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": 5.995743928336292,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.8857,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 5.84898784008959,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.8246,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 5.857608354960542,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.8554,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 5.94674149859297,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.8919,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 5.43602453917174,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.8292,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.28081625481142,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.8366,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.489576339278193,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.825,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.073251535786362,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.768,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.489999229218354,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.7854,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.168371898477801,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.7637,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.052918963435025,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.7767,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.778210969167731,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.7556,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.0077503984302725,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.7627,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 4.054530332506414,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.7536,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.9319314933387703,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.7584,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 3.0972798165890962,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.7062,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.9143863222978093,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.7012,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.3403383285263883,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.7079,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.6219977493102338,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.6801,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.512470478458768,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.6852,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.6304081774803931,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.6562,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.6995664068058072,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.615,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.5337708749857943,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.6388,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.2542158007737387,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.6284,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 1.092124090029604,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.6171,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.229233126237603,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.6457,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 1.1601272318472469,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.6167,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.899739958156907,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.619,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.7363156311915442,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.6108,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.8553798126272922,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.5843,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.8946197809894906,
+      "learning_rate": 1e-05,
+      "loss": 0.5927,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.6731614628891218,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.5926,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.6688226887065257,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.5635,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.699650301253786,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.5836,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.6354440590805875,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.5785,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.6864894058344223,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.5626,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6346358526473035,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.5558,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.6774719264353274,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.5888,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.5525475470617214,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.5417,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.6285196626322251,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.5725,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.6344843918126871,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.5308,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.5179659861430788,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.5649,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.5233889002828399,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.544,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.6262417416842713,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.5738,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.51773993194665,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.5331,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.49386180079256703,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.5374,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.5320891003132276,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.5668,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.47087345537627984,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.543,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.48097976618302724,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.5454,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.4516991963719062,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.5431,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.5525067079558014,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.5392,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.4630095890872355,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.5364,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.4106961133453071,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.5226,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.5144076207878077,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.5186,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.5870588609533756,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.5264,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.4890501154756294,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.5258,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.6528993703185187,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.5429,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.41484464434520596,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.5293,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.4818274994832776,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.5349,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.45503785088201804,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.5389,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.5588323380137873,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.5086,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.48343596932925886,
+      "learning_rate": 2e-05,
+      "loss": 0.5154,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.5017765550436015,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.5385,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.510889213264833,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.5268,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.556942207791306,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.5436,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.5487159375704805,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.5394,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.5907196567333026,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.5348,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.6282845535528181,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.5336,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.5449264542534015,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.5272,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.546137770729985,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.5066,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.46736138924117526,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.5007,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.5991168869182589,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.5346,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.6613410780727123,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.5119,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.5925836561678439,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.5396,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.4950729077688411,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.515,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.596258674979846,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.5454,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.5632471729033339,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.5048,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.5112719478743412,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.5094,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.5581713443516965,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.5566,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.5362529658205217,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.5321,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.5623148709657463,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.5183,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.5386075203571785,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.5358,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.5970033031449734,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.5286,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.5630604319135873,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.5133,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.47165668071121136,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.4896,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.549190829017185,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.5324,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.5671983620367546,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.5168,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.6469598959122926,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.508,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.8368618838007912,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.5315,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.6326011669673275,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.5265,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.6572454696359964,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.5172,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.5970238358309806,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.5155,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.7038410316108608,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.5107,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.563623734251756,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.5054,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.5853927953627995,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.52,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.8197699703370552,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.5185,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.9683742866732078,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.5072,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.6867907021183738,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.4869,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.7744216554716171,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.5265,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 0.5729085189471416,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.5118,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.9561280421981412,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.5171,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 1.2144127777470848,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.5079,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.6526618468206604,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.517,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.8341857700061324,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.5084,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 1.041011044388291,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.5138,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 0.7511329469038062,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.5131,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 0.7182593664950879,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.5174,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 1.1423187141325901,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.5027,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.7391220067816657,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.467,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 0.7621429471162965,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.5108,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.8027238450157105,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.52,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.7364183951535875,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.5114,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 0.671416289524356,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.5104,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.7542058035038927,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.5148,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.7163517520508228,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.525,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 0.7469484368435906,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.5217,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.7583132267258107,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.5236,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.7111343454481455,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.5131,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.7515099938366177,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.5101,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.7914446149132931,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.5193,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.6876777848012163,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.4935,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.659713248940656,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.5042,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.7066946747791878,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.5122,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.6529805693159042,
+      "learning_rate": 4e-05,
+      "loss": 0.5076,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.6219987134539365,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.5314,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 0.5729428755220142,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.5239,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.7213126049069796,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.5168,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.5651768160902834,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.5004,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 0.6797900448633926,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.4971,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.624537517584349,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.4732,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.6656227935242758,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.5054,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.6310665334113505,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.4857,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.5705947952338835,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.477,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.6528581639264867,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.5025,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.6743424266176532,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.5081,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.5497303534343602,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.5073,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.7806647635843135,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.5117,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.6391936278993572,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.5131,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.7951362172376135,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.5129,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.783155544424267,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.4983,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.7065839653749313,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.4913,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.7985771601810658,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.4948,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.5494532598945226,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.4912,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.6829001351314193,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.5048,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.6375033789500041,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.5069,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.6097771248421863,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.4977,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.7047986439972876,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.5197,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.5842994806673211,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.5361,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.7112696607753898,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.5248,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.4441283724936416,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.4875,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.559136540766387,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.4788,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.5466437603297271,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.4877,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.5747846341349281,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.4747,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.48403418324078984,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.4971,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.5227955263485186,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.4941,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.4488921927439625,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.4967,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.5128385820718162,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.4961,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.5258261004327839,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.5087,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.5184691396768745,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.5092,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.5061003961178774,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.4883,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.5882524297395124,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.4639,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.4791321037143229,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.5188,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.5871757680131341,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.484,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.4038631948087999,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.5176,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.6175477761848346,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.4697,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.589869812151643,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.5183,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.500710008423322,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.5043,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.5783310364821709,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.5009,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.4920898196528024,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.5003,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.4893362466942712,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.4919,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.5922797924592116,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.5122,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.5519138655663703,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.5133,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.5065523893593061,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.4982,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.44480687701524096,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.4817,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.5481381929172949,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.4848,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.5494121173818133,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.4864,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.5205679048449935,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.4846,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.4757480969873058,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.5099,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.7295334035091855,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.5377,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.5783984635840089,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.4963,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.5611899895598097,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.4829,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.5129000609974703,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.5083,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.5791873750817697,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.4917,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.5557753051340072,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.4811,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.6372890317693242,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.521,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.5924646721322726,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.502,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.5729258396539204,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.4927,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.4748667064648297,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.5178,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.581119271764251,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.4803,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.5599820097461098,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.4943,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.5112918053280527,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.4853,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.5085297553831242,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.4733,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.48739320921818124,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.5063,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.48628074804633986,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.4804,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.5126316059955808,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.4874,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.6532524225885543,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.4903,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.44671327038555164,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.4676,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.599764666204435,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.509,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.5417029517423546,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.4874,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.5857016584292014,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.4886,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.48936358470126357,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.4838,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.5805962832816671,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.495,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.5477083720933879,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.4911,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.4743292975971122,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.4877,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.5582322184391303,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.4931,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.47878137259798703,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.4908,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.48126802299278515,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.5156,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.5080919263683078,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.4657,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.4807386674421494,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.493,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.5124905158112727,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.4933,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.525175412790541,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.5141,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.522842688118544,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.4969,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.4699337440156443,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.4899,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.48317064086601835,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.5066,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.44085757895278466,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.4873,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.4919260446164858,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.5256,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.4250128833421103,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.4825,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.4694998126464947,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.487,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.43891137988857426,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.492,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.4635876568211123,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.4752,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.3961372074374294,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.48,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.41453133313143825,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.4897,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.3704941345769845,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.4892,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.37972368094932807,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.4833,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.3673772350848498,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.4909,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.3496730660162604,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.4655,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.3952259194293458,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.4874,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.4447070187716413,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.5147,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.35476624716819843,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.529,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.3968840608290535,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.4819,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.40317075213521447,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.4954,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.47448357666552105,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.5026,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.3839870357224418,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.5097,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.5422281352963545,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.506,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.5039642164174918,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.489,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.39527120781661856,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.4833,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.39687859443376816,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.4914,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.3848056280455947,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.5133,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.4098643546193354,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.4541,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.42358965067715937,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.4931,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.4066609616312249,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.4852,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.3890356615706711,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.4896,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.44190300476573857,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.4757,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.4203219258047266,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.5097,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.4479889992559476,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.4634,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.38184340087915125,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.4752,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4163340846997262,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.5083,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.46318046365655413,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.4285,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.39790040280222244,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.4197,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.4748334335646317,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.4234,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.4351273818110961,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.4143,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.46549969647389783,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.4263,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.4226993787878022,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.4196,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.4710170946167376,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.4153,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.44025692809795197,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.4267,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.43683602417399375,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.4322,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.42615619665911714,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.4355,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.42868954178908325,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.4129,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.42298843416015086,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.4007,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.40193769044636807,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.4012,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.41158545530839635,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.4295,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.4039566895860258,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.4081,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.3900418903263525,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.4223,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.4195644798337369,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.4123,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.40315503642992634,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.4336,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.39262422369656513,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.4282,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.38418270596260623,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.413,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.4466522933145431,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.407,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.37254601434364604,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.4224,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.5250092287572736,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.4346,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.46770942932334486,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.4277,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.4591432994833295,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.4095,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.4909458156363743,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.3995,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.4757373546500382,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.4575,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.3702260829214202,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.4003,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.4104803406826697,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.4091,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.7039916617441615,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.4101,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 1.0789179964283158,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.4239,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.4818372356920153,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.4372,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.4427533182994016,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.4185,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.4052716850062573,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.4023,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.4395215535371073,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.41,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.4116967746311307,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.4183,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.5934951571636621,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.4283,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.5179043576598195,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.4162,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.513930825808106,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.4188,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.6521151796823224,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.414,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.5414141784835865,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.4054,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.6759463141759497,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.4488,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.46419639221114556,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.4521,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.5751072821308439,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.4202,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.585556158686014,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.4344,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.44886742222580317,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.4226,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.5921182684474755,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.4379,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.45401919805811797,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.4308,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.49266732697163446,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.4202,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.4123007468494242,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.4337,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.5721445257640198,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.4357,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.3885658914751434,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.4241,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.5952776257607042,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.398,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.45451439002329125,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.4132,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.48540849406230385,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.4232,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.431971248431566,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.4319,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.47034333443012094,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.3868,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.3723175496868396,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.4298,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.5303852798194886,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.4197,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.6064088409774209,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.428,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.41788065750405173,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.4203,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.4227691426195425,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.4222,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 11.580183477413307,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.4614,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.6541933814135152,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.4194,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.5260737365833653,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.4416,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.6151305378441528,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.4232,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 1.815160663242574,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.4061,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.7690485824237858,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.4446,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.611932790982559,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.4374,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.8057724460216963,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.4444,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.4447179393546682,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.4111,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.7706807797262977,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.4244,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.5499810478249263,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.4342,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.6851570434510014,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.4371,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.5612632300295017,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.4018,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.697659984649047,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.4225,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.5143987068902428,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.3998,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.6557262127206981,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.4212,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.45828454819371983,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.4455,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.6044918643593126,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.4315,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.4520988981998993,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.4144,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.5624132816593491,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.4168,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.5262468175398626,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.4408,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.5212260624177614,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.4308,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.6846377824003947,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.4278,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.6063480099465425,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.4155,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.45196218665551235,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.4173,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.5192042801265918,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.413,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.4578678088835355,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.3987,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.43482875462252557,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.4133,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.4506999577202459,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.4282,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.39405564631982926,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.4127,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.4851247392022014,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.4314,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.41938300681854873,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.4336,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.46213754431050874,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.4231,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.47379978120324145,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.4079,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.4811878391406682,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.4293,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.33131272375114906,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.4249,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.43245877966211005,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.4404,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.3241974139456826,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.4173,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.41256480048666555,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.4174,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.35519772229627306,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.3938,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.3748262410642998,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.4207,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.3738441051810799,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.4345,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.37778166217219034,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.4143,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.44121569578271597,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.4296,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.3909070509836028,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.4229,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.40644786161417057,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.4403,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.4011800155162489,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.4109,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.39174780489610583,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.4197,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.3842627371120732,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.4211,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.40507630852657317,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.4186,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.37839133615381176,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.4038,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.3633650183688817,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.4043,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.3607301237147278,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.4241,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.4379803974781942,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.4375,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.3700103797414678,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.4587,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.4295254067583161,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.4292,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.3501926640223395,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.4198,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.43792508133849906,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.4162,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.4229308220136951,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.4311,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.36626644142467385,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.4173,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.4068493636432903,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.4131,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.3215847808202276,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.4011,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.41939615956120396,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.4195,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.3951314466397384,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.4173,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.41559847886735,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.4265,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.38510351508006724,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.435,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.4044765439856461,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.4048,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.41024583749121535,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.4015,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.38563103894466233,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.4121,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.40527774247737625,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.4357,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.3484086540013349,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.4339,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.36892216544534623,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.4164,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.32858620611217376,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.4242,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.40068872289074414,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.4182,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.4541441078238831,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.4469,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.3753851789922739,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.4252,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.37957162675764233,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.4098,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.4479087729756309,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.4571,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.36329586236857775,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.4203,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.39389103031415723,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.4274,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.3380337814020961,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.4216,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.3494711618684675,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.4181,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.43093522326536127,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.4318,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.3185447572323538,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.4263,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.48053039982587775,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.4186,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.38720245799286146,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.4075,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.3791207538269182,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.4379,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.3926162955196774,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.4198,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.3217613753690778,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.3923,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.41607228132681,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.414,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.2977476587041196,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.4069,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.38012973719880894,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.4349,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.35751965654684426,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.4178,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.3596268475003013,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.4205,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 1.5292370777912874,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.419,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.4499914966572882,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.4044,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 2.446876127279502,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.4244,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 8.340665759556156,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.4654,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 1.8047061669396987,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.4545,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 3.134456927458385,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.4973,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 3.773349184515084,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.479,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 1.0026023784811282,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.4507,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.8054625911359287,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.4116,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.8807782853215519,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.4345,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 659.3301715096749,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 10.0331,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 1.5233329265996467,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.4583,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 2.6513228855609334,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.4344,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 1.4991366964050343,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.4618,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.7351410252916444,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.4518,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 1.014451879968675,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.4375,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.6312270516817623,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.4325,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.7442817374427869,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.429,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.5432862671182281,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.4184,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.7262101008357207,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.4215,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.7405100640419382,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.4282,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.7332907838126113,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.4354,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.5646530087010274,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.4215,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.5633179554196002,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.4096,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.5150570790608688,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.3998,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.4837040521239469,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.4078,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.564084944434986,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.4193,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.45782677925803306,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.4336,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.4259136958767686,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.4276,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.5100763976826861,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.418,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.3627073939292151,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.4179,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.5359873995595973,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.4242,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.3786817891664447,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.446,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.516335588268405,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.4215,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.3699383975980625,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.4242,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.3720479499466603,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.4229,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.3917204060972954,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.4309,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.35841144560404686,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.4177,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.40139199019618904,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.4168,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.3740276598024876,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.4078,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.362358479970685,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.4218,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.4004523034560373,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.4157,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.36626217534973915,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.4295,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.3954406522946314,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.4256,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.38179579314382445,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.4096,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.34938593161339165,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.391,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.3405166277803267,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.3934,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.33104853705469,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.4161,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.34419616990740026,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.4211,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.3578589448035697,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.4316,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.32869290435046544,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.4235,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.3381784846190314,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.4232,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.3238609862780929,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.4387,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.30501705341526925,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.4189,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.35585775667790415,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.422,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.32054912408957154,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.4026,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.29960090354570923,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.4094,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.29829506036156206,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.4166,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.30956524552764136,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.4128,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.3679650154814362,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.4014,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.3310962643127605,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.4272,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.33940380965513445,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.414,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.29518343939065267,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.388,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.3539791917092375,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.4338,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.2978050166327431,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.4408,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.3192705516079902,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.3894,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.3115689359082826,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.4158,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.3116686761883348,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.4213,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.30926188076524314,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.4194,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.32033317706778985,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.4012,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.3396280139406618,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.4165,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.3638992196074814,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.4137,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.3253031355969175,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.4182,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.3831518486700153,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.435,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.35123810308294756,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.4102,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.399057268830187,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.4222,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.4321843237583283,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.4372,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.39703763618698024,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.4238,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.42075954415856864,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.4009,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.36231662053174657,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.4031,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.399292386974108,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.407,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.3294965527383642,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.4189,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.3550451185541771,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.432,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.33796083303466584,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.4078,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.34647462486687886,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.4333,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.3230135826672183,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.4135,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.3392571455123131,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.4182,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.325665051108273,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.4028,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.3262571928339191,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.4427,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.32075878226002197,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.4143,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.33575725254145894,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.4125,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.4564630882690283,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.354,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.42220053090431103,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.3129,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.4619012754399503,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.3287,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.46316438137070465,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.3376,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.4011546796143333,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.3734,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.35609312674615884,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.3248,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.3865299331368283,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.335,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.46603443224181645,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.3343,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.39801182241489996,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.3229,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.43199905733460847,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.3323,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.4408058980396711,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.3291,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.4025573758686661,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.3602,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.43365024611531006,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.3181,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.40137362613512395,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.3428,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.4001959224626208,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.3218,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.3738888793573515,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.3203,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.35647284032302784,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.33,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.39850812021334153,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.3381,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.34790594349028003,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.3166,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.3574369559264932,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.3246,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.39211283333773617,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.3134,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.3298328521689692,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.3341,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.390338536832132,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.3336,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.3744689214741357,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.3306,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.32183221087053326,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.3446,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.342733203832684,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.3351,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.3661237624427145,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.3593,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.3369837397820888,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.3078,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.36416722695565784,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.3276,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.4032208806363377,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.3282,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.3891577701634072,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.3336,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.38141062233129414,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.3344,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.32160676327738896,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.3104,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.41592731672196287,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.3463,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.35075727273940716,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.3358,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.3850038695746571,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.3258,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.38430068440559373,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.3167,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.3730925627559726,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.354,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.36433939931212006,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.3267,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.3895120892968634,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.3279,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.3927464460572885,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.3352,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.4263742547200056,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.3337,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.33788752465506683,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.3094,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.4258702383004806,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.3498,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.33603142324536506,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.3339,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.39501965009373763,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.3312,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.3206315982057152,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.3345,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.37516842305698606,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.3167,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.4464760527536208,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.3314,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.350033926980126,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.3242,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.39402420218019674,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.3426,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.3334309036706509,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.3185,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.37583278311590007,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.3263,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.32419143700860037,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.3233,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.39174211242812274,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.3287,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.3262207409481195,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.318,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.35380498111043945,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.3147,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.31104645310621143,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.3273,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.33141608432256725,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.3265,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.3317227323902745,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.346,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.312037880991474,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.3543,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.35738297170854033,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.3263,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.3140806249624721,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.3274,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.32246647472410245,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.3246,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.33845435697802356,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.3202,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.3300376379457373,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.3391,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.3367698200713191,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.3342,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.3257469599407618,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.3282,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.3376886147652718,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.3407,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.3550506535159614,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.3489,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.30333356458641075,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.317,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.3844374640929318,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.3286,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.31487828627053355,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.3435,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.39270828048084816,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.336,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.3347690601218965,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.3408,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.3817281941492656,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.3374,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.37144612903137075,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.3404,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.33678139487955366,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.3417,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.3647852773806986,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.34,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.37093734898077196,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.3359,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.34362469644572075,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.3397,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.4279123180620762,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.3319,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.30231149267941354,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.3353,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.40622532388365434,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.3128,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.3597630714020983,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.3353,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.3803912578668225,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.3247,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.40356074890226074,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.3475,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.3846187457280564,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.3273,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.4090600795118469,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.3285,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.4005741502498713,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.3225,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.32831197338077417,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.3481,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.38357221854048684,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.3425,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.3447655314306072,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.3459,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.347667711910716,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.3486,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.3630821811726249,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.338,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.3185726816977044,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.3368,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.35399815678620095,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.3286,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3031387765091462,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.3207,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.3518937470963418,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.3351,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.30474005522865305,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.3166,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.3353270104416083,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.337,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.3202350010426476,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.3257,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.32356399748626646,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.3397,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.4249582171655473,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.3491,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.29911742811475417,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.327,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.3174629721376187,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.3377,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.3273797770919448,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.334,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.3289622393582968,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.3381,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.31600901781300306,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.3269,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.30057571451769655,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.3229,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.31287531134913876,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.3405,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.31969989089238265,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.3419,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.3758766132216779,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.3309,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.33909479633666656,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.3503,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.2752366980020035,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.3238,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.3084569240454886,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.3266,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.2998076074184059,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.3495,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.2826077596070367,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.3313,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.3520045251522789,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.3522,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.2793498689797315,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.3277,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.3194321323821889,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.3458,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.2907182552342607,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.3447,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.30740988151979126,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.3322,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.2778936368936946,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.3304,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.30900817422980364,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.3428,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.28820749265179885,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.3198,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.31040474075832153,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.3438,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.2909067550447811,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.3323,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.28817931670005936,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.3132,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.28618671320650646,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.316,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.2764269180338971,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.3176,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.3826434361669781,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.3226,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.2986261261516348,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.3432,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.33202864765970647,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.3359,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.2961957958990913,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.3258,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.3045821163505706,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.3479,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.3077739119893418,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.3374,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.31030459344651906,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.327,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.3207058314073198,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.3519,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.2955618514484793,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.3407,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.310817914780091,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.3406,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.28518605368041705,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.3261,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.3058810386971839,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.3242,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.34724949653597814,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.3147,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.310957169792725,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.3372,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.3035155829138116,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.3252,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.29792138696595477,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.3301,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.307381118612827,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.342,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.3175616645198012,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.3178,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.3052929181522019,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.3149,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.2961381418729913,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.3253,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.3271347225626524,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.3084,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.32180658955715835,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.3361,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.31265244181997554,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.3563,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.31130573619853186,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.3387,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.2964644989847796,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.3452,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.3143884684307729,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.3351,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.2976998763014312,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.3365,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.29446007275802866,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.3253,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.3078791306085053,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.3454,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.2759369881247033,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.3136,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.29231410647587375,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.3288,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.31054888032397515,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.3473,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.2938440383770707,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.3365,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.30374452555341674,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.3376,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.2885914130304289,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.3181,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.3293712744394787,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.3422,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.2758960804175602,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.3342,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.35184922986296435,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.3452,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.2925989420446461,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.3395,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.27892794444516966,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.3365,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.31402112032893836,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.3257,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.28316676233765836,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.3433,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.2791649632296786,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.3483,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.3100791319944055,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.3267,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.2790678378065266,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.3338,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.31181611957139166,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.3462,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.2726778057398488,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.3575,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.28887400349416714,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.3412,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.2977273323431278,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.3274,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.2819160042068189,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.3105,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.31614379412402654,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.3435,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.2807285581432801,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.3357,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.32005641060505485,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.3288,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.3085385029374961,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.3485,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.30567599175961696,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.3239,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.2885488268123726,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.3245,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.29091217246939605,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.323,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.30051732346879984,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.3327,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.29975349665973916,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.3339,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.32833881856676994,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.3454,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.2907075067249056,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.3316,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.34962781230115575,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.325,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.3013670408026995,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.3466,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.3554382875346681,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.3426,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.29548802235729976,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.3089,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.304873866421489,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.3426,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.32487133863176765,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.3067,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.28613605449674495,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.3309,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.31182605840208677,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.3458,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.3292220416535557,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.3371,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.3262625955785152,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.3343,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.3266497519820725,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.3321,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.33675611364169006,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.3499,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.26607403811488,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.3416,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.34864431569181953,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.3202,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.29801140974043605,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.3382,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.29560848755632996,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.3186,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.3308693946947564,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.3411,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.277136672722498,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.3191,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.32605600386740463,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.337,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.287573957448844,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.3217,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.31997161950065733,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.3215,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.2872675797656785,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.3278,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.32211137291895486,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.3445,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.2897365238103541,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.3412,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.30768210268341656,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.3177,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.32334319818750046,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.3376,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.28375166277307273,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.339,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.31910346965201153,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.3334,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.27521198926854573,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.325,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.3128391074071525,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.3216,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.28841270694062904,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.3303,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.2820691621156627,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.3181,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.2900843934576689,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.3246,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.25834734643218965,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.3319,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.29306652584073656,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.3286,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.2579819557834044,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.333,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.26800467109920895,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.3312,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.28606209638823865,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.3448,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.26374048452332677,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.3339,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.30117420399476175,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.3413,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.2863083686742939,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.3284,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.26365440398982776,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.3127,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.2725491019389563,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.3377,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.30231513358656154,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.3304,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.27950446204192475,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.3382,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.3199939257694519,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.3455,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.30620532254440486,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.3064,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.2997835700145586,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.3462,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.29075892438972256,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.3423,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.2733493724807016,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.3254,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.25903871296947645,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.3217,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.2682334265845026,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.3376,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.2915546381267614,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.3288,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.28689839593632094,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.3242,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.2554392536103866,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.3444,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.5835751168394387,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2522,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.3133219462009255,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.2448,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.733158651468192,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.2547,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.3505095733711323,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.2358,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.4796493766918344,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.2372,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.43971303952751045,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.2389,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.35524500337641923,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.2419,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.3864811972234915,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.2619,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.4217494854827549,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.2479,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.33370492443219024,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.2679,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.35422281880069617,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.2454,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.36661215895968224,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.2326,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.3490933939668861,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.239,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.33279752693708203,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.2419,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.3452480958016108,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.2484,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.33795447882673496,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2403,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.3298860286577796,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.2421,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.29522747811416683,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.2263,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.3080696418649201,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.253,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.32022048427286265,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.2509,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.2899424319661325,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2354,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.30157642923811695,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.2357,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.30880978011564836,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.2372,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.2994526272734142,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.2522,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.2956781720561229,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.234,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.30053841291557465,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.2564,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.29899280626265123,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.251,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.2914430418024895,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2534,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.3060682358272995,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.2391,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.3196362130547634,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.2382,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.28426835659188476,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.2323,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.2756638209962817,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.2269,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.30226822237466744,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.2327,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.2943006107731347,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.2557,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.29240730361214773,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.2496,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.2866745788094937,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.2492,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.2998450673537633,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.239,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.2866627496996306,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.2441,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.27768096791818725,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.2443,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.27843664652306155,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2417,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.27605630969518163,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.2427,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.2724996064648268,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.2522,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.3016034099768857,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.2242,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.28501480437687243,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.2384,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.26451194640624076,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.2326,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.2732721994203457,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.246,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.26548020694638386,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.2576,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.27362591662593094,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.227,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.27590116871718423,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.242,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.27791658522055923,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.2432,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.27024839853956584,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.2322,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.271805523826188,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.2421,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.27816707000661495,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2303,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.2951099926945175,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.233,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.3042516333748742,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.2423,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.27471920976994946,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.2407,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.2881452781929505,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.2532,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.2890956095461252,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2375,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.27111979056957547,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.2319,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.27816037975472274,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2538,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.30467381310296215,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.2572,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.2616981345863463,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.2612,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.29136412485569635,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.2529,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.284036579793863,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.2351,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.2782433827862521,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.2399,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.2858400242572286,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.2348,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.28563151487453475,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.267,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.2692089576421457,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.249,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.2836631351834005,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.2507,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.2633215900777347,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.2242,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.298182478991203,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2475,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.28604486372012217,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.2423,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.2832100536189015,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2585,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.28285349032368834,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.2356,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.29693093673820886,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.2486,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.29924068865015896,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.2464,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.2728229456952813,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2471,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.27274644180273594,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.2484,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.2948175836571744,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.236,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.302136391804252,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2597,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.2718016492535063,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2375,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.27660890822485984,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.2438,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.2707001723998004,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.234,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.2708250002149788,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.2546,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.2590956496825162,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.2461,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.2568693218701219,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.2412,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.2633948571790117,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2424,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.2726104496804457,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.2378,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.2488887711233269,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.2449,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.26928066763845554,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.2524,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.2726437940478812,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.2333,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.2706769117922981,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.2404,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.2771330690398696,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2499,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.281786928214495,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.2568,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.27947950265126564,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.2637,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.2809863888388822,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.2651,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.2629913946481208,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.2444,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.2818572784858291,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.2671,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.27780140831511524,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.2183,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.26701146110390755,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.2523,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.2703443758556053,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.2441,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.28605328319442636,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.2363,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.28613221621893886,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.2553,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.2675348039038383,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.2501,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.278474724193619,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.2632,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.2751583966756434,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.2522,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.2684225597843277,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.2302,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.2783040584347159,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.2244,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.2710190014432921,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.2263,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.287128491492998,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.2573,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.268952236825472,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.2371,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.2703366582824677,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.2469,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.2889693160833185,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2277,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.2682797907741322,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.2486,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.2653050444413486,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.2504,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.2708808987310403,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.2532,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.2801774351971736,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.2375,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.26394613671147527,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.2527,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.2677995605167786,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.25,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.2866110433741485,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.2336,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.27496856759726185,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2354,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.27250457063139527,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.2456,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.2649846622550655,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.2401,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.2580332945247223,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2298,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.28221954734273347,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.2463,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.25628193747524836,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2308,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.27080444873945264,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.241,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.27328099327042543,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.2638,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.2784885086856091,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.2315,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.2747358216518299,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.2532,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.26631451866029154,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.2314,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.25958115831024736,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.2461,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.2646281970189627,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.2276,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.27583700456663024,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2607,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.24210924457359229,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.2368,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.25549071142719443,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.253,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.24930881749578254,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2337,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.25363248450862896,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.2521,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.26407278854845,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.2452,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.27173226020083624,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.2291,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.2695066371910974,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.2527,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.2580811293768525,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.2292,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.2704419521776085,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.2554,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.2666007015367646,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.2507,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.2525346515602053,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.2442,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.2571497888556597,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.2486,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.2599045999102285,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.2427,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.2666943890399845,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.2398,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.2712804819193819,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.2478,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.2638050509785591,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.2479,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.2694886903597673,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.2333,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.2625227130304416,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.263,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.2781926564697027,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2668,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.27046428654847476,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.2477,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.24727119400858397,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.263,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.2633316881991464,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.2384,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.26130926664806686,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.2299,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.3227300281953672,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.2377,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.26910361139995137,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.2443,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.2640606649867982,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.271,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.2529675129547032,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.2424,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.25467393863924603,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.264,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.27197156624409874,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.2434,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.2787119999711179,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.25,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.25486896209849946,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2526,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.2594689231069863,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.2192,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.2731438117817346,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.256,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.27974223145452737,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.2551,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.25291729971000904,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.2323,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.2747962525239562,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.2488,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.2776814163156164,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2573,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.2843807866262877,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2432,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.26679347488199884,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.2366,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.2649343798020637,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.2384,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.26402884167926916,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.2529,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.26193417427309407,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.2282,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.2677962229427931,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.235,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.2557988062938005,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.2516,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.25153146333548715,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2422,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.2644465037280319,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.2549,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.2710445891348643,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.24,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.2617187413673325,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.2595,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.25977055137803,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2372,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.259807904589199,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.2563,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.2639973084944849,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.2407,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.2682342930866284,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.2454,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.2552083680115714,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.2535,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.25568247948527995,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.2525,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.2791107211462181,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.252,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.27088056994126164,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.2392,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.2623512719565068,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.224,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.2672264935361282,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.2524,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.25237730855506935,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.2303,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.2670494655143998,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.2435,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.2565429682150129,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.244,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.27493513811288356,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.2278,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.2544233299410556,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.2458,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.2830780590933559,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.2596,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.2630279882086839,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.243,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.2666004344821763,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.2478,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.28221326307793687,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.2153,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.2713629943540288,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.2375,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.2760630011126762,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.2433,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.27291762546476567,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.2669,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.26261301041241636,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.2414,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.2530315415000248,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2314,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.26959305945605955,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.2471,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.2638173061699087,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.2594,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.2577979489969081,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.2427,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.2622045046072471,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.232,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.256183153984749,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.2364,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.26268906988425333,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.2435,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.24616635253094205,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.2551,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.2718888981481694,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2538,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.26781478385803875,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.2411,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.26868141677940843,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.2428,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.2582053610285048,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.2427,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.2553042841780978,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.2431,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.25295914438294465,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.2472,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.2486830110482166,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.2342,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.26564354637158977,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.2459,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.2568670903762963,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.2427,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.27101961340335434,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.2622,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.2777466714054332,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.2148,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.26210811783605714,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.242,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.2539110232219965,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.2534,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.25412140677066153,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.2409,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.25230767769273194,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.2557,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.27280448201258567,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.2338,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.25157819360885036,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.2416,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.25342590837902834,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.2418,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.24341962879398196,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2262,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.24058157653621018,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.2137,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.2992037610414381,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.249,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.2622131117566907,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.2175,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.23449190160222827,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.2428,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.24226270019515397,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2433,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.2723097132922341,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2224,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.25143351574295225,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.248,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.26229534750007666,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.253,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.26749995581423774,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2571,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.43636410730459435,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.2496,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.2443845017402691,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.2367,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.2593078549857545,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.223,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.2749218446725211,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.2311,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.26292575823219977,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.25,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.2588926689872232,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.2385,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.45958882403005147,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.1949,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.34322391385737105,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.1735,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.2861255409153952,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.1942,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.3303740295330826,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.1772,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.4263446216793085,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.1957,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.3969727781200364,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.1859,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.2999781840854303,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.1745,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.35321146761367755,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.1843,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.39104423939351907,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.1728,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.4011701899376963,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.1682,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.2915518128503706,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.1717,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.3046736985436904,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.1987,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.3392203044522923,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.1771,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.3421555442991155,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.1889,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.31049570412168237,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.1788,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.27281385089723126,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.1766,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.26793679970035084,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.1714,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.2942059425893595,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.1918,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.2854969623898522,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.1626,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.2668507885216925,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.2053,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.2534188289530066,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.1752,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.262791952790782,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.1692,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.271391287937614,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.1808,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.2790494907365424,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.1794,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.262668703619211,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.1839,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.2525451254792512,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.1828,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.2522078361508901,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.1807,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.25232292567852926,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.18,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.25018125748504416,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.1653,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.2635145064087615,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.1834,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.25680133353290063,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.1879,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.2683426899182466,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.1972,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.2593571625791963,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.1846,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.2521464973424621,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.191,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.25339537489598574,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.1951,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.24580620662402414,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.1929,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.2462757561105887,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.1909,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.24026438705908848,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.1683,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.2549479777918683,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.1822,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.24435785548956002,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.1891,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.24372066746458346,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.1872,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.2637588317444383,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.2062,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.24432847555485085,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.2012,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.2536183538650344,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.2013,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.25867211525356565,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.1747,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.26352912877285284,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.1728,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.25051323880072723,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.173,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.2518973021200358,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.175,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.24949428960728676,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.1927,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.2508118219523418,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.1812,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.2420097825542755,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1788,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.23498982359896367,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.1717,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.24062125252448371,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.1948,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.24769048090190052,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.1925,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.2559266240268413,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.1764,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.2554084300189258,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.1917,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.25289098044150654,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.1711,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.2502623408788126,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.1881,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.24062447346598634,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.1934,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.23803297340828122,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.1862,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.2503908808066172,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.1992,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.24137344912671838,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.2012,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.24841490796897794,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.1766,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.2574652218219172,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.1753,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.24267743188989754,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.1741,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.2430888560540027,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.1763,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.25750268496518086,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.1656,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.2575066312791849,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.1728,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.25558387927835047,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.1835,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.23543427301964537,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.2083,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.26382548251923954,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.1894,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.2545414864867143,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.1797,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.24420982337044317,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.188,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.24799952347041393,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.1774,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.24991666224458392,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.1875,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.24000659230170487,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.1899,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.24936805047132354,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.1776,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.2497062417131247,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.1938,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.2617211005943497,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.186,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.2491120448360401,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.1811,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.2493507318095294,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.1819,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.23294385128436054,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.1834,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.24299262208058786,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.1747,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.23928152000741404,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.1784,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.24275876929257362,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.1754,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.25097153500904684,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.182,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.2399822372971045,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.1797,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.2547524342618445,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.1777,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.2497082292858563,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.1688,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.23951073393219982,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.1654,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.2479571554534125,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.1874,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.24173726295198406,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.186,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.5043742552693209,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.1896,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.2482020321165998,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.1783,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.24913379253134166,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.1873,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.2533652670664892,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.1714,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.24260496443249,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.1921,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.24196142680005817,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.1764,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.24384099926489586,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.1896,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.2342281933219824,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.2005,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.2481561979383119,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.1968,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.23527261592990972,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.1907,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.2374536687537135,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.1792,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.24020333434639674,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.1838,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.2358034518902722,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.1712,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.23994298377037163,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.1714,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.24832211482346822,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.1732,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.24566122284636654,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.2077,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.2389851553370132,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.171,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.25282016213288644,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.171,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.2641603623785364,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.1857,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.25131257439332944,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.175,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.2453766120051595,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.1832,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.24674056888963605,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.1663,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.2391558119774073,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.1844,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.24954706194959347,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.1748,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.23692070808436785,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.1603,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.24250666532333082,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.1759,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.25234218269474024,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.2004,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.24815078698516627,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.1888,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.25345193121745735,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.1742,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.24860113288484675,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.1964,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.24265541432653576,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.1944,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.23853570945232014,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.171,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.2404391971087448,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.1733,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.2381560942375561,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.1713,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.24245503667527044,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.1912,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.2425213967134462,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.1912,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.24410515342577088,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.1766,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.23464276273341736,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.1664,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.2491168211790467,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.1912,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.2293539340977046,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.1733,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.24241939658348707,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.1682,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.2495298672413952,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.1921,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.24500269876567515,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.1875,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.24092694969979173,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.1819,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.23966953248019388,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.1841,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.2371964740598655,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.1753,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.2482020595617604,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.1896,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.23691762465985414,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.173,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.2460533590413968,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.1888,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.24504050788432383,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.1763,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.24560968235322453,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.1708,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.2534668514558345,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.1806,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.24727299184521717,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.2096,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.2497003938720705,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.1836,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.2510900510914232,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.1642,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.25077435540983856,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.1768,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.252059877106192,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.1704,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.2526029086710936,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.1726,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.24526994926849857,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.1811,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.24088362856057344,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.1697,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.24320841817336963,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.1787,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.25746344257372894,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.1791,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.2321367989843686,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.1965,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.24227254013973412,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.1899,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.2633981555093553,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.1779,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.2327094853910794,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.1779,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.2454903550926458,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.1874,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.24445744053992896,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.169,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.2376515322625982,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.1898,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.24350109538690975,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.1887,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.23486793898877278,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.1843,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.24815044728038493,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.191,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.24461785554920998,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.1673,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.23482337325301075,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.1939,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.2522295625129565,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.1764,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.2399305222806714,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.1757,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.23929780989844648,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.1743,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.24129909331402344,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.1683,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.23726391643260308,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.185,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.24487945866342717,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.206,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.24874504577086443,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.193,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.23766272214474932,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1843,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.24738426419517967,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.1787,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.24946251040049586,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.1877,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.23419248703765017,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.1973,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.22892978695557,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.1929,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.23777105199791887,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.1872,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.2423749416038842,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.181,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.2456660472158873,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.1763,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.241115852126767,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.1782,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.23897936686571458,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.1617,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.23946682004037323,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.1717,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.2503807773972539,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.1865,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.24745374614559806,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.1707,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.2444087775160195,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.1894,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.22769013596716556,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.203,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.24025309395958536,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.1845,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.23545429297505985,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.1917,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.24977148744332883,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.1875,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.2433083914348605,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.1754,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.25088337611835076,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.1945,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.2286120693329966,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.1834,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.24066468449663536,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.1849,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.23673945734829652,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.2103,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.25595531536378585,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.1969,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.23703066714225518,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.1881,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.2558385840250548,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.1913,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.24273657370214885,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.1862,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.2536164046595826,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.1675,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.24333544090623185,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.2008,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.22602900602720413,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.1641,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.24554744579860718,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.1861,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.2393690461706231,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.1861,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.23369459265874257,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.1876,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.24942501289696686,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.184,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.2474152353527098,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.1752,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.23545231274035797,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.1963,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.240128601320219,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.1912,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.24007146902563267,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.1933,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.25715096417239,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.1846,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.24224142777950441,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.1938,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.2419148648697214,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.1917,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.24787971737161088,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.1812,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.23118736979458968,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.1688,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.24539559011205608,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.175,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.2424780477744415,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.1932,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.25045497658627186,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.174,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.23868578108029087,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.1708,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.2471586980085277,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.1729,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.24563755276145824,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.173,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.23880664700648552,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.1635,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.24553837570929143,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.1739,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.25205748883771983,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.1967,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.2373114924024068,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.1938,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.24530940549425737,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.177,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.24349223376778317,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.1873,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.24514695123374303,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.1677,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.2424411104029982,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.1836,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.24346994698086533,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.1851,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.2353693440204151,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.187,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.2569604073183458,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.1649,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.2468261577526337,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.1874,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.2414254322676255,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.205,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.24271602413493468,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.1672,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.2747157152138343,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.1701,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.2567495833498511,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.173,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.2489453104893538,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.1608,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.25099912699031884,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.1814,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.24161064046732697,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.1866,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.23709496138941188,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.1926,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.2436847804584222,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.1775,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.24596550010546106,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.1842,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.24201994868032708,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.1826,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.2419894072729756,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.2163,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.23968011461618022,
+      "learning_rate": 0.0,
+      "loss": 0.1695,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 2.6027063082850714e+18,
+      "train_loss": 0.351337015146186,
+      "train_runtime": 28524.9486,
+      "train_samples_per_second": 5.539,
+      "train_steps_per_second": 0.043
+    }
+  ],
+  "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": 2.6027063082850714e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}