.1698,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.25063524753067423,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.1203,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.2778316414055335,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.1463,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.27260349327111283,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.169,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.2690862079208882,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.1517,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.2758105789591853,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.1593,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.2553564784080644,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.1652,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.266239891182543,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.1461,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.2832039108444528,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.1595,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.2795224474240657,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.1481,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.25757166719892827,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.1523,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.28270209944471425,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.1542,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.23906778775001541,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.1311,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.26350740132382383,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.1553,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.2741600319160243,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.1345,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.27398777842659744,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.1478,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.2836139420431483,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.1872,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.267481389985896,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.1531,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.2772443357342693,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.1682,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.2797762616444277,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.1421,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.2560675829720179,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.1684,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.26323281935603676,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.1477,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.2755617892878139,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.1212,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.24948455287222754,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.1792,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.2701368552836132,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.1571,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.2560653095999954,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.1504,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.26478594152469037,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.1574,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.25726605680364156,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.1481,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.26338156872466634,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.1516,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.2606956619105573,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.1452,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.26743444961958274,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.1386,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.2769508326504015,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.1586,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.25702312360443264,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.1548,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.2672366090480976,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.1545,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.27150790299364785,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.1436,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.26791763938197566,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.1815,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.25651671434577294,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.1326,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.2789936620427434,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.1624,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.27237776458761526,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.1542,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.26576679090796307,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.1527,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.2620441711174661,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.152,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.25808588710438146,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.1531,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.2713782328780971,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.1756,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.2557697284002739,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.1459,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.2725418287700313,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.1599,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.2740471582521143,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.1516,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.27035047898772413,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.1493,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.2671246207003109,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.1507,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.2661518331729954,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.1696,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.2568902894928704,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.1298,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.2633045142746357,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.1473,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.27414991706065933,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.1409,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.26769294485532014,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.1393,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.2781453381221924,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.1745,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.26849493817663295,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.1438,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.24940287104809353,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.1479,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.27006524612709654,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.1205,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.28978351708261646,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.1519,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.2581172493997515,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.1249,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.2744979778669244,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.1593,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.2791892505061453,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.1647,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.2679808065814096,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.1409,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.25380489503318726,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.1416,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.2576886292914907,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.1229,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.2707807886350941,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.1592,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.2509048073735879,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.1665,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.2513387175181226,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.1605,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.2646882973120831,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.1581,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.2705439601081349,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.1676,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.26945303468892323,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.1435,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.27494957774434103,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.1686,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.2581661628256064,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.1328,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.2888241382030328,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.174,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.25008261324698633,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.1271,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.2750193713714384,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.152,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.2512868559314059,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.1236,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.25533245689030964,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.1539,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.2594595516686247,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.1465,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.26066420051729716,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.1364,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.27000414797404,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.1471,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.2763548895527103,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.1858,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.25995021437832055,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.1676,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.2683420097971914,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.1678,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.24745093211480654,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.1386,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.26160621353586316,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.139,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.2542764691638737,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.1536,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.25742679407052066,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1407,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.2634511146200828,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.1749,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.26376740922963954,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.1445,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.26280547660585024,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.1488,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.2615200768781053,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.1464,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.2406002165880025,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.1315,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.27630251945677453,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.1841,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.2583988374278461,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.171,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.25773143161701156,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.1615,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.2635491512438516,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.1306,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.261081964337847,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.1435,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.27737570296268604,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.1564,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.2563083937686209,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.1566,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.2616725465048692,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.1729,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.2753340802221566,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.1594,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.2751767011689632,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.1448,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.26920333094289906,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.1496,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.2654075423473389,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.1898,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.2520789934208493,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.1512,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.2660838245541023,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.1286,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.2791913578598037,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.1398,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.25564068815512725,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.1674,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.25960492709093735,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.148,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.2601889539267207,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.1468,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.26860690126172115,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.1293,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.26622303396745045,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.1626,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.2770773435307469,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.1683,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.2681980140723361,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.1337,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.2702221056766108,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.1457,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.2521886526476772,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.1525,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.2633671362888063,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.1569,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.2639708241818931,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.1391,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.25783614332200444,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.1584,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.2635219086613323,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.1592,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.2647928089382436,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.1636,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.27373601033123746,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.1573,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.2531721278477169,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.1573,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.26729866801929714,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.1709,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.2670109143984944,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.1624,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.25973711458799864,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.1316,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.26503802368769946,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.1334,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.25976832016224743,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.1321,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.2621784711386421,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.1717,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.26774025349946995,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.1392,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.27005774052272163,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.1522,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.253636886689584,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.1747,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.2641741214265016,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.1448,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.26456859275204575,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.1632,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.26323848920691467,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.1415,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.27762446471820956,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.1546,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.2535005904983007,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.1293,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.2658509198788494,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.158,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.25854081431482073,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.1428,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.28049082877618187,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.1832,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.26228005045526925,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.1515,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.27061860739403126,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.1501,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.2475941208757802,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.1578,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.26688874167743365,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.159,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.27143999258282686,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.1818,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.26860101076375453,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.1589,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.2788663715075823,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.1635,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.27502117581912255,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.1708,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.25891086859989254,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.1644,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.2634828275460281,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.1362,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.25863960122455565,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.1601,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.2746198252498032,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.1408,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.27264551941062,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.1365,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.27122796055587706,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.1565,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.2539152363474782,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.206,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.25654913573877275,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.1407,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.2603434384687774,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.1354,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.27035207965256175,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.1424,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.25972115646815563,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.1729,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.2479527515659695,
+      "learning_rate": 0.0,
+      "loss": 0.1538,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 1.484981906855428e+18,
+      "train_loss": 0.4082452781890568,
+      "train_runtime": 19492.8427,
+      "train_samples_per_second": 8.106,
+      "train_steps_per_second": 0.063
+    }
+  ],
+  "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": 1.484981906855428e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}