diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8687 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1235,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004048582995951417,
+      "grad_norm": 6.644279178915882,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.0453,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.660618197197681,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.0505,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.588813020010519,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.0505,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.528319702581216,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.0555,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.1674875192039496,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.0345,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 6.131938043546551,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.0544,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.710846522961344,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 1.0025,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.345396651722498,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.9895,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.9178090677271653,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.9725,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.6753094474612054,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.957,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.551427944969323,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.9544,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 4.025409348008579,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.939,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.233503482693255,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.9355,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.928664937935533,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.9213,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.686532407031982,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.9219,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 2.7514067779116846,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.8802,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.478862690134577,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.8583,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 1.9836247690613302,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.8469,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.4185382149961347,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.8249,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.039157423087331,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.8149,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.1789449832876038,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.8049,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.3055882366276084,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.8062,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.0430263914864433,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.7834,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 0.883502385316422,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.7717,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 1.0987395701733462,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.7617,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.9592626244123573,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.7727,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 0.7287445254969134,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.745,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.9233982852839271,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.7506,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.866844586266404,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.7343,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.8399311895451818,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.7394,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.7745038463150135,
+      "learning_rate": 1e-05,
+      "loss": 0.7318,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.768294528160851,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.7399,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.7213376876830613,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.7216,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.7663141159997094,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.7201,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.6653463816370688,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.7145,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.5940054575137701,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.7205,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.5907581197731459,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.7147,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.5935524164228826,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.7094,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.6349140930689,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.7115,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.558733507943147,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.7013,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.6238508339730252,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.7001,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.5499906960705063,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.7012,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.5263718382612358,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.7017,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.6743455369035659,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.6939,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.4770048375550748,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.6791,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.5095347917565992,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.6985,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.5596410998091869,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.6919,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.512463547261227,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.6981,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.5762432186888099,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.6789,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.7651972191208034,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.6873,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.6924550749049182,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.6892,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.6016645527060835,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.6798,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.5538975370437724,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.6769,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.6280654829366895,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.6815,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.7281532549079759,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.6824,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.6311363464691543,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.6737,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.6592483226658035,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.6804,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.6950277575066879,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.6744,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.8818445773881636,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.664,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.8870267458286897,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.6886,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.838326222086243,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.6646,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.5991253917117679,
+      "learning_rate": 2e-05,
+      "loss": 0.6658,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.8778005060858262,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.6686,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 1.0066572136152632,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.6768,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.9442745383931374,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.6489,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.9198737998517754,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.6594,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 1.1029954781017632,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.6559,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.7211240438327966,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.6608,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.6756238494701883,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.6422,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.7558488147112351,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.6599,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 1.0064652497364552,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.6598,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 1.1740377314822825,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.6505,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.6480956682889627,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.6479,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 1.3215526780912923,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.6591,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 1.109459277322029,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.6566,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.8780047060235355,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.6466,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 1.4610672173017065,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.6509,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.8125669252374204,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.6484,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 1.7835804265278434,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.6573,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 1.3468924569900385,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.6565,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 1.7535059987246657,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.6584,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 1.9057813454454324,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.6612,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 1.1290570638591688,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.6487,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 1.3437618767811266,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.6475,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 1.2541032500182845,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.6635,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.9817064249376433,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.649,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 1.6165828804246782,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.6517,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.9029216150760554,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.6439,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 1.3933565133394787,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.6412,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 1.0057764457704943,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.6596,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.1257003940760488,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.629,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.9869671029687375,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.6517,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 1.3803060639223272,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.647,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.9634052683285093,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.6477,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 1.390907878327017,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.6461,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 1.1094953726992975,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.6475,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 1.155226305498302,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.6418,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 1.1582515816878929,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.6327,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 1.0330569089868598,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.6306,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.218295855326828,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.6484,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 1.139383545055863,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.6303,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.8878043700339012,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.6518,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.9028949830450792,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.6462,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.9359482212793011,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.6348,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 1.256993154385998,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.6474,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.3444055577343528,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.6471,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 1.0316571616730148,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.6538,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 1.3113142722700304,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.6477,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.8610872384559258,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.6379,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 1.5746651614869789,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.6412,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.64275674935368,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.6499,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.9296135377140105,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.6457,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.5880571814361026,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.6355,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 1.451571146756334,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.6426,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 1.049979041443754,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.6367,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 2.0724858269989834,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.6336,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 1.0890398545688507,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.6479,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 2.28284853163501,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.646,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 1.774994013537396,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.64,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 1.8088342796843342,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.6359,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 1.6401077534116404,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.6486,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 1.6838940216444975,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.6259,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 1.3110328128620337,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.6416,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 1.7797081478075254,
+      "learning_rate": 4e-05,
+      "loss": 0.637,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 1.4530986448055285,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.6186,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 1.7373553561149775,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.633,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 1.436591063738894,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.6395,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.4106075758388663,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.626,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 1.3047977907495114,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.6277,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 1.2646229984012702,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.6231,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.9418622006139367,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.6328,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.9610916388567909,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.6351,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.9795583717201587,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.6271,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 1.5362156341420696,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.64,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 1.0511299955928757,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.641,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 1.1685356148000106,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.6136,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 1.1783133961090546,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.6321,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 1.113903638879973,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.6304,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 1.0519863191776377,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.6349,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 1.0386557473567475,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.6247,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 1.1584928560767862,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.6405,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.7247626782056692,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.625,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.7249174520810667,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.6305,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.877311920163919,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.6265,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.6958146827562957,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.6397,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.7308095843673169,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.6053,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.7791638486118695,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.6027,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.6651727896018046,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.6313,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.8907866505803324,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.6352,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.7693846911859984,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.6251,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.6103765450077763,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.6068,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.7682026977718596,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.6368,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.8738551693885728,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.6193,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 1.0686619151322838,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.6245,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 1.2616821128805153,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.6287,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.6117071576240567,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.6267,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.6559689517776095,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.6374,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 1.0585223199589464,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.6406,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 1.2934049335824958,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.6266,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.7104494840843393,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.6309,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.9517015899369179,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.6148,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 1.2198140943707976,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.6294,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.7629937996779949,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.6267,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.8798840558681256,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.6353,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.9553193496872417,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.6106,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.8316378458111402,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.6118,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.7255928212152443,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.6314,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.6641440678004681,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.6288,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.7561850130533885,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.6312,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.9587311790986834,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.6166,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.7893293532870499,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.6105,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.8250902129572306,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.6179,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.7998649891673635,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.6218,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.5098472140088725,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.6307,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.7152730917536478,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.6107,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.5101079085390692,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.6198,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.5206433513362172,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.6053,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.5388884149068982,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.6105,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.5300800421074492,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.6212,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.5811994874775896,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.6053,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.5358956744788286,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.6088,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.4592659863510017,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.6157,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.5228479735989797,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.6159,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.488698391348089,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.62,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.5660136917234865,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.6111,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.6167485327033599,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.5993,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.536316014276257,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.5993,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.7758715462900541,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.6044,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.7129797089774318,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.6141,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.5690306166215001,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.5978,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.73670554812312,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.6029,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.6888391975139231,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.6125,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.5964967443997315,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.6127,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.5753977279715803,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.6013,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.6054940343020011,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.6011,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.7119178102463964,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.6148,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.5182703062287777,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.5977,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.6280666745861233,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.6165,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.7678929724090783,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.6034,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.8017394744030375,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.6169,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.8035341153329643,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.6098,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.765743746376143,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.6132,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.6275864837632175,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.6128,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.5293784108423736,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.6091,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.6913854833904995,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.6031,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.8580306193088436,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.6103,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.6909512120696261,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.6149,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.5818945667648737,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.6048,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.6281660000688745,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.6195,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.6717755202926563,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.6057,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.7256189831575979,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.612,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.5715796119143937,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.5987,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.4712330859754628,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.5977,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.5054337909198059,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.6079,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.5938132427354355,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.6114,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.5472375877675302,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.6029,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.5344065092040523,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.6034,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.4622059448127343,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.5884,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.4354135873376074,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.6039,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.547417062173511,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.5995,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.5666509936683791,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.5961,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.5275497544526428,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.6043,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.5630477217856776,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.5903,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.5125747252849693,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.6062,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.5089079056172859,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.5987,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.4842740416724658,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.6061,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.5696010244073215,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.5997,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.6423328625722938,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.5948,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.5829237537912094,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.5988,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.4885436262712049,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.6013,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.5665433801210567,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.5938,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.5552175904898833,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.5921,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.5776991724063905,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.5933,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.4722165791781639,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.5998,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.5274769448928928,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.6009,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.42133401039028057,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.5916,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.45180849899061687,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.6004,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.4173777902724058,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.5974,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.5122675081967469,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.601,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.43548850133499195,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.6008,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.5238617961980937,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.6006,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.6094263595765882,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.5907,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.5483641839799974,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.5906,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.47326847635409197,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.6073,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.49653500927604216,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.5839,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.6390695678203948,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.6069,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5230310634989168,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.6021,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.5996973653914693,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.5672,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.6633840064970885,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.5642,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.7395870514880833,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.5613,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.6334109626513504,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.564,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.4844309568182706,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.5451,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.5688037253314784,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.5584,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.6149426901605988,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.5524,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.8249030313024623,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.5303,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.7040047057197637,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.5528,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.6760044210644877,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.5567,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.8971223250735665,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.5426,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.5698041209739483,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.5544,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.5861460850711591,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.5567,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.6384790026869763,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.5458,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.6583853759522699,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.5637,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.7125995512908061,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.5463,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.5214963120908258,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.5457,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.4894776373064181,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.5652,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.5152212967677737,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.5464,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.4716106169247167,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.5494,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.4789799113749134,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.5405,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.4202166761170006,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.5457,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.4783508755263806,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.5523,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.5766669873894382,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.5465,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.6581045938686508,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.5491,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.4596495921703166,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.5492,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.4895799677627537,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.5445,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.40849508271213175,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.5456,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.4436756562602175,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.5369,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.4906402317412793,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.5426,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.5653008062561473,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.5522,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.6042303979393374,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.5471,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.7040897618860568,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.5396,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.6881020830955791,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.5507,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.7120622304081756,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.5532,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.45106148792916156,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.5544,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.574660694230039,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.5561,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.6607549719794341,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.5512,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.624025866837171,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.5476,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.6199504785081619,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.5425,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.49759633201406117,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.5587,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.5544569269703341,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.5551,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.5153489634781586,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.5609,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.571600656136921,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.5329,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.49689465783136705,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.555,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.6040256653506153,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.5498,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.5916300016829027,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.5457,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.5171012779617895,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.5579,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.7390583538232579,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.5435,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.5235671569141701,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.5554,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.4852408205376249,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.5355,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.5939585979911635,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.5424,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.4726669504961403,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.5677,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.490862714759545,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.5507,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.5769710584381224,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.5514,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.47293554139888144,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.5563,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.5610325964573735,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.5551,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.5944064929137497,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.5602,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.47236668705426493,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.5401,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.5674849372780645,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.5434,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.5878850999709666,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.5452,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.4954866903564486,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.5376,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.5725012781635953,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.566,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.5352678615521882,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.5398,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.5031297956539871,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.5417,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.6146720872713393,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.5601,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.557595731650302,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.5537,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.4638371591109581,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.5517,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.4734888665312576,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.5386,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.49017579618522866,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.5337,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.5000155854168788,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.5483,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.5243242814366735,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.5441,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.4395963008113188,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.5495,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.4897608313545995,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.5534,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.545208947660422,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.5431,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.5639648846385711,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.5607,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.5463095731021912,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.5388,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.46351797443801984,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.5479,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.4976546684727901,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.5412,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.4165470011094131,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.54,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.4633657645675872,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.5444,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.5633385365747823,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.5313,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.4644396196806399,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.5466,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.6122258329317359,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.5578,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.4925174973235737,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.5513,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.657370007447965,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.5605,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.5091697464264187,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.5427,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.4262419059581584,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.5399,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.4681330324915583,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.5445,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.45465972709314245,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.5432,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.44168199159299254,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.5477,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.5144597306538352,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.5409,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.40214643427013413,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.5363,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.5741375667123828,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.5376,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.5014497669872663,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.5692,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.40962023042147494,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.5522,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.5763914922131427,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.5484,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.48200438603435636,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.5463,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.48443095293675487,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.5555,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.48312701421289594,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.5372,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.48381357587419344,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.5365,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.5623623984069153,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.5451,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.49482639433890513,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.5509,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.5377518584403501,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.5301,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.6667992532896702,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.5364,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.49491970788256867,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.5426,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.7155392699216947,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.5394,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.5898033338750854,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.5488,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.5417155995204117,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.5341,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.5611336800050591,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.549,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.5111576996210427,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5406,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.6307456268383987,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.5444,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.5179663113631297,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.5523,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.5167653079857267,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.5345,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.5187317062012451,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.5353,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.5872597379988399,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.5432,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.5506054085985105,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.5387,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.5465210489616645,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.5583,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.4804790685398863,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.5464,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.4587645235626149,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.5474,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.5316142289541731,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.5476,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.4511058550090798,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.5409,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.3981458424686601,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.5326,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.42234965100230343,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.5338,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.4227602392960576,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.5476,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.47330541016410693,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.5451,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.4630226209314684,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.5519,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.38023326427576853,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.5374,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.4087854186999823,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.5279,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.40643367504547406,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.5348,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.40645904252446097,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.5364,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.43285668317091186,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.5379,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.48320337629443155,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.5321,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.3734106195317183,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.5255,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.46844540035206467,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.556,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.4633015367171415,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.539,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.5253957517412741,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.5397,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.43169474408899444,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.5453,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.44640155048530494,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.5388,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.5277838714946654,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.5382,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.3882588254651016,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.5414,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.4466803843494803,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.5301,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.40576489603394295,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.5394,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.3840896422933079,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.5438,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.3758890515708499,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.54,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.3719880734989641,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.5402,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.39779458325545164,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.5525,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.4222612937497095,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.5281,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.3612450104923946,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.5479,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.43195565043444,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.5395,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.3962485064341307,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.5304,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.4108564153586002,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.5301,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.4315233208245491,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.5564,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.45474416596285516,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.5324,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.40272737547056864,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.5469,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.3995232977335872,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.5465,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.477526196231165,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.5365,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.3916424835765797,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.5328,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.45466536250195067,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.5404,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.4928925896012322,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.5311,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.3479274744868383,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.5457,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.5285516319535377,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.5417,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.42359147883986825,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.5332,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.41241225058332576,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.5373,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.38277155992616885,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.5425,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.4339956728742657,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.5391,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.4536426387646384,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.5493,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.35596795158484806,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.5394,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.4990945510559225,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.5307,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.46911803102108773,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.5352,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.35348397983151486,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.5362,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.3916822774483654,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.5393,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.4392363399747517,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.5364,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.39131009114719206,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.5501,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.45538323422217836,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.5422,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.3316413169141353,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.525,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.4498494004736659,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.5215,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.4231017194500246,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.5354,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.4531969354573154,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5253,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.49630294424747856,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.5364,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.5121522015430743,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.5433,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.6211061113613041,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.5397,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.34688495820106263,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.5277,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.4796921322686133,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.5322,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.5191458559694259,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.5477,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.4372854874604615,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.5042,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.44113531655462296,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.5357,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.46053406010753534,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.532,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.4319448454321786,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.5374,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.4125521306073699,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.5372,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.5022092616705462,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.539,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.4339206822999285,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.5346,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.4675971325779938,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.5374,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.43786857975648885,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.5315,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.46583600566587413,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.5431,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.39779452863371456,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.5248,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.3786162980122791,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.5214,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.41485412747968275,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.543,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.37830630040749674,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.5323,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.35689762629315813,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.5199,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.33993849615259103,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.5402,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.44422102348545023,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.5512,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.4536461357578301,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.5361,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.36374681645012874,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.5315,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.45330955767214476,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.5434,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.45369120367052035,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.5434,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.39503068517199125,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.5358,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.5251807883894896,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.5307,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.45679710797628886,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.5398,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.37091646857069926,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.5257,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.3447442720805886,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.5331,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.3933486080466037,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.5351,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.34300344369885327,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.5366,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.33606922150298557,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.5367,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.4363795480118051,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.531,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.34249200764550386,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.5119,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.4402827519832939,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.5334,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.34491420379108034,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.5407,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.39656655899075755,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.5293,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.3474799206273165,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.5303,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.32924844143527815,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.5401,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.3874980024261998,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.5362,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.33986722329620733,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.5144,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.3492581490493562,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.53,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.3695848841397429,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.5435,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.3316219961373921,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.5459,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.36500948392364774,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.541,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.36513190479844965,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.5437,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.4095989832284771,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.5506,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.4089230087440065,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.5424,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.36386155345153925,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.529,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.3892035582054759,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.5353,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.4171165040712449,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.5404,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.38044419671629975,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.5317,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.41849926516126246,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.5401,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.3942621775059082,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.5197,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.4106762079799909,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.5294,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.3992488446805513,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.5243,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.42743228300152586,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.537,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.3811252199410639,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.5335,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.38408840339377026,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.5365,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.3012404643294816,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.5281,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.4013009217206677,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.5248,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.3325585073615987,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.5426,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.38059964008077585,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.541,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.351504546156416,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.5318,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.36414255304922094,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.5221,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.5232350121262543,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.4597,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 1.3277940695758852,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.4639,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.7779608893185085,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.4782,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.4145921099112085,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.4742,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.6752865330916001,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.4649,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.6302561889108218,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.4576,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.5449886048491142,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.4673,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.645045816100685,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.4636,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.5871953425667866,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.468,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.5288300918564713,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.4548,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.5903193953238077,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.452,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.42467450933195416,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.4614,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.47367419663018534,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.4585,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.4581428069465376,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.4432,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.42226211601322333,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.4466,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.45652915184185056,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.4643,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.4180442174491647,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.4635,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.3981467331298054,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.4693,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.42746201010690615,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.4494,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.3618202619104069,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.4537,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.4103623766175304,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.4388,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.3956075131376683,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.4514,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.411472589429134,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.457,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.39728891294541513,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.4667,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.413734022426781,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.4704,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.42848062890708466,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.4574,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.4239320677291933,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.4597,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.45043576134157653,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.476,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.39625136259653954,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.4541,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.4486344322434649,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.4668,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.4784505098487413,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.4435,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.3689624414523661,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.444,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.4724847389112914,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.4592,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.39232041340768525,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.4467,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.4055515736269868,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.472,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.4293861913763718,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.4757,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.42280305116991945,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.4632,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.4225256003983785,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.4569,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.41538236634857745,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.4664,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.3712262430456189,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.4616,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.4912752524509773,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.4535,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.3361218453522477,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.4501,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.44174811564502314,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.4714,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.34167929083486537,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.4478,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.40416147583250805,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.4496,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.3310959874850001,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.4429,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.3670887037847941,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.4689,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.33648156976017507,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.467,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.39829396158636476,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.4695,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.36351749424112273,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.4627,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.39186748867795224,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.4612,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.39778290778959174,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.4625,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.37985110333691396,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.4544,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.44488273580161847,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.4457,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.386682953259662,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.446,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.4462063215491627,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.4601,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.3897885492900284,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.4652,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.3939045498009429,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.4494,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.4430059277548974,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.4543,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.3734869681183456,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.4433,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.4147532202840976,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.4491,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.3904852752960415,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.4564,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.3950278167506924,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.4637,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.3321873559446057,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.4566,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.396356605822097,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.4522,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.4326366427242158,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.4651,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.3850050004663423,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.4505,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.435009128249092,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.4551,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.4040129215334809,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.4695,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.3684718814134234,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.4527,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.4116032603169294,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.4611,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.34007582093953964,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.4313,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.40272024395147826,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.452,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.33478126995633267,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.4426,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.38362501488659934,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.4514,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.3207419378224572,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.4531,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.3340869065130413,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.4361,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.39699932690168027,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.4761,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.2956439489901709,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.4607,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.4201284334512277,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.4509,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.3135823992094322,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.458,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.375933523666462,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.4633,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.34167996170966486,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.4512,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.3412121029025413,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.4662,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.3720055775230108,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.4663,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.3840321858959443,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.4824,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.3869683113759484,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.4572,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.41333633233340045,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.4517,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.4037884368007713,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.4623,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.4166080991584015,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.4535,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.39291808516399634,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.455,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.4566385012834712,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.466,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.36076927183002805,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.4721,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.44656413137305645,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.4478,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.36304387583349457,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.459,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.37708378988581376,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.4593,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.33157055107808936,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.4632,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3336837079136447,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.4586,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.3408253214355332,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.4563,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.35282433458446033,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.4551,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.3913086763659134,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.4519,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.37115412295242345,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.4746,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.4221179772916654,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.468,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.41195953373667865,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.4683,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.4945300225132774,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.4538,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.43281144889666084,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.4512,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.3195424013883972,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.4564,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.3647975525180299,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.4602,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.3206326934578989,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.4635,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.339392703397709,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.4568,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.34856896453994657,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.4505,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.3385935616270814,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.452,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.30193251791676384,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.4594,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.33605734841524926,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.4341,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.3257666031190517,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.4471,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.3366882456844802,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.4552,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.34146685077990063,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.4673,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.33374560761725325,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.4462,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.3543143087110726,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.4553,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.33844295325419355,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.445,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.3797307656799684,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.4692,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.3382765466376028,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.4642,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.36440867569399377,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.4619,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.3133423694899667,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.4571,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.2912415803811802,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.4523,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.33712633985083396,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.4655,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.32912460022324264,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.4587,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.3301773241710167,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.4746,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.34907409411748086,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.4636,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.3214672607249311,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.471,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.3448162913734872,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.4454,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.30947500917296195,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.4658,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.3260756927408186,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.4547,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.3132002381764203,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.4609,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.3317926878245557,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.454,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.3302219520357828,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.4663,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.33972212457901163,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.4543,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3101765285866312,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.4439,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.34474928706604774,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.463,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.31971201024938073,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.4543,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.3143794064266405,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.4572,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.35264721372857116,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.4578,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.29176902171280433,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.4445,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.3596871370109816,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.4443,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.3006679463192175,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.4574,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.3027315888012854,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.4482,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.3186728092782899,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.4581,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.30339142673339375,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.4514,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.31492037876273726,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.4668,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.2993952959614342,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.4521,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.30516640870968126,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.4622,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.2990738422325984,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.4503,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.34464760939289707,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.4553,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.2962911156305182,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.4509,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.30458897398325174,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.4583,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.3183852764679703,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.4556,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.3313765348014383,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.4424,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.33223436113752386,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.457,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.34011415234591996,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.4575,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.3447274463870287,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.4468,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.3318798385509554,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.4425,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.3296049563879899,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.4628,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.3109226602290027,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.4636,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.34689502451310894,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.4339,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.31190358823207415,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.4491,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.32338088025081335,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.4624,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.31618402776152543,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.4786,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.315967108000882,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.4476,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.3024999279741678,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.4678,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.3126687411881154,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.4604,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.31712971420675307,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.4541,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.3144562871858621,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.4604,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.3172551794539477,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.4542,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.3223043587656376,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.4707,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.31358369230993316,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.4552,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.37516884597446953,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.473,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.30247677952676294,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.4608,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.37026879104884436,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.4646,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.3015312198865216,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.4521,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.3464944875032031,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.4569,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.34227926283156757,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.4583,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.33682219491544946,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.454,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.3254624072643625,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.4427,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.3463569944674373,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.4489,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.3063669127722146,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.4492,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.31372995231092043,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.4533,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.36599225197599783,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.4462,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.305981538935038,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.4483,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.3367494914475327,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.4493,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.32133790946702423,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.4593,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.30232991483532634,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.4325,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.3324228490937353,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.4649,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.31026958517890174,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.4489,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.3057200876032334,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.4393,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.2990342167460102,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.4584,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.2817166448135425,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.4474,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.27619935648501137,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.4522,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.2834710467811884,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.4475,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.2949685102544608,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.4437,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.32172241556021086,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.4478,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.3015269013940913,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.462,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.31512271170964945,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.4602,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.3153381555352568,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.4394,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.29887290827373314,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.4574,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.31210945235982107,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.4541,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.3094472870786131,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.4508,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.290173016925453,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.4379,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.29338082611814453,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.4549,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.29404198860755204,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.4354,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.2876559948454653,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.452,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.2979238482664578,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.4548,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.27181870891910725,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.4562,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.31196954376904923,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.4538,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.30385602246168775,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.4641,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.32076308796025843,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.4389,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.33201736864355424,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.447,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.35040961408230453,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.4591,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.2975840374886278,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.4618,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.3453379117744565,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.4582,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.3322194219693047,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.4487,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.35185809731234774,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.45,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.3517732213421286,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.456,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.3131862066147641,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.4275,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.2968220066701647,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.4543,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.33605287851451865,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.4329,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.306036521424426,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.458,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.3519275826531444,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.4404,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.35403497843800913,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.4808,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.3669520077662926,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.4466,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.3649035533237724,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.4333,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.37070818096696373,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.4535,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.3433997429712202,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.4425,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.3755758282972286,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.4366,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.3160563366519908,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.4512,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.3265285712933192,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.452,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.33419260187767635,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.4366,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.3006314137859374,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.4315,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.29672594974571426,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.4687,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.2988621173434511,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.4416,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.28794300197896044,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.4502,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.3225905479633513,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.4487,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.3397870502421634,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.4595,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.279440450182005,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.4349,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.3251659778361387,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.4475,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.3213324668659527,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.4569,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.30363989284023535,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.4522,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.3338803407372244,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.4412,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.5264536459581134,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.3773,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.7579351442222891,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.3768,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.43769277016016384,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.3787,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.4398765116369617,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.3741,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.42259618233706353,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.3757,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.36429769701614795,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.3766,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.3581129594325574,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.3852,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.3694917910516005,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.3737,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.33256074513123673,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.3618,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.3548609761591031,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.392,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.3503830395597747,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.3635,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.3128535339955102,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.3635,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.3400140565034549,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.3771,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.3158781784305377,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.3782,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.31145448235887474,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.3774,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.33872184973145775,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.3594,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.30710511703523463,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.3715,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.3635848941407083,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.3624,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.336510156612485,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.3875,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.3313851592498203,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.3766,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.3125573019496866,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.3677,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.32205775794118363,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.3572,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.3157623380008486,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.3531,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.33496125779409863,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.3929,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.2752865080555921,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.3742,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.39295491227550106,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.3598,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.3115756963773282,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.3851,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.3362170023234101,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.3554,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.3022728683500289,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.3739,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.36834729101202346,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.3564,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.30583922994083507,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.3728,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.32347037422213126,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.3707,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.3149941618026953,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.3568,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.31592486682571486,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.3704,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.3198341596935681,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.3825,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.2994878428552517,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.3798,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.29504117262145996,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.3795,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.3273460694382824,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.3704,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.2932354891953905,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.3567,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.3044374587006039,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.3852,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.3231115869938127,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.3557,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.30100107327208386,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.3651,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.31838282554236785,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.3587,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.3017764107423124,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.3696,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.29275875976490834,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.3717,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.3237656871988336,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.3669,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.2918191062976081,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.3669,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.3416234980396444,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.3577,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.3199849242887322,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.3585,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.3200345660463951,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.3556,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.29795867752755023,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.3773,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.3239807153424229,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.3673,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.2988660237659406,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.3585,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.31274676288314435,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.3616,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.28209764787388064,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.3647,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.2837516924243957,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.3602,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.2795483543836006,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.3836,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.27745833510771967,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.3673,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.2879450791944728,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.3525,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.2876764445675511,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.3825,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.2896494585054712,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.3643,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.2888296523026882,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.3599,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.2821588346951458,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.3694,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.32628229029279754,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.3753,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.29343957485484345,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.3513,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.3255727879533873,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.3721,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.3157764829751568,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.3558,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.28016493886221927,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.3554,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.3529322120224148,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.3572,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.2936681190084139,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.3689,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.3143441714312162,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.3749,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.2955276048657247,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.3794,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.2904794947634555,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.3812,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.28954574077724066,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.3577,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.296566668013867,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.3752,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.28485634889208034,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.3585,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.3133681850665042,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.378,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.30283080695378084,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.3793,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.29387280631184476,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.372,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.29448961657583017,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.3706,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.3020091855914785,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.3805,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.2794168900233411,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.3586,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.28744655103845246,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.4058,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.3270430258344182,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.3695,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.2574991058596014,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.3659,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.34639490848783283,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.377,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.2857270152828784,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.358,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.3144906391851912,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.3825,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.31157784398417254,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.3601,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.3497211436822914,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.3741,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.31487586713431853,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.3723,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.29825291568301837,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.3603,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.3173439741314425,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.3711,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.28812315221468543,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.367,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.2970964090036489,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.3619,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.30073876652216625,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.3743,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.30928096005200506,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.3788,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.2837015095471583,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.3666,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.2870185692730357,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.3737,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.30133680894751247,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.3587,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.280615500630527,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.384,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.333653420743374,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.376,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.32298396468196333,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.366,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.3121550090956952,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.3918,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.28126567685873244,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.3594,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.29030151976689117,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.3772,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.29898438016764334,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.366,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.2710238950235004,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.3742,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.3026821510521035,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.3539,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.3131079237596804,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.3852,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.29758277252036075,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.3677,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.3116761597908294,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.3429,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.3565488021781788,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.3549,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.31306551551463446,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.3676,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.28415441945218145,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.3957,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.30861985780345397,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.358,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.26869196609743284,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.3749,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.2955545205444327,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.3711,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.30891782910687715,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.3759,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.3050310773549002,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.3848,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.31242038645795134,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.3679,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.3436346131883073,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.3893,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.2989825412153237,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.3679,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.29125385309959606,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.3725,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.31405221245923165,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.3575,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.2983205604459634,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.3766,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.2769584459164951,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.3751,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.2914557322296731,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.3567,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.2921493415261811,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.3557,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.278756039151302,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.3803,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.293677259499646,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.3626,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.31662442783779166,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.3937,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.27241381732381775,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.3848,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.295902065451502,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.3826,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.2921080865852624,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.3689,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.2821994408258389,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.3582,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.28038527801380164,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.3677,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.2907054978075269,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.3889,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.27518894454372583,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.3799,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.2729157809832527,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.3747,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.27864187865516354,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.3856,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.26739155167744927,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.3657,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.25708891544650087,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.3711,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.28381254024198405,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.3733,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.2720077256616206,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.3686,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.2625941937658976,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.3591,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.2682163135614963,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.3593,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.2886538840337985,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.3646,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.2686498488220876,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.3626,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.2618124999798667,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.356,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.2729534417439116,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.3657,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.3038458069005821,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.3658,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.26295873827072647,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.3869,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.2756676478550577,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.365,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.287617819829148,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.3606,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.2931321200476769,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.3675,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.27244127348573566,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.3259,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.3251652107701318,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.3738,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.3017709875524237,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.3389,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.3058465584277325,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.3664,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.2710122652214794,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.3575,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.2926305704468557,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.3667,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.30140032245976955,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.3462,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.2754238448930551,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.3727,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.33527241262076135,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.3518,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.2842769179242986,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.3628,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.27410587232883005,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.3662,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.32707519455579603,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.3537,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.25665711464855895,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.3593,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.2960070895449358,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.3723,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.2890360715746594,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.3699,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.25818058787290266,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.3706,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.277944928134189,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.3365,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.27890701823048164,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.3655,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.2794557333688538,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.3683,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.2810163707633457,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.3622,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.3061992758059518,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.3579,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.28070088112658664,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.3793,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.27099194566215623,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.3735,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.3337111618735937,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.3644,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.2567799334618748,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.3676,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.293245036940227,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.3588,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.2873389644416834,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.372,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.3062073877015776,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.3683,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.3037160911374242,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.3661,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.26170712285979053,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.3647,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.26398491185354206,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.3583,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.2685778266469691,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.3717,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.25030431124196956,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.364,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.2550587165275028,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.3769,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.25615255153642535,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.3759,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.2581239572004828,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.3651,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.24963570589026102,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.3596,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.25999593352947664,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.3547,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.3088075462460981,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.349,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.25840855379204103,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.3504,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.2566939367870555,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.3731,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.2632765850967536,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.3668,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.25269397205903427,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.3582,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.25553194688408604,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.3619,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.27667932814869217,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.3791,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.27172889543595047,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.3754,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.2825290904574483,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.3404,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.2650099858015969,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.3559,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.2512478559823303,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.3781,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.25937479953368403,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.366,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.2700477263376578,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.3808,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.2661712235964209,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.3574,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.24762897871515077,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.3617,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.26522506296264065,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.3623,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.2706008150677547,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.3645,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.2747809223381473,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.3707,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.2574761126136214,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.3599,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.26063366015918044,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.38,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.2621028511557982,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.3665,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.2689620486258455,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.3633,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.25401306593798517,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.3763,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.25293317881458977,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.3504,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.24764021381582987,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.3665,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.2480345967178225,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.3672,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.2531608611888458,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.3453,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.26370046619333104,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.3591,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.2428858729790407,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.365,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.2761782199335176,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.3812,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.28774218183090444,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.3697,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.2577349518626733,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.3382,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.26134158211066877,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.3749,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.2667580928106077,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.3651,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.27169602777996804,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.3522,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.24701904769732738,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.3827,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.24578966548136408,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.3635,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.26291745858628757,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.3592,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.24916670683210856,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.3586,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.23574038896145308,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.3652,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.2586242495112813,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.3711,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.25712355704282536,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.3556,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.26398728665727256,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.337,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.2437801368989198,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.3483,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.2543736528942834,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.3711,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.2672946142542543,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.3737,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.26632012566535174,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.3645,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.24986164901067884,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.3365,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.2530574845021584,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.3548,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.26260909689318723,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.3544,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.24225146722417748,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.3686,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.2490432462842458,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.3637,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.24246541582392672,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.3484,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.4789881723863431,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.3173,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.3038213496197589,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.3089,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.46631121380159896,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.3098,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.2850909397314578,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.2965,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.3011564850623842,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.3011,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.4079713665959092,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.3036,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.3013585349257194,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.3146,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.3040258445660262,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.2884,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.33716509016007795,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.2938,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.26053972200795084,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.3162,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.28498547667857943,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.2891,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.2793277089295537,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.3172,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.2934076968538314,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.3005,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.25869748160198536,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.3005,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.2768508279947955,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.2996,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.2940519402262035,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.2945,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.2736930500783596,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.294,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.2675937435878702,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.3123,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.2738210270979187,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.3149,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.26737055597264725,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.309,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.2541867879363145,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.3121,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.28572104672543563,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.3086,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.2748012893501311,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.3194,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.2484946168643767,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.3163,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.25931143889138253,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.3091,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.26614141693509263,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.319,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.27714700432769346,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.2789,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.25143657380112033,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.3088,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.2535639789060187,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.2808,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.2680638739654191,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.3015,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.29313633649054327,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.3201,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.2613804980486943,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.3232,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.25291524568288504,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.3207,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.24814612322820737,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.3392,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.30603033771551574,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.3006,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.2534654240356485,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.3225,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.25978593166666114,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.3132,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.2653812810518497,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.3109,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.2820305139298887,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.2906,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.25446570054954226,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.3106,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.2652643649423743,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.2979,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.2839294787403395,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.2777,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.24492591839703742,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.3118,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.2536705202043279,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.2922,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.2746659241425904,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.3128,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.2634112777661462,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.292,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.2527630192108976,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.3052,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.25740114423499416,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.2964,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.2665032572120207,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.2854,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.28035450755427765,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.3035,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.2591257027121836,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.2823,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.2356375006729914,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.3037,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.28814287266838684,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.2989,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.2971692388013822,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.32,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.27228447745124296,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.3109,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.25707082680913657,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.2742,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.2686016995365054,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.3055,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.25068994096328157,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.3241,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.25968595624712654,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.2975,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.25023278814053984,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.3111,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.25224276573124965,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.2993,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.2575849263711171,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.3019,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.27296102039191084,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.322,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.2588811449213892,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.3028,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.2606290863079884,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.3076,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.25628536174695465,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.3061,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.269956623642217,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.2978,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.26392970356508455,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.2961,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.2553410093941273,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.3003,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.26171995903508766,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.2805,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.24907198165988964,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.2986,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.25186419727738885,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.3023,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.2561322245750696,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.2936,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.2651069716194942,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.2666,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.24073872266315086,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.318,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.2528736377280061,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.2929,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.2530568534226102,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.29,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.24961347117398056,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.2893,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.2517596447547836,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.3091,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.24510162928908796,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.3077,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.24915538970305706,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.2975,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.2499297185915091,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.2875,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.255465863992542,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.302,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.259716422240975,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.2826,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.2631614567054725,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.2919,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.2585577551380169,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.296,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.2595104406603113,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.2904,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.24920652231567106,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.2831,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.256925068649078,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.2972,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.24471497031263104,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.2974,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.25914944275796203,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.3044,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.24255183646563092,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.2958,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.24174051733275448,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.3174,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.25831024294275423,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.3237,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.24248400439274168,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.3123,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.2582770632415517,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.2982,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.2468221273134114,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.3018,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.24532540831568106,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.3033,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.25263882838971036,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.2975,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.24780272672253373,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.3092,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.2629448805004677,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.3162,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.25253865544128584,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.3188,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.23901500469622727,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.274,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.25093367048731274,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.2877,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.25579775045668535,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.3191,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.24926754694514952,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.2979,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.24651463046445268,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.3023,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.24489650557097503,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.3096,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.25179651517264773,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.3145,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.26261340645921216,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.2985,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.2516354805394906,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.3101,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.24404951359553773,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.3066,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.24774894447074303,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.2994,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.2506922185438074,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.3125,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.2369136238542681,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.3106,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.2350550264439098,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.329,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.24656732205919885,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.2954,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.2423813979168101,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.3139,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.24433669691200893,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.3232,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.25374002673435603,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.2871,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.26545878441533066,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.3145,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.24998817543618126,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.2968,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.23835858719002315,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.2829,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.25130275607446956,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.3178,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.25341601471228664,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.3126,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.24528789347750854,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.309,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.23930486427026929,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.3123,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.25067498102567654,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.3,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.24435538568215406,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.3035,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.24271288260916793,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.3288,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.24637825069207678,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.3101,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.24398387720954445,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.3102,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.24378234877951685,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.2884,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.23521574138801224,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.3046,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.2533879322136089,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.2914,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.23677170451251686,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.3236,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.24184381345022826,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.3101,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.23353382582969665,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.3079,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.23644970997603212,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.2918,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.24390600279839125,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.282,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.24214368572701755,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.3105,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.24764141758510355,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.2873,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.23900987543114616,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.3112,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.24054353548352053,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.3101,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.24798441942567637,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.2851,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.258033492730356,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.2851,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.2451950381073837,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.3224,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.2453195886510755,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.3044,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.2673570731383936,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.2677,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.25200079357329935,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.3035,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.24186176618688976,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.2953,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.25191841658885766,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.2831,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.2550056118893045,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.3154,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.2399206703947867,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.2677,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.25874147185932356,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.2855,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.24726058998340378,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.3354,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.2369469896673008,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.3019,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.23826436271540458,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.3052,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.2415930121065221,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.3126,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.23969548774843272,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.2908,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.2438667928120989,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.3185,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.24241147910659494,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.2987,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.2371377937098121,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.3235,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.22991408658054752,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.2947,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.23831693996604306,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.308,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.2393431301217548,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.2749,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.24631228400159555,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.3025,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.23564318390392555,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.3191,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.24831351334136365,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.321,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.24098043003536998,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.3011,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.2503251990713473,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.3168,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.24428280626751497,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.3111,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.24489817503983866,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.3124,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.2443653163353411,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.3075,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.24246294531655568,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.2926,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.23918562198897833,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.3118,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.24516326009817482,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.3088,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.25101075020059865,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.3064,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.2484875109808209,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.3017,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.2497078074578728,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.2962,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.25394871717996725,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.2911,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.24425160118613942,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.3149,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.24106539749663525,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.2876,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.2396426071716354,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.3007,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.23408499545636174,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.2981,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.2375740426604001,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.2929,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.22911245687795537,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.2868,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.25210827286660653,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.3109,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.23031599442975353,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.2888,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.24158669596611773,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.3106,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.24844275482414444,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.3109,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.24563850644098018,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.3204,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.23618787666692462,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.2948,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.24451817160529732,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.3057,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.2361628560719932,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.2891,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.2458367222985705,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.2856,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.24922956271456564,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.3026,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.24079857324075093,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.298,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.2456044953946978,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.2909,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.23608673148705356,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.3026,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.2441435860223073,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.2896,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.23848180930422078,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.3019,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.2408199440494385,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.2996,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.2486383449032548,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.3104,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.24696339035363096,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.2957,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.2432570798706974,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.33,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.2478047433601094,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.3164,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.24250248897807491,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.3049,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.24016126759061923,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.3246,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.2416594017803563,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.3071,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.23248281445101127,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.3019,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.22658934213489526,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.3193,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.23956597268658236,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.302,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.23518615496390616,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.3179,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.23954025485440672,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.2762,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.23618001748923573,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.3159,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.24597080945045793,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.2893,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.2372225894757917,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.3028,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.24421081012176374,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.3033,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.24722324142315488,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.31,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.2525497967132827,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.2987,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.22852799155886377,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.2993,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.22282305254192925,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.3267,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.25525198217690737,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.3051,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.22957882507719615,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.3068,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.2311446937058213,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.2975,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.24238570130246823,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.3028,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.25008289918078663,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.2989,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.23464571796879197,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.2949,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.24437334450314863,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.3068,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.24873594964023754,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.281,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.23614521432962063,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.3125,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.23978482171516266,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.3103,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.2472525742504935,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.3063,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.2494356199479332,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.2911,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.233842265480168,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.2979,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.2460165805111994,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.2855,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.24066640720749613,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.2935,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.24332796392375436,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.2787,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.24007652207308,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.2946,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.23766715359712529,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.3039,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.2530300681101982,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.317,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.232644264229771,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.286,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.24910440287066535,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.3053,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.23939404603564,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.3061,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.24378070966267226,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.2937,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.23837611016075247,
+      "learning_rate": 0.0,
+      "loss": 0.2955,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 3251999295930368.0,
+      "train_loss": 0.4668179716175867,
+      "train_runtime": 29567.1729,
+      "train_samples_per_second": 5.344,
+      "train_steps_per_second": 0.042
+    }
+  ],
+  "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": 3251999295930368.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}