diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6307 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.995135510771369,
+  "eval_steps": 500,
+  "global_step": 895,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.005559416261292564,
+      "grad_norm": 7.9358195450351205,
+      "learning_rate": 8.88888888888889e-07,
+      "loss": 1.2622,
+      "step": 1
+    },
+    {
+      "epoch": 0.011118832522585128,
+      "grad_norm": 7.942651735525788,
+      "learning_rate": 1.777777777777778e-06,
+      "loss": 1.2714,
+      "step": 2
+    },
+    {
+      "epoch": 0.01667824878387769,
+      "grad_norm": 7.8124967595464785,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.2541,
+      "step": 3
+    },
+    {
+      "epoch": 0.022237665045170257,
+      "grad_norm": 7.377301625314602,
+      "learning_rate": 3.555555555555556e-06,
+      "loss": 1.258,
+      "step": 4
+    },
+    {
+      "epoch": 0.02779708130646282,
+      "grad_norm": 5.956675332171705,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 1.2091,
+      "step": 5
+    },
+    {
+      "epoch": 0.03335649756775538,
+      "grad_norm": 3.181434103411455,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.1656,
+      "step": 6
+    },
+    {
+      "epoch": 0.03891591382904795,
+      "grad_norm": 2.540442378830216,
+      "learning_rate": 6.222222222222223e-06,
+      "loss": 1.1705,
+      "step": 7
+    },
+    {
+      "epoch": 0.04447533009034051,
+      "grad_norm": 6.203334552578263,
+      "learning_rate": 7.111111111111112e-06,
+      "loss": 1.1541,
+      "step": 8
+    },
+    {
+      "epoch": 0.05003474635163308,
+      "grad_norm": 6.731756614308619,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.1547,
+      "step": 9
+    },
+    {
+      "epoch": 0.05559416261292564,
+      "grad_norm": 6.247745131639911,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 1.1347,
+      "step": 10
+    },
+    {
+      "epoch": 0.061153578874218205,
+      "grad_norm": 6.29730189833508,
+      "learning_rate": 9.777777777777779e-06,
+      "loss": 1.1331,
+      "step": 11
+    },
+    {
+      "epoch": 0.06671299513551077,
+      "grad_norm": 5.132108221825742,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 1.117,
+      "step": 12
+    },
+    {
+      "epoch": 0.07227241139680333,
+      "grad_norm": 3.2550227184562375,
+      "learning_rate": 1.1555555555555556e-05,
+      "loss": 1.0767,
+      "step": 13
+    },
+    {
+      "epoch": 0.0778318276580959,
+      "grad_norm": 2.294739996569744,
+      "learning_rate": 1.2444444444444446e-05,
+      "loss": 1.0591,
+      "step": 14
+    },
+    {
+      "epoch": 0.08339124391938846,
+      "grad_norm": 2.257537007260575,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.0411,
+      "step": 15
+    },
+    {
+      "epoch": 0.08895066018068103,
+      "grad_norm": 1.919991827155782,
+      "learning_rate": 1.4222222222222224e-05,
+      "loss": 1.0206,
+      "step": 16
+    },
+    {
+      "epoch": 0.0945100764419736,
+      "grad_norm": 1.543897719994616,
+      "learning_rate": 1.5111111111111112e-05,
+      "loss": 1.0078,
+      "step": 17
+    },
+    {
+      "epoch": 0.10006949270326616,
+      "grad_norm": 1.2679636773521212,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.9897,
+      "step": 18
+    },
+    {
+      "epoch": 0.10562890896455872,
+      "grad_norm": 1.180397863768546,
+      "learning_rate": 1.688888888888889e-05,
+      "loss": 0.9982,
+      "step": 19
+    },
+    {
+      "epoch": 0.11118832522585129,
+      "grad_norm": 0.9555181178645678,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.9708,
+      "step": 20
+    },
+    {
+      "epoch": 0.11674774148714386,
+      "grad_norm": 1.0269402752613572,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.9798,
+      "step": 21
+    },
+    {
+      "epoch": 0.12230715774843641,
+      "grad_norm": 1.0708000870410468,
+      "learning_rate": 1.9555555555555557e-05,
+      "loss": 0.9699,
+      "step": 22
+    },
+    {
+      "epoch": 0.12786657400972898,
+      "grad_norm": 1.0055337623395493,
+      "learning_rate": 2.0444444444444446e-05,
+      "loss": 0.9427,
+      "step": 23
+    },
+    {
+      "epoch": 0.13342599027102153,
+      "grad_norm": 0.9229760705571243,
+      "learning_rate": 2.1333333333333335e-05,
+      "loss": 0.9471,
+      "step": 24
+    },
+    {
+      "epoch": 0.13898540653231412,
+      "grad_norm": 1.0000200092904565,
+      "learning_rate": 2.2222222222222227e-05,
+      "loss": 0.9358,
+      "step": 25
+    },
+    {
+      "epoch": 0.14454482279360667,
+      "grad_norm": 0.9339129127635016,
+      "learning_rate": 2.3111111111111112e-05,
+      "loss": 0.9482,
+      "step": 26
+    },
+    {
+      "epoch": 0.15010423905489922,
+      "grad_norm": 1.4607014103386948,
+      "learning_rate": 2.4e-05,
+      "loss": 0.9356,
+      "step": 27
+    },
+    {
+      "epoch": 0.1556636553161918,
+      "grad_norm": 0.8421272457585499,
+      "learning_rate": 2.4888888888888893e-05,
+      "loss": 0.9225,
+      "step": 28
+    },
+    {
+      "epoch": 0.16122307157748436,
+      "grad_norm": 1.1844488591141227,
+      "learning_rate": 2.577777777777778e-05,
+      "loss": 0.9203,
+      "step": 29
+    },
+    {
+      "epoch": 0.16678248783877692,
+      "grad_norm": 1.2189146868730585,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.927,
+      "step": 30
+    },
+    {
+      "epoch": 0.1723419041000695,
+      "grad_norm": 0.883273254038553,
+      "learning_rate": 2.755555555555556e-05,
+      "loss": 0.9089,
+      "step": 31
+    },
+    {
+      "epoch": 0.17790132036136205,
+      "grad_norm": 1.3537906784503,
+      "learning_rate": 2.8444444444444447e-05,
+      "loss": 0.8963,
+      "step": 32
+    },
+    {
+      "epoch": 0.1834607366226546,
+      "grad_norm": 0.8379832204564226,
+      "learning_rate": 2.9333333333333333e-05,
+      "loss": 0.8902,
+      "step": 33
+    },
+    {
+      "epoch": 0.1890201528839472,
+      "grad_norm": 1.0943609442451472,
+      "learning_rate": 3.0222222222222225e-05,
+      "loss": 0.9132,
+      "step": 34
+    },
+    {
+      "epoch": 0.19457956914523974,
+      "grad_norm": 1.317940648780496,
+      "learning_rate": 3.111111111111112e-05,
+      "loss": 0.8984,
+      "step": 35
+    },
+    {
+      "epoch": 0.20013898540653233,
+      "grad_norm": 0.8848172619144172,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.9093,
+      "step": 36
+    },
+    {
+      "epoch": 0.20569840166782488,
+      "grad_norm": 1.8109414393395322,
+      "learning_rate": 3.288888888888889e-05,
+      "loss": 0.9033,
+      "step": 37
+    },
+    {
+      "epoch": 0.21125781792911744,
+      "grad_norm": 1.2554883307684124,
+      "learning_rate": 3.377777777777778e-05,
+      "loss": 0.9143,
+      "step": 38
+    },
+    {
+      "epoch": 0.21681723419041002,
+      "grad_norm": 1.3729869755187412,
+      "learning_rate": 3.466666666666667e-05,
+      "loss": 0.9062,
+      "step": 39
+    },
+    {
+      "epoch": 0.22237665045170257,
+      "grad_norm": 1.3452250001878445,
+      "learning_rate": 3.555555555555555e-05,
+      "loss": 0.897,
+      "step": 40
+    },
+    {
+      "epoch": 0.22793606671299513,
+      "grad_norm": 1.4414798342602804,
+      "learning_rate": 3.644444444444445e-05,
+      "loss": 0.8827,
+      "step": 41
+    },
+    {
+      "epoch": 0.2334954829742877,
+      "grad_norm": 1.1308801320597703,
+      "learning_rate": 3.733333333333334e-05,
+      "loss": 0.8888,
+      "step": 42
+    },
+    {
+      "epoch": 0.23905489923558026,
+      "grad_norm": 1.5239027018360913,
+      "learning_rate": 3.8222222222222226e-05,
+      "loss": 0.8797,
+      "step": 43
+    },
+    {
+      "epoch": 0.24461431549687282,
+      "grad_norm": 1.712956853860375,
+      "learning_rate": 3.9111111111111115e-05,
+      "loss": 0.8739,
+      "step": 44
+    },
+    {
+      "epoch": 0.2501737317581654,
+      "grad_norm": 0.9900656403191864,
+      "learning_rate": 4e-05,
+      "loss": 0.8806,
+      "step": 45
+    },
+    {
+      "epoch": 0.25573314801945796,
+      "grad_norm": 1.6460944571950658,
+      "learning_rate": 4.088888888888889e-05,
+      "loss": 0.8763,
+      "step": 46
+    },
+    {
+      "epoch": 0.26129256428075054,
+      "grad_norm": 1.2766480112920402,
+      "learning_rate": 4.177777777777778e-05,
+      "loss": 0.8736,
+      "step": 47
+    },
+    {
+      "epoch": 0.26685198054204307,
+      "grad_norm": 1.3233659252241343,
+      "learning_rate": 4.266666666666667e-05,
+      "loss": 0.8542,
+      "step": 48
+    },
+    {
+      "epoch": 0.27241139680333565,
+      "grad_norm": 1.634661434972772,
+      "learning_rate": 4.355555555555556e-05,
+      "loss": 0.8695,
+      "step": 49
+    },
+    {
+      "epoch": 0.27797081306462823,
+      "grad_norm": 0.8670372531726727,
+      "learning_rate": 4.444444444444445e-05,
+      "loss": 0.8671,
+      "step": 50
+    },
+    {
+      "epoch": 0.28353022932592076,
+      "grad_norm": 1.8672796712712416,
+      "learning_rate": 4.5333333333333335e-05,
+      "loss": 0.8745,
+      "step": 51
+    },
+    {
+      "epoch": 0.28908964558721334,
+      "grad_norm": 1.1637792401354623,
+      "learning_rate": 4.6222222222222224e-05,
+      "loss": 0.8571,
+      "step": 52
+    },
+    {
+      "epoch": 0.2946490618485059,
+      "grad_norm": 2.2610360534030116,
+      "learning_rate": 4.711111111111112e-05,
+      "loss": 0.8766,
+      "step": 53
+    },
+    {
+      "epoch": 0.30020847810979845,
+      "grad_norm": 1.996143461729675,
+      "learning_rate": 4.8e-05,
+      "loss": 0.8775,
+      "step": 54
+    },
+    {
+      "epoch": 0.30576789437109103,
+      "grad_norm": 1.4081005708876793,
+      "learning_rate": 4.88888888888889e-05,
+      "loss": 0.8662,
+      "step": 55
+    },
+    {
+      "epoch": 0.3113273106323836,
+      "grad_norm": 3.134853339618935,
+      "learning_rate": 4.9777777777777785e-05,
+      "loss": 0.8539,
+      "step": 56
+    },
+    {
+      "epoch": 0.31688672689367614,
+      "grad_norm": 1.9084219748530682,
+      "learning_rate": 5.066666666666667e-05,
+      "loss": 0.8798,
+      "step": 57
+    },
+    {
+      "epoch": 0.3224461431549687,
+      "grad_norm": 3.5153547411633674,
+      "learning_rate": 5.155555555555556e-05,
+      "loss": 0.8634,
+      "step": 58
+    },
+    {
+      "epoch": 0.3280055594162613,
+      "grad_norm": 2.6122587526657894,
+      "learning_rate": 5.244444444444445e-05,
+      "loss": 0.848,
+      "step": 59
+    },
+    {
+      "epoch": 0.33356497567755383,
+      "grad_norm": 2.975460445642716,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 0.8518,
+      "step": 60
+    },
+    {
+      "epoch": 0.3391243919388464,
+      "grad_norm": 2.1222203908542396,
+      "learning_rate": 5.422222222222223e-05,
+      "loss": 0.874,
+      "step": 61
+    },
+    {
+      "epoch": 0.344683808200139,
+      "grad_norm": 2.796857287788194,
+      "learning_rate": 5.511111111111112e-05,
+      "loss": 0.8533,
+      "step": 62
+    },
+    {
+      "epoch": 0.3502432244614315,
+      "grad_norm": 2.0600145849078917,
+      "learning_rate": 5.6e-05,
+      "loss": 0.8874,
+      "step": 63
+    },
+    {
+      "epoch": 0.3558026407227241,
+      "grad_norm": 2.930616478920807,
+      "learning_rate": 5.6888888888888895e-05,
+      "loss": 0.872,
+      "step": 64
+    },
+    {
+      "epoch": 0.3613620569840167,
+      "grad_norm": 1.999313770783877,
+      "learning_rate": 5.777777777777778e-05,
+      "loss": 0.8517,
+      "step": 65
+    },
+    {
+      "epoch": 0.3669214732453092,
+      "grad_norm": 2.777297306851716,
+      "learning_rate": 5.8666666666666665e-05,
+      "loss": 0.8569,
+      "step": 66
+    },
+    {
+      "epoch": 0.3724808895066018,
+      "grad_norm": 2.2731615451600593,
+      "learning_rate": 5.955555555555556e-05,
+      "loss": 0.8616,
+      "step": 67
+    },
+    {
+      "epoch": 0.3780403057678944,
+      "grad_norm": 1.998295903749419,
+      "learning_rate": 6.044444444444445e-05,
+      "loss": 0.8543,
+      "step": 68
+    },
+    {
+      "epoch": 0.38359972202918696,
+      "grad_norm": 1.6203916734242618,
+      "learning_rate": 6.133333333333334e-05,
+      "loss": 0.8446,
+      "step": 69
+    },
+    {
+      "epoch": 0.3891591382904795,
+      "grad_norm": 1.76915202879457,
+      "learning_rate": 6.222222222222223e-05,
+      "loss": 0.8576,
+      "step": 70
+    },
+    {
+      "epoch": 0.39471855455177207,
+      "grad_norm": 1.3152162421151332,
+      "learning_rate": 6.311111111111112e-05,
+      "loss": 0.8563,
+      "step": 71
+    },
+    {
+      "epoch": 0.40027797081306465,
+      "grad_norm": 1.3643468996208965,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.8426,
+      "step": 72
+    },
+    {
+      "epoch": 0.4058373870743572,
+      "grad_norm": 1.908069980328843,
+      "learning_rate": 6.488888888888889e-05,
+      "loss": 0.8531,
+      "step": 73
+    },
+    {
+      "epoch": 0.41139680333564976,
+      "grad_norm": 1.685214282010214,
+      "learning_rate": 6.577777777777777e-05,
+      "loss": 0.8441,
+      "step": 74
+    },
+    {
+      "epoch": 0.41695621959694235,
+      "grad_norm": 1.1524590649427782,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.8496,
+      "step": 75
+    },
+    {
+      "epoch": 0.4225156358582349,
+      "grad_norm": 2.5121776075082476,
+      "learning_rate": 6.755555555555557e-05,
+      "loss": 0.8517,
+      "step": 76
+    },
+    {
+      "epoch": 0.42807505211952745,
+      "grad_norm": 2.010450049558264,
+      "learning_rate": 6.844444444444445e-05,
+      "loss": 0.8621,
+      "step": 77
+    },
+    {
+      "epoch": 0.43363446838082004,
+      "grad_norm": 1.6558455850767062,
+      "learning_rate": 6.933333333333334e-05,
+      "loss": 0.8535,
+      "step": 78
+    },
+    {
+      "epoch": 0.43919388464211256,
+      "grad_norm": 2.0116967221199697,
+      "learning_rate": 7.022222222222222e-05,
+      "loss": 0.8534,
+      "step": 79
+    },
+    {
+      "epoch": 0.44475330090340515,
+      "grad_norm": 1.2758552339008475,
+      "learning_rate": 7.11111111111111e-05,
+      "loss": 0.8475,
+      "step": 80
+    },
+    {
+      "epoch": 0.45031271716469773,
+      "grad_norm": 2.171298513799551,
+      "learning_rate": 7.2e-05,
+      "loss": 0.8557,
+      "step": 81
+    },
+    {
+      "epoch": 0.45587213342599026,
+      "grad_norm": 1.492569972617151,
+      "learning_rate": 7.28888888888889e-05,
+      "loss": 0.8599,
+      "step": 82
+    },
+    {
+      "epoch": 0.46143154968728284,
+      "grad_norm": 1.8374241836084444,
+      "learning_rate": 7.377777777777779e-05,
+      "loss": 0.8585,
+      "step": 83
+    },
+    {
+      "epoch": 0.4669909659485754,
+      "grad_norm": 1.7493375110413492,
+      "learning_rate": 7.466666666666667e-05,
+      "loss": 0.8403,
+      "step": 84
+    },
+    {
+      "epoch": 0.47255038220986795,
+      "grad_norm": 1.1188241074603504,
+      "learning_rate": 7.555555555555556e-05,
+      "loss": 0.8301,
+      "step": 85
+    },
+    {
+      "epoch": 0.47810979847116053,
+      "grad_norm": 2.1803527841540045,
+      "learning_rate": 7.644444444444445e-05,
+      "loss": 0.8579,
+      "step": 86
+    },
+    {
+      "epoch": 0.4836692147324531,
+      "grad_norm": 2.605263047586671,
+      "learning_rate": 7.733333333333333e-05,
+      "loss": 0.862,
+      "step": 87
+    },
+    {
+      "epoch": 0.48922863099374564,
+      "grad_norm": 1.3879012695020032,
+      "learning_rate": 7.822222222222223e-05,
+      "loss": 0.8344,
+      "step": 88
+    },
+    {
+      "epoch": 0.4947880472550382,
+      "grad_norm": 1.3012238789432584,
+      "learning_rate": 7.911111111111112e-05,
+      "loss": 0.8587,
+      "step": 89
+    },
+    {
+      "epoch": 0.5003474635163307,
+      "grad_norm": 2.513237057025125,
+      "learning_rate": 8e-05,
+      "loss": 0.868,
+      "step": 90
+    },
+    {
+      "epoch": 0.5059068797776234,
+      "grad_norm": 1.8055394762212944,
+      "learning_rate": 7.999969539471858e-05,
+      "loss": 0.8533,
+      "step": 91
+    },
+    {
+      "epoch": 0.5114662960389159,
+      "grad_norm": 1.5304364449758314,
+      "learning_rate": 7.999878158351353e-05,
+      "loss": 0.8452,
+      "step": 92
+    },
+    {
+      "epoch": 0.5170257123002084,
+      "grad_norm": 2.1193756766009084,
+      "learning_rate": 7.999725858030245e-05,
+      "loss": 0.864,
+      "step": 93
+    },
+    {
+      "epoch": 0.5225851285615011,
+      "grad_norm": 1.2380452088121536,
+      "learning_rate": 7.999512640828105e-05,
+      "loss": 0.8458,
+      "step": 94
+    },
+    {
+      "epoch": 0.5281445448227936,
+      "grad_norm": 2.058829749102372,
+      "learning_rate": 7.999238509992291e-05,
+      "loss": 0.8558,
+      "step": 95
+    },
+    {
+      "epoch": 0.5337039610840861,
+      "grad_norm": 1.6580103531528805,
+      "learning_rate": 7.998903469697887e-05,
+      "loss": 0.8646,
+      "step": 96
+    },
+    {
+      "epoch": 0.5392633773453788,
+      "grad_norm": 1.5955718716301766,
+      "learning_rate": 7.998507525047644e-05,
+      "loss": 0.8436,
+      "step": 97
+    },
+    {
+      "epoch": 0.5448227936066713,
+      "grad_norm": 1.4398145473558983,
+      "learning_rate": 7.998050682071902e-05,
+      "loss": 0.8497,
+      "step": 98
+    },
+    {
+      "epoch": 0.5503822098679638,
+      "grad_norm": 1.3793681111087452,
+      "learning_rate": 7.997532947728505e-05,
+      "loss": 0.8381,
+      "step": 99
+    },
+    {
+      "epoch": 0.5559416261292565,
+      "grad_norm": 1.4240010090800914,
+      "learning_rate": 7.99695432990268e-05,
+      "loss": 0.8624,
+      "step": 100
+    },
+    {
+      "epoch": 0.561501042390549,
+      "grad_norm": 1.9629255774107022,
+      "learning_rate": 7.99631483740693e-05,
+      "loss": 0.8387,
+      "step": 101
+    },
+    {
+      "epoch": 0.5670604586518415,
+      "grad_norm": 1.177962963776449,
+      "learning_rate": 7.995614479980896e-05,
+      "loss": 0.845,
+      "step": 102
+    },
+    {
+      "epoch": 0.5726198749131342,
+      "grad_norm": 1.5104993823032296,
+      "learning_rate": 7.994853268291205e-05,
+      "loss": 0.8377,
+      "step": 103
+    },
+    {
+      "epoch": 0.5781792911744267,
+      "grad_norm": 1.3395795088646747,
+      "learning_rate": 7.994031213931312e-05,
+      "loss": 0.8435,
+      "step": 104
+    },
+    {
+      "epoch": 0.5837387074357192,
+      "grad_norm": 1.5424073475176412,
+      "learning_rate": 7.993148329421323e-05,
+      "loss": 0.8447,
+      "step": 105
+    },
+    {
+      "epoch": 0.5892981236970118,
+      "grad_norm": 1.944768632226559,
+      "learning_rate": 7.992204628207802e-05,
+      "loss": 0.846,
+      "step": 106
+    },
+    {
+      "epoch": 0.5948575399583044,
+      "grad_norm": 1.1063119498625391,
+      "learning_rate": 7.991200124663568e-05,
+      "loss": 0.8304,
+      "step": 107
+    },
+    {
+      "epoch": 0.6004169562195969,
+      "grad_norm": 2.3248288094528635,
+      "learning_rate": 7.990134834087473e-05,
+      "loss": 0.8473,
+      "step": 108
+    },
+    {
+      "epoch": 0.6059763724808895,
+      "grad_norm": 1.402163201884845,
+      "learning_rate": 7.989008772704177e-05,
+      "loss": 0.8383,
+      "step": 109
+    },
+    {
+      "epoch": 0.6115357887421821,
+      "grad_norm": 1.7492854215090492,
+      "learning_rate": 7.987821957663889e-05,
+      "loss": 0.8505,
+      "step": 110
+    },
+    {
+      "epoch": 0.6170952050034746,
+      "grad_norm": 1.2887514260164146,
+      "learning_rate": 7.986574407042118e-05,
+      "loss": 0.8327,
+      "step": 111
+    },
+    {
+      "epoch": 0.6226546212647672,
+      "grad_norm": 1.307242590555259,
+      "learning_rate": 7.985266139839389e-05,
+      "loss": 0.8525,
+      "step": 112
+    },
+    {
+      "epoch": 0.6282140375260598,
+      "grad_norm": 1.6231277109038205,
+      "learning_rate": 7.983897175980957e-05,
+      "loss": 0.8385,
+      "step": 113
+    },
+    {
+      "epoch": 0.6337734537873523,
+      "grad_norm": 1.3006915329623785,
+      "learning_rate": 7.982467536316502e-05,
+      "loss": 0.8359,
+      "step": 114
+    },
+    {
+      "epoch": 0.6393328700486449,
+      "grad_norm": 1.7752541708560459,
+      "learning_rate": 7.980977242619814e-05,
+      "loss": 0.8497,
+      "step": 115
+    },
+    {
+      "epoch": 0.6448922863099374,
+      "grad_norm": 0.963970001633594,
+      "learning_rate": 7.97942631758846e-05,
+      "loss": 0.8348,
+      "step": 116
+    },
+    {
+      "epoch": 0.65045170257123,
+      "grad_norm": 1.8279125029062835,
+      "learning_rate": 7.977814784843438e-05,
+      "loss": 0.8291,
+      "step": 117
+    },
+    {
+      "epoch": 0.6560111188325226,
+      "grad_norm": 1.1016041521938733,
+      "learning_rate": 7.976142668928815e-05,
+      "loss": 0.837,
+      "step": 118
+    },
+    {
+      "epoch": 0.6615705350938151,
+      "grad_norm": 1.9123138383948801,
+      "learning_rate": 7.97440999531136e-05,
+      "loss": 0.8284,
+      "step": 119
+    },
+    {
+      "epoch": 0.6671299513551077,
+      "grad_norm": 1.4770547180557922,
+      "learning_rate": 7.972616790380151e-05,
+      "loss": 0.8498,
+      "step": 120
+    },
+    {
+      "epoch": 0.6726893676164003,
+      "grad_norm": 1.4177848951880787,
+      "learning_rate": 7.970763081446172e-05,
+      "loss": 0.834,
+      "step": 121
+    },
+    {
+      "epoch": 0.6782487838776928,
+      "grad_norm": 1.4568355456555888,
+      "learning_rate": 7.968848896741896e-05,
+      "loss": 0.8436,
+      "step": 122
+    },
+    {
+      "epoch": 0.6838082001389854,
+      "grad_norm": 1.417327263712158,
+      "learning_rate": 7.966874265420866e-05,
+      "loss": 0.8416,
+      "step": 123
+    },
+    {
+      "epoch": 0.689367616400278,
+      "grad_norm": 1.1607989065407163,
+      "learning_rate": 7.964839217557237e-05,
+      "loss": 0.8487,
+      "step": 124
+    },
+    {
+      "epoch": 0.6949270326615705,
+      "grad_norm": 1.1655546055699306,
+      "learning_rate": 7.962743784145323e-05,
+      "loss": 0.8309,
+      "step": 125
+    },
+    {
+      "epoch": 0.700486448922863,
+      "grad_norm": 1.6283722594753118,
+      "learning_rate": 7.960587997099132e-05,
+      "loss": 0.8476,
+      "step": 126
+    },
+    {
+      "epoch": 0.7060458651841557,
+      "grad_norm": 0.855140445880346,
+      "learning_rate": 7.958371889251868e-05,
+      "loss": 0.8167,
+      "step": 127
+    },
+    {
+      "epoch": 0.7116052814454482,
+      "grad_norm": 1.181067166006884,
+      "learning_rate": 7.956095494355438e-05,
+      "loss": 0.8273,
+      "step": 128
+    },
+    {
+      "epoch": 0.7171646977067407,
+      "grad_norm": 1.3100507927777372,
+      "learning_rate": 7.95375884707994e-05,
+      "loss": 0.8438,
+      "step": 129
+    },
+    {
+      "epoch": 0.7227241139680334,
+      "grad_norm": 1.6550084905324673,
+      "learning_rate": 7.951361983013127e-05,
+      "loss": 0.8377,
+      "step": 130
+    },
+    {
+      "epoch": 0.7282835302293259,
+      "grad_norm": 1.2643405442051172,
+      "learning_rate": 7.948904938659872e-05,
+      "loss": 0.8449,
+      "step": 131
+    },
+    {
+      "epoch": 0.7338429464906184,
+      "grad_norm": 1.2710601497668057,
+      "learning_rate": 7.946387751441609e-05,
+      "loss": 0.8404,
+      "step": 132
+    },
+    {
+      "epoch": 0.7394023627519111,
+      "grad_norm": 1.4879557131368137,
+      "learning_rate": 7.943810459695764e-05,
+      "loss": 0.8242,
+      "step": 133
+    },
+    {
+      "epoch": 0.7449617790132036,
+      "grad_norm": 1.1635532204185386,
+      "learning_rate": 7.941173102675172e-05,
+      "loss": 0.8153,
+      "step": 134
+    },
+    {
+      "epoch": 0.7505211952744962,
+      "grad_norm": 1.2694310707160308,
+      "learning_rate": 7.938475720547477e-05,
+      "loss": 0.8372,
+      "step": 135
+    },
+    {
+      "epoch": 0.7560806115357888,
+      "grad_norm": 1.071426149452144,
+      "learning_rate": 7.93571835439452e-05,
+      "loss": 0.8311,
+      "step": 136
+    },
+    {
+      "epoch": 0.7616400277970813,
+      "grad_norm": 1.561376324467928,
+      "learning_rate": 7.932901046211715e-05,
+      "loss": 0.8182,
+      "step": 137
+    },
+    {
+      "epoch": 0.7671994440583739,
+      "grad_norm": 0.9569705113881821,
+      "learning_rate": 7.930023838907411e-05,
+      "loss": 0.8285,
+      "step": 138
+    },
+    {
+      "epoch": 0.7727588603196665,
+      "grad_norm": 1.3769869224499816,
+      "learning_rate": 7.927086776302235e-05,
+      "loss": 0.8223,
+      "step": 139
+    },
+    {
+      "epoch": 0.778318276580959,
+      "grad_norm": 1.0661634612442863,
+      "learning_rate": 7.924089903128425e-05,
+      "loss": 0.821,
+      "step": 140
+    },
+    {
+      "epoch": 0.7838776928422516,
+      "grad_norm": 1.7150494701801486,
+      "learning_rate": 7.921033265029153e-05,
+      "loss": 0.8318,
+      "step": 141
+    },
+    {
+      "epoch": 0.7894371091035441,
+      "grad_norm": 1.1760603467329838,
+      "learning_rate": 7.917916908557822e-05,
+      "loss": 0.8141,
+      "step": 142
+    },
+    {
+      "epoch": 0.7949965253648367,
+      "grad_norm": 1.195770766355331,
+      "learning_rate": 7.914740881177365e-05,
+      "loss": 0.8086,
+      "step": 143
+    },
+    {
+      "epoch": 0.8005559416261293,
+      "grad_norm": 1.0311554098100406,
+      "learning_rate": 7.911505231259516e-05,
+      "loss": 0.8082,
+      "step": 144
+    },
+    {
+      "epoch": 0.8061153578874218,
+      "grad_norm": 1.6912543359730465,
+      "learning_rate": 7.908210008084082e-05,
+      "loss": 0.8023,
+      "step": 145
+    },
+    {
+      "epoch": 0.8116747741487144,
+      "grad_norm": 1.0813659299032288,
+      "learning_rate": 7.904855261838179e-05,
+      "loss": 0.8315,
+      "step": 146
+    },
+    {
+      "epoch": 0.817234190410007,
+      "grad_norm": 1.5581335270047423,
+      "learning_rate": 7.901441043615479e-05,
+      "loss": 0.8268,
+      "step": 147
+    },
+    {
+      "epoch": 0.8227936066712995,
+      "grad_norm": 1.00596454320661,
+      "learning_rate": 7.897967405415425e-05,
+      "loss": 0.8386,
+      "step": 148
+    },
+    {
+      "epoch": 0.828353022932592,
+      "grad_norm": 1.144194212812256,
+      "learning_rate": 7.894434400142447e-05,
+      "loss": 0.8325,
+      "step": 149
+    },
+    {
+      "epoch": 0.8339124391938847,
+      "grad_norm": 1.1625675356230636,
+      "learning_rate": 7.890842081605148e-05,
+      "loss": 0.8504,
+      "step": 150
+    },
+    {
+      "epoch": 0.8394718554551772,
+      "grad_norm": 1.1297535406659176,
+      "learning_rate": 7.887190504515485e-05,
+      "loss": 0.8075,
+      "step": 151
+    },
+    {
+      "epoch": 0.8450312717164697,
+      "grad_norm": 1.2053386499038343,
+      "learning_rate": 7.883479724487946e-05,
+      "loss": 0.8247,
+      "step": 152
+    },
+    {
+      "epoch": 0.8505906879777624,
+      "grad_norm": 1.1106411060998684,
+      "learning_rate": 7.879709798038685e-05,
+      "loss": 0.8298,
+      "step": 153
+    },
+    {
+      "epoch": 0.8561501042390549,
+      "grad_norm": 1.0971069205166395,
+      "learning_rate": 7.875880782584683e-05,
+      "loss": 0.8273,
+      "step": 154
+    },
+    {
+      "epoch": 0.8617095205003474,
+      "grad_norm": 1.159414773344552,
+      "learning_rate": 7.871992736442852e-05,
+      "loss": 0.8326,
+      "step": 155
+    },
+    {
+      "epoch": 0.8672689367616401,
+      "grad_norm": 1.4910863801494305,
+      "learning_rate": 7.868045718829166e-05,
+      "loss": 0.8263,
+      "step": 156
+    },
+    {
+      "epoch": 0.8728283530229326,
+      "grad_norm": 0.8170303946123546,
+      "learning_rate": 7.864039789857743e-05,
+      "loss": 0.8193,
+      "step": 157
+    },
+    {
+      "epoch": 0.8783877692842251,
+      "grad_norm": 0.8610983641895286,
+      "learning_rate": 7.859975010539937e-05,
+      "loss": 0.8119,
+      "step": 158
+    },
+    {
+      "epoch": 0.8839471855455178,
+      "grad_norm": 0.9841408044834976,
+      "learning_rate": 7.855851442783414e-05,
+      "loss": 0.8291,
+      "step": 159
+    },
+    {
+      "epoch": 0.8895066018068103,
+      "grad_norm": 1.9526023275945634,
+      "learning_rate": 7.851669149391198e-05,
+      "loss": 0.8103,
+      "step": 160
+    },
+    {
+      "epoch": 0.8950660180681028,
+      "grad_norm": 0.7191259063946382,
+      "learning_rate": 7.847428194060722e-05,
+      "loss": 0.8214,
+      "step": 161
+    },
+    {
+      "epoch": 0.9006254343293955,
+      "grad_norm": 1.8842244115014957,
+      "learning_rate": 7.843128641382856e-05,
+      "loss": 0.8198,
+      "step": 162
+    },
+    {
+      "epoch": 0.906184850590688,
+      "grad_norm": 1.0514604879384255,
+      "learning_rate": 7.838770556840923e-05,
+      "loss": 0.821,
+      "step": 163
+    },
+    {
+      "epoch": 0.9117442668519805,
+      "grad_norm": 1.6003009095500897,
+      "learning_rate": 7.834354006809699e-05,
+      "loss": 0.8396,
+      "step": 164
+    },
+    {
+      "epoch": 0.9173036831132731,
+      "grad_norm": 1.2604980474829044,
+      "learning_rate": 7.829879058554411e-05,
+      "loss": 0.8169,
+      "step": 165
+    },
+    {
+      "epoch": 0.9228630993745657,
+      "grad_norm": 1.3754424425834815,
+      "learning_rate": 7.8253457802297e-05,
+      "loss": 0.8248,
+      "step": 166
+    },
+    {
+      "epoch": 0.9284225156358582,
+      "grad_norm": 1.2970707388137186,
+      "learning_rate": 7.820754240878593e-05,
+      "loss": 0.8242,
+      "step": 167
+    },
+    {
+      "epoch": 0.9339819318971508,
+      "grad_norm": 1.1094402817946039,
+      "learning_rate": 7.816104510431446e-05,
+      "loss": 0.821,
+      "step": 168
+    },
+    {
+      "epoch": 0.9395413481584434,
+      "grad_norm": 1.0619969692986742,
+      "learning_rate": 7.811396659704884e-05,
+      "loss": 0.7994,
+      "step": 169
+    },
+    {
+      "epoch": 0.9451007644197359,
+      "grad_norm": 1.268154938816547,
+      "learning_rate": 7.806630760400712e-05,
+      "loss": 0.8231,
+      "step": 170
+    },
+    {
+      "epoch": 0.9506601806810285,
+      "grad_norm": 0.8114888834391076,
+      "learning_rate": 7.80180688510484e-05,
+      "loss": 0.8138,
+      "step": 171
+    },
+    {
+      "epoch": 0.9562195969423211,
+      "grad_norm": 1.1353293038885697,
+      "learning_rate": 7.79692510728616e-05,
+      "loss": 0.8363,
+      "step": 172
+    },
+    {
+      "epoch": 0.9617790132036136,
+      "grad_norm": 1.1710537031908355,
+      "learning_rate": 7.791985501295437e-05,
+      "loss": 0.8124,
+      "step": 173
+    },
+    {
+      "epoch": 0.9673384294649062,
+      "grad_norm": 0.9925849663427073,
+      "learning_rate": 7.786988142364175e-05,
+      "loss": 0.8062,
+      "step": 174
+    },
+    {
+      "epoch": 0.9728978457261988,
+      "grad_norm": 0.9839146497835408,
+      "learning_rate": 7.781933106603472e-05,
+      "loss": 0.8202,
+      "step": 175
+    },
+    {
+      "epoch": 0.9784572619874913,
+      "grad_norm": 1.4171846242465553,
+      "learning_rate": 7.776820471002856e-05,
+      "loss": 0.7936,
+      "step": 176
+    },
+    {
+      "epoch": 0.9840166782487839,
+      "grad_norm": 0.7277915858989548,
+      "learning_rate": 7.771650313429117e-05,
+      "loss": 0.8091,
+      "step": 177
+    },
+    {
+      "epoch": 0.9895760945100764,
+      "grad_norm": 0.8641953694944955,
+      "learning_rate": 7.76642271262512e-05,
+      "loss": 0.8146,
+      "step": 178
+    },
+    {
+      "epoch": 0.995135510771369,
+      "grad_norm": 1.334464922937482,
+      "learning_rate": 7.761137748208606e-05,
+      "loss": 0.7959,
+      "step": 179
+    },
+    {
+      "epoch": 1.0006949270326615,
+      "grad_norm": 0.8723143516212564,
+      "learning_rate": 7.75579550067098e-05,
+      "loss": 0.8144,
+      "step": 180
+    },
+    {
+      "epoch": 1.0062543432939541,
+      "grad_norm": 0.9676877145889436,
+      "learning_rate": 7.750396051376082e-05,
+      "loss": 0.7972,
+      "step": 181
+    },
+    {
+      "epoch": 1.0118137595552468,
+      "grad_norm": 0.8078199257078474,
+      "learning_rate": 7.74493948255895e-05,
+      "loss": 0.7873,
+      "step": 182
+    },
+    {
+      "epoch": 1.0173731758165392,
+      "grad_norm": 0.9618995502446188,
+      "learning_rate": 7.739425877324567e-05,
+      "loss": 0.806,
+      "step": 183
+    },
+    {
+      "epoch": 1.0229325920778318,
+      "grad_norm": 1.1903105794790574,
+      "learning_rate": 7.733855319646598e-05,
+      "loss": 0.8063,
+      "step": 184
+    },
+    {
+      "epoch": 1.0284920083391245,
+      "grad_norm": 1.113260237892017,
+      "learning_rate": 7.72822789436611e-05,
+      "loss": 0.8055,
+      "step": 185
+    },
+    {
+      "epoch": 1.0340514246004169,
+      "grad_norm": 1.2466501606809175,
+      "learning_rate": 7.722543687190271e-05,
+      "loss": 0.8,
+      "step": 186
+    },
+    {
+      "epoch": 1.0396108408617095,
+      "grad_norm": 0.8935894806107648,
+      "learning_rate": 7.716802784691061e-05,
+      "loss": 0.7886,
+      "step": 187
+    },
+    {
+      "epoch": 1.0451702571230022,
+      "grad_norm": 1.0645110228278147,
+      "learning_rate": 7.711005274303939e-05,
+      "loss": 0.8025,
+      "step": 188
+    },
+    {
+      "epoch": 1.0507296733842946,
+      "grad_norm": 0.655899538337792,
+      "learning_rate": 7.70515124432652e-05,
+      "loss": 0.7939,
+      "step": 189
+    },
+    {
+      "epoch": 1.0562890896455872,
+      "grad_norm": 0.7149867223741726,
+      "learning_rate": 7.699240783917226e-05,
+      "loss": 0.7819,
+      "step": 190
+    },
+    {
+      "epoch": 1.0618485059068798,
+      "grad_norm": 0.9227738003310058,
+      "learning_rate": 7.693273983093932e-05,
+      "loss": 0.7815,
+      "step": 191
+    },
+    {
+      "epoch": 1.0674079221681723,
+      "grad_norm": 1.1743470172337884,
+      "learning_rate": 7.687250932732587e-05,
+      "loss": 0.8034,
+      "step": 192
+    },
+    {
+      "epoch": 1.072967338429465,
+      "grad_norm": 1.1326402294789017,
+      "learning_rate": 7.681171724565841e-05,
+      "loss": 0.8039,
+      "step": 193
+    },
+    {
+      "epoch": 1.0785267546907575,
+      "grad_norm": 0.981410769406106,
+      "learning_rate": 7.675036451181638e-05,
+      "loss": 0.7856,
+      "step": 194
+    },
+    {
+      "epoch": 1.08408617095205,
+      "grad_norm": 0.8619461846062771,
+      "learning_rate": 7.668845206021812e-05,
+      "loss": 0.7938,
+      "step": 195
+    },
+    {
+      "epoch": 1.0896455872133426,
+      "grad_norm": 1.0839380849451241,
+      "learning_rate": 7.662598083380664e-05,
+      "loss": 0.7848,
+      "step": 196
+    },
+    {
+      "epoch": 1.0952050034746352,
+      "grad_norm": 1.0392746014030274,
+      "learning_rate": 7.656295178403519e-05,
+      "loss": 0.7911,
+      "step": 197
+    },
+    {
+      "epoch": 1.1007644197359276,
+      "grad_norm": 1.0359884367143106,
+      "learning_rate": 7.649936587085285e-05,
+      "loss": 0.7861,
+      "step": 198
+    },
+    {
+      "epoch": 1.1063238359972203,
+      "grad_norm": 1.0028428027595355,
+      "learning_rate": 7.643522406268985e-05,
+      "loss": 0.7883,
+      "step": 199
+    },
+    {
+      "epoch": 1.111883252258513,
+      "grad_norm": 0.8262483797533237,
+      "learning_rate": 7.63705273364429e-05,
+      "loss": 0.7801,
+      "step": 200
+    },
+    {
+      "epoch": 1.1174426685198053,
+      "grad_norm": 0.793966822779039,
+      "learning_rate": 7.630527667746022e-05,
+      "loss": 0.7808,
+      "step": 201
+    },
+    {
+      "epoch": 1.123002084781098,
+      "grad_norm": 0.5841255495857546,
+      "learning_rate": 7.623947307952655e-05,
+      "loss": 0.7875,
+      "step": 202
+    },
+    {
+      "epoch": 1.1285615010423906,
+      "grad_norm": 0.641770336653931,
+      "learning_rate": 7.617311754484809e-05,
+      "loss": 0.7769,
+      "step": 203
+    },
+    {
+      "epoch": 1.134120917303683,
+      "grad_norm": 0.49349990660077014,
+      "learning_rate": 7.610621108403714e-05,
+      "loss": 0.7812,
+      "step": 204
+    },
+    {
+      "epoch": 1.1396803335649757,
+      "grad_norm": 0.7063714078210745,
+      "learning_rate": 7.603875471609677e-05,
+      "loss": 0.7771,
+      "step": 205
+    },
+    {
+      "epoch": 1.1452397498262683,
+      "grad_norm": 0.6252197448112845,
+      "learning_rate": 7.59707494684053e-05,
+      "loss": 0.7807,
+      "step": 206
+    },
+    {
+      "epoch": 1.1507991660875607,
+      "grad_norm": 0.5127771372908517,
+      "learning_rate": 7.590219637670055e-05,
+      "loss": 0.7707,
+      "step": 207
+    },
+    {
+      "epoch": 1.1563585823488534,
+      "grad_norm": 0.7846508910498406,
+      "learning_rate": 7.583309648506429e-05,
+      "loss": 0.7866,
+      "step": 208
+    },
+    {
+      "epoch": 1.161917998610146,
+      "grad_norm": 0.6435801852453142,
+      "learning_rate": 7.576345084590606e-05,
+      "loss": 0.7807,
+      "step": 209
+    },
+    {
+      "epoch": 1.1674774148714384,
+      "grad_norm": 0.6129447763782494,
+      "learning_rate": 7.569326051994735e-05,
+      "loss": 0.7947,
+      "step": 210
+    },
+    {
+      "epoch": 1.173036831132731,
+      "grad_norm": 0.7393929046755997,
+      "learning_rate": 7.562252657620536e-05,
+      "loss": 0.7898,
+      "step": 211
+    },
+    {
+      "epoch": 1.1785962473940237,
+      "grad_norm": 0.7875515383471235,
+      "learning_rate": 7.555125009197675e-05,
+      "loss": 0.7749,
+      "step": 212
+    },
+    {
+      "epoch": 1.184155663655316,
+      "grad_norm": 0.8502781736276134,
+      "learning_rate": 7.547943215282116e-05,
+      "loss": 0.796,
+      "step": 213
+    },
+    {
+      "epoch": 1.1897150799166087,
+      "grad_norm": 0.9381926799528068,
+      "learning_rate": 7.54070738525448e-05,
+      "loss": 0.7866,
+      "step": 214
+    },
+    {
+      "epoch": 1.1952744961779014,
+      "grad_norm": 1.1151942570209228,
+      "learning_rate": 7.533417629318367e-05,
+      "loss": 0.7807,
+      "step": 215
+    },
+    {
+      "epoch": 1.2008339124391938,
+      "grad_norm": 1.2041485011685347,
+      "learning_rate": 7.526074058498687e-05,
+      "loss": 0.7905,
+      "step": 216
+    },
+    {
+      "epoch": 1.2063933287004864,
+      "grad_norm": 0.9555182120392091,
+      "learning_rate": 7.518676784639962e-05,
+      "loss": 0.7931,
+      "step": 217
+    },
+    {
+      "epoch": 1.211952744961779,
+      "grad_norm": 1.0723231936146647,
+      "learning_rate": 7.511225920404626e-05,
+      "loss": 0.7937,
+      "step": 218
+    },
+    {
+      "epoch": 1.2175121612230715,
+      "grad_norm": 1.026526722316571,
+      "learning_rate": 7.503721579271308e-05,
+      "loss": 0.7878,
+      "step": 219
+    },
+    {
+      "epoch": 1.2230715774843641,
+      "grad_norm": 0.9506008956242211,
+      "learning_rate": 7.496163875533105e-05,
+      "loss": 0.7697,
+      "step": 220
+    },
+    {
+      "epoch": 1.2286309937456568,
+      "grad_norm": 1.131493716226611,
+      "learning_rate": 7.488552924295843e-05,
+      "loss": 0.7785,
+      "step": 221
+    },
+    {
+      "epoch": 1.2341904100069492,
+      "grad_norm": 0.7811448856254682,
+      "learning_rate": 7.480888841476319e-05,
+      "loss": 0.7744,
+      "step": 222
+    },
+    {
+      "epoch": 1.2397498262682418,
+      "grad_norm": 0.5750988813544421,
+      "learning_rate": 7.473171743800535e-05,
+      "loss": 0.7697,
+      "step": 223
+    },
+    {
+      "epoch": 1.2453092425295345,
+      "grad_norm": 0.6043177617706299,
+      "learning_rate": 7.46540174880193e-05,
+      "loss": 0.7774,
+      "step": 224
+    },
+    {
+      "epoch": 1.2508686587908269,
+      "grad_norm": 0.6185743964153164,
+      "learning_rate": 7.457578974819578e-05,
+      "loss": 0.7745,
+      "step": 225
+    },
+    {
+      "epoch": 1.2564280750521195,
+      "grad_norm": 0.5315823047799536,
+      "learning_rate": 7.449703540996393e-05,
+      "loss": 0.7818,
+      "step": 226
+    },
+    {
+      "epoch": 1.2619874913134121,
+      "grad_norm": 0.4889499415667462,
+      "learning_rate": 7.44177556727731e-05,
+      "loss": 0.7811,
+      "step": 227
+    },
+    {
+      "epoch": 1.2675469075747046,
+      "grad_norm": 0.6576085919723266,
+      "learning_rate": 7.433795174407465e-05,
+      "loss": 0.7887,
+      "step": 228
+    },
+    {
+      "epoch": 1.2731063238359972,
+      "grad_norm": 0.7111593144278238,
+      "learning_rate": 7.425762483930347e-05,
+      "loss": 0.7748,
+      "step": 229
+    },
+    {
+      "epoch": 1.2786657400972898,
+      "grad_norm": 0.6484637513039345,
+      "learning_rate": 7.417677618185955e-05,
+      "loss": 0.776,
+      "step": 230
+    },
+    {
+      "epoch": 1.2842251563585823,
+      "grad_norm": 0.6761213806310967,
+      "learning_rate": 7.409540700308927e-05,
+      "loss": 0.7906,
+      "step": 231
+    },
+    {
+      "epoch": 1.289784572619875,
+      "grad_norm": 0.72586401747787,
+      "learning_rate": 7.401351854226673e-05,
+      "loss": 0.7863,
+      "step": 232
+    },
+    {
+      "epoch": 1.2953439888811675,
+      "grad_norm": 0.8761480055698481,
+      "learning_rate": 7.393111204657478e-05,
+      "loss": 0.7808,
+      "step": 233
+    },
+    {
+      "epoch": 1.30090340514246,
+      "grad_norm": 1.155943642457972,
+      "learning_rate": 7.384818877108617e-05,
+      "loss": 0.7679,
+      "step": 234
+    },
+    {
+      "epoch": 1.3064628214037526,
+      "grad_norm": 0.8100094228240489,
+      "learning_rate": 7.376474997874422e-05,
+      "loss": 0.7976,
+      "step": 235
+    },
+    {
+      "epoch": 1.3120222376650452,
+      "grad_norm": 0.5445781974209539,
+      "learning_rate": 7.368079694034381e-05,
+      "loss": 0.7746,
+      "step": 236
+    },
+    {
+      "epoch": 1.3175816539263376,
+      "grad_norm": 0.6679063542986605,
+      "learning_rate": 7.359633093451187e-05,
+      "loss": 0.7883,
+      "step": 237
+    },
+    {
+      "epoch": 1.3231410701876303,
+      "grad_norm": 0.958523219858762,
+      "learning_rate": 7.351135324768797e-05,
+      "loss": 0.7757,
+      "step": 238
+    },
+    {
+      "epoch": 1.328700486448923,
+      "grad_norm": 1.1832452619017166,
+      "learning_rate": 7.342586517410473e-05,
+      "loss": 0.7866,
+      "step": 239
+    },
+    {
+      "epoch": 1.3342599027102153,
+      "grad_norm": 0.6620192952316521,
+      "learning_rate": 7.333986801576808e-05,
+      "loss": 0.7787,
+      "step": 240
+    },
+    {
+      "epoch": 1.339819318971508,
+      "grad_norm": 0.5706874893778034,
+      "learning_rate": 7.325336308243747e-05,
+      "loss": 0.7753,
+      "step": 241
+    },
+    {
+      "epoch": 1.3453787352328006,
+      "grad_norm": 0.7293610495624486,
+      "learning_rate": 7.316635169160585e-05,
+      "loss": 0.779,
+      "step": 242
+    },
+    {
+      "epoch": 1.350938151494093,
+      "grad_norm": 0.9230133745177531,
+      "learning_rate": 7.307883516847968e-05,
+      "loss": 0.7855,
+      "step": 243
+    },
+    {
+      "epoch": 1.3564975677553857,
+      "grad_norm": 0.8597946100934724,
+      "learning_rate": 7.299081484595874e-05,
+      "loss": 0.7712,
+      "step": 244
+    },
+    {
+      "epoch": 1.3620569840166783,
+      "grad_norm": 0.7770773303527336,
+      "learning_rate": 7.290229206461578e-05,
+      "loss": 0.7765,
+      "step": 245
+    },
+    {
+      "epoch": 1.3676164002779707,
+      "grad_norm": 1.018019133526188,
+      "learning_rate": 7.281326817267612e-05,
+      "loss": 0.7827,
+      "step": 246
+    },
+    {
+      "epoch": 1.3731758165392633,
+      "grad_norm": 1.2608059715209885,
+      "learning_rate": 7.272374452599717e-05,
+      "loss": 0.7801,
+      "step": 247
+    },
+    {
+      "epoch": 1.378735232800556,
+      "grad_norm": 0.5427742215121941,
+      "learning_rate": 7.263372248804768e-05,
+      "loss": 0.769,
+      "step": 248
+    },
+    {
+      "epoch": 1.3842946490618484,
+      "grad_norm": 0.9707313981078285,
+      "learning_rate": 7.254320342988707e-05,
+      "loss": 0.7879,
+      "step": 249
+    },
+    {
+      "epoch": 1.389854065323141,
+      "grad_norm": 1.3746763366179666,
+      "learning_rate": 7.245218873014451e-05,
+      "loss": 0.8022,
+      "step": 250
+    },
+    {
+      "epoch": 1.3954134815844337,
+      "grad_norm": 0.8168605853299865,
+      "learning_rate": 7.236067977499791e-05,
+      "loss": 0.7879,
+      "step": 251
+    },
+    {
+      "epoch": 1.400972897845726,
+      "grad_norm": 1.5862041867625931,
+      "learning_rate": 7.22686779581528e-05,
+      "loss": 0.787,
+      "step": 252
+    },
+    {
+      "epoch": 1.4065323141070187,
+      "grad_norm": 0.6237034602452829,
+      "learning_rate": 7.217618468082118e-05,
+      "loss": 0.7659,
+      "step": 253
+    },
+    {
+      "epoch": 1.4120917303683114,
+      "grad_norm": 1.248354943849073,
+      "learning_rate": 7.208320135170004e-05,
+      "loss": 0.7944,
+      "step": 254
+    },
+    {
+      "epoch": 1.4176511466296038,
+      "grad_norm": 0.5870554519430948,
+      "learning_rate": 7.19897293869501e-05,
+      "loss": 0.7942,
+      "step": 255
+    },
+    {
+      "epoch": 1.4232105628908964,
+      "grad_norm": 1.2147306579223598,
+      "learning_rate": 7.189577021017402e-05,
+      "loss": 0.7768,
+      "step": 256
+    },
+    {
+      "epoch": 1.428769979152189,
+      "grad_norm": 0.5689684573208176,
+      "learning_rate": 7.180132525239488e-05,
+      "loss": 0.7777,
+      "step": 257
+    },
+    {
+      "epoch": 1.4343293954134815,
+      "grad_norm": 0.891498824230005,
+      "learning_rate": 7.170639595203434e-05,
+      "loss": 0.7697,
+      "step": 258
+    },
+    {
+      "epoch": 1.4398888116747741,
+      "grad_norm": 0.5981417552374583,
+      "learning_rate": 7.16109837548907e-05,
+      "loss": 0.7845,
+      "step": 259
+    },
+    {
+      "epoch": 1.4454482279360668,
+      "grad_norm": 0.6749443606812077,
+      "learning_rate": 7.151509011411693e-05,
+      "loss": 0.7751,
+      "step": 260
+    },
+    {
+      "epoch": 1.4510076441973592,
+      "grad_norm": 0.54518347236071,
+      "learning_rate": 7.14187164901985e-05,
+      "loss": 0.7787,
+      "step": 261
+    },
+    {
+      "epoch": 1.4565670604586518,
+      "grad_norm": 0.5880443192755512,
+      "learning_rate": 7.132186435093115e-05,
+      "loss": 0.7901,
+      "step": 262
+    },
+    {
+      "epoch": 1.4621264767199444,
+      "grad_norm": 0.5541773489903166,
+      "learning_rate": 7.122453517139854e-05,
+      "loss": 0.7697,
+      "step": 263
+    },
+    {
+      "epoch": 1.4676858929812369,
+      "grad_norm": 0.6059693677431078,
+      "learning_rate": 7.112673043394976e-05,
+      "loss": 0.7935,
+      "step": 264
+    },
+    {
+      "epoch": 1.4732453092425295,
+      "grad_norm": 0.4757882619272586,
+      "learning_rate": 7.10284516281768e-05,
+      "loss": 0.7735,
+      "step": 265
+    },
+    {
+      "epoch": 1.4788047255038221,
+      "grad_norm": 0.611919607902747,
+      "learning_rate": 7.092970025089183e-05,
+      "loss": 0.7914,
+      "step": 266
+    },
+    {
+      "epoch": 1.4843641417651146,
+      "grad_norm": 0.542145528258939,
+      "learning_rate": 7.08304778061044e-05,
+      "loss": 0.7774,
+      "step": 267
+    },
+    {
+      "epoch": 1.4899235580264072,
+      "grad_norm": 0.5325510989772689,
+      "learning_rate": 7.073078580499854e-05,
+      "loss": 0.7761,
+      "step": 268
+    },
+    {
+      "epoch": 1.4954829742876998,
+      "grad_norm": 0.5986995332394465,
+      "learning_rate": 7.063062576590975e-05,
+      "loss": 0.7838,
+      "step": 269
+    },
+    {
+      "epoch": 1.5010423905489922,
+      "grad_norm": 0.4616995964663547,
+      "learning_rate": 7.052999921430189e-05,
+      "loss": 0.7755,
+      "step": 270
+    },
+    {
+      "epoch": 1.5066018068102849,
+      "grad_norm": 0.49058298770662495,
+      "learning_rate": 7.04289076827439e-05,
+      "loss": 0.7861,
+      "step": 271
+    },
+    {
+      "epoch": 1.5121612230715775,
+      "grad_norm": 0.44812644938940693,
+      "learning_rate": 7.032735271088652e-05,
+      "loss": 0.7564,
+      "step": 272
+    },
+    {
+      "epoch": 1.51772063933287,
+      "grad_norm": 0.475517524463409,
+      "learning_rate": 7.022533584543877e-05,
+      "loss": 0.7649,
+      "step": 273
+    },
+    {
+      "epoch": 1.5232800555941626,
+      "grad_norm": 0.3956398801829029,
+      "learning_rate": 7.012285864014445e-05,
+      "loss": 0.7748,
+      "step": 274
+    },
+    {
+      "epoch": 1.5288394718554552,
+      "grad_norm": 0.3897447425893131,
+      "learning_rate": 7.001992265575846e-05,
+      "loss": 0.7783,
+      "step": 275
+    },
+    {
+      "epoch": 1.5343988881167476,
+      "grad_norm": 0.4597977578717938,
+      "learning_rate": 6.991652946002302e-05,
+      "loss": 0.7635,
+      "step": 276
+    },
+    {
+      "epoch": 1.5399583043780403,
+      "grad_norm": 0.4666034369576422,
+      "learning_rate": 6.981268062764383e-05,
+      "loss": 0.7683,
+      "step": 277
+    },
+    {
+      "epoch": 1.545517720639333,
+      "grad_norm": 0.4753869261509179,
+      "learning_rate": 6.9708377740266e-05,
+      "loss": 0.7666,
+      "step": 278
+    },
+    {
+      "epoch": 1.5510771369006253,
+      "grad_norm": 0.43411039480677927,
+      "learning_rate": 6.960362238645008e-05,
+      "loss": 0.775,
+      "step": 279
+    },
+    {
+      "epoch": 1.556636553161918,
+      "grad_norm": 0.35532156974575624,
+      "learning_rate": 6.949841616164774e-05,
+      "loss": 0.7715,
+      "step": 280
+    },
+    {
+      "epoch": 1.5621959694232106,
+      "grad_norm": 0.36018397788729956,
+      "learning_rate": 6.939276066817759e-05,
+      "loss": 0.7695,
+      "step": 281
+    },
+    {
+      "epoch": 1.567755385684503,
+      "grad_norm": 0.4363262505492209,
+      "learning_rate": 6.928665751520067e-05,
+      "loss": 0.8013,
+      "step": 282
+    },
+    {
+      "epoch": 1.5733148019457956,
+      "grad_norm": 0.49701805746941385,
+      "learning_rate": 6.918010831869604e-05,
+      "loss": 0.7855,
+      "step": 283
+    },
+    {
+      "epoch": 1.5788742182070883,
+      "grad_norm": 0.5598941853744219,
+      "learning_rate": 6.907311470143609e-05,
+      "loss": 0.779,
+      "step": 284
+    },
+    {
+      "epoch": 1.5844336344683807,
+      "grad_norm": 0.6100802112728656,
+      "learning_rate": 6.896567829296185e-05,
+      "loss": 0.7794,
+      "step": 285
+    },
+    {
+      "epoch": 1.5899930507296733,
+      "grad_norm": 0.7281021082315408,
+      "learning_rate": 6.885780072955824e-05,
+      "loss": 0.7709,
+      "step": 286
+    },
+    {
+      "epoch": 1.595552466990966,
+      "grad_norm": 0.8642915362417339,
+      "learning_rate": 6.874948365422899e-05,
+      "loss": 0.7772,
+      "step": 287
+    },
+    {
+      "epoch": 1.6011118832522584,
+      "grad_norm": 1.0745648538812815,
+      "learning_rate": 6.864072871667177e-05,
+      "loss": 0.7766,
+      "step": 288
+    },
+    {
+      "epoch": 1.606671299513551,
+      "grad_norm": 1.0956998824768283,
+      "learning_rate": 6.853153757325303e-05,
+      "loss": 0.78,
+      "step": 289
+    },
+    {
+      "epoch": 1.6122307157748437,
+      "grad_norm": 1.2750864854683548,
+      "learning_rate": 6.842191188698267e-05,
+      "loss": 0.7857,
+      "step": 290
+    },
+    {
+      "epoch": 1.617790132036136,
+      "grad_norm": 0.8795380185955686,
+      "learning_rate": 6.831185332748889e-05,
+      "loss": 0.7788,
+      "step": 291
+    },
+    {
+      "epoch": 1.6233495482974287,
+      "grad_norm": 0.8009577214005955,
+      "learning_rate": 6.820136357099256e-05,
+      "loss": 0.7716,
+      "step": 292
+    },
+    {
+      "epoch": 1.6289089645587214,
+      "grad_norm": 0.7908950069367938,
+      "learning_rate": 6.80904443002819e-05,
+      "loss": 0.7834,
+      "step": 293
+    },
+    {
+      "epoch": 1.6344683808200138,
+      "grad_norm": 0.7087524595566469,
+      "learning_rate": 6.797909720468665e-05,
+      "loss": 0.7799,
+      "step": 294
+    },
+    {
+      "epoch": 1.6400277970813064,
+      "grad_norm": 0.8154001152670927,
+      "learning_rate": 6.786732398005254e-05,
+      "loss": 0.7729,
+      "step": 295
+    },
+    {
+      "epoch": 1.645587213342599,
+      "grad_norm": 1.0032770151573074,
+      "learning_rate": 6.775512632871522e-05,
+      "loss": 0.7743,
+      "step": 296
+    },
+    {
+      "epoch": 1.6511466296038915,
+      "grad_norm": 1.0526489795713596,
+      "learning_rate": 6.76425059594746e-05,
+      "loss": 0.7755,
+      "step": 297
+    },
+    {
+      "epoch": 1.656706045865184,
+      "grad_norm": 0.9590121679714712,
+      "learning_rate": 6.752946458756862e-05,
+      "loss": 0.7709,
+      "step": 298
+    },
+    {
+      "epoch": 1.6622654621264767,
+      "grad_norm": 0.8034166003807086,
+      "learning_rate": 6.741600393464725e-05,
+      "loss": 0.7584,
+      "step": 299
+    },
+    {
+      "epoch": 1.6678248783877692,
+      "grad_norm": 0.6265399191838994,
+      "learning_rate": 6.730212572874618e-05,
+      "loss": 0.7701,
+      "step": 300
+    },
+    {
+      "epoch": 1.6733842946490618,
+      "grad_norm": 0.7017430730297519,
+      "learning_rate": 6.718783170426055e-05,
+      "loss": 0.7709,
+      "step": 301
+    },
+    {
+      "epoch": 1.6789437109103544,
+      "grad_norm": 0.7032001111839687,
+      "learning_rate": 6.707312360191854e-05,
+      "loss": 0.7714,
+      "step": 302
+    },
+    {
+      "epoch": 1.6845031271716469,
+      "grad_norm": 0.3993478869941896,
+      "learning_rate": 6.695800316875484e-05,
+      "loss": 0.7599,
+      "step": 303
+    },
+    {
+      "epoch": 1.6900625434329395,
+      "grad_norm": 0.5249562467408584,
+      "learning_rate": 6.684247215808407e-05,
+      "loss": 0.7692,
+      "step": 304
+    },
+    {
+      "epoch": 1.6956219596942321,
+      "grad_norm": 0.5899591681953836,
+      "learning_rate": 6.6726532329474e-05,
+      "loss": 0.7621,
+      "step": 305
+    },
+    {
+      "epoch": 1.7011813759555245,
+      "grad_norm": 0.5069877083978761,
+      "learning_rate": 6.661018544871884e-05,
+      "loss": 0.7845,
+      "step": 306
+    },
+    {
+      "epoch": 1.7067407922168172,
+      "grad_norm": 0.488248778365674,
+      "learning_rate": 6.649343328781232e-05,
+      "loss": 0.7832,
+      "step": 307
+    },
+    {
+      "epoch": 1.7123002084781098,
+      "grad_norm": 0.4690318525581349,
+      "learning_rate": 6.637627762492067e-05,
+      "loss": 0.7804,
+      "step": 308
+    },
+    {
+      "epoch": 1.7178596247394022,
+      "grad_norm": 0.4199909982678252,
+      "learning_rate": 6.625872024435559e-05,
+      "loss": 0.7816,
+      "step": 309
+    },
+    {
+      "epoch": 1.7234190410006949,
+      "grad_norm": 0.38057764950277645,
+      "learning_rate": 6.6140762936547e-05,
+      "loss": 0.7753,
+      "step": 310
+    },
+    {
+      "epoch": 1.7289784572619875,
+      "grad_norm": 0.3918174751770195,
+      "learning_rate": 6.602240749801588e-05,
+      "loss": 0.7688,
+      "step": 311
+    },
+    {
+      "epoch": 1.73453787352328,
+      "grad_norm": 0.3670670542000639,
+      "learning_rate": 6.59036557313468e-05,
+      "loss": 0.7671,
+      "step": 312
+    },
+    {
+      "epoch": 1.7400972897845726,
+      "grad_norm": 0.3846828078401982,
+      "learning_rate": 6.578450944516051e-05,
+      "loss": 0.7923,
+      "step": 313
+    },
+    {
+      "epoch": 1.7456567060458652,
+      "grad_norm": 0.39298089631112915,
+      "learning_rate": 6.566497045408645e-05,
+      "loss": 0.771,
+      "step": 314
+    },
+    {
+      "epoch": 1.7512161223071576,
+      "grad_norm": 0.370122117562583,
+      "learning_rate": 6.554504057873498e-05,
+      "loss": 0.7544,
+      "step": 315
+    },
+    {
+      "epoch": 1.7567755385684505,
+      "grad_norm": 0.3303038453558157,
+      "learning_rate": 6.542472164566979e-05,
+      "loss": 0.7655,
+      "step": 316
+    },
+    {
+      "epoch": 1.762334954829743,
+      "grad_norm": 0.5075770625427899,
+      "learning_rate": 6.530401548738001e-05,
+      "loss": 0.7576,
+      "step": 317
+    },
+    {
+      "epoch": 1.7678943710910353,
+      "grad_norm": 0.500351509483819,
+      "learning_rate": 6.51829239422523e-05,
+      "loss": 0.7668,
+      "step": 318
+    },
+    {
+      "epoch": 1.7734537873523282,
+      "grad_norm": 0.4365766487802718,
+      "learning_rate": 6.506144885454285e-05,
+      "loss": 0.7539,
+      "step": 319
+    },
+    {
+      "epoch": 1.7790132036136206,
+      "grad_norm": 0.45527067954463785,
+      "learning_rate": 6.493959207434934e-05,
+      "loss": 0.7666,
+      "step": 320
+    },
+    {
+      "epoch": 1.784572619874913,
+      "grad_norm": 0.4757862949658658,
+      "learning_rate": 6.481735545758273e-05,
+      "loss": 0.7609,
+      "step": 321
+    },
+    {
+      "epoch": 1.7901320361362059,
+      "grad_norm": 0.6179343909917517,
+      "learning_rate": 6.469474086593894e-05,
+      "loss": 0.7594,
+      "step": 322
+    },
+    {
+      "epoch": 1.7956914523974983,
+      "grad_norm": 0.7997107662830784,
+      "learning_rate": 6.45717501668706e-05,
+      "loss": 0.7559,
+      "step": 323
+    },
+    {
+      "epoch": 1.8012508686587907,
+      "grad_norm": 1.0969367036016504,
+      "learning_rate": 6.444838523355852e-05,
+      "loss": 0.7796,
+      "step": 324
+    },
+    {
+      "epoch": 1.8068102849200836,
+      "grad_norm": 1.2089537958976195,
+      "learning_rate": 6.432464794488323e-05,
+      "loss": 0.7724,
+      "step": 325
+    },
+    {
+      "epoch": 1.812369701181376,
+      "grad_norm": 0.6596723262517451,
+      "learning_rate": 6.42005401853963e-05,
+      "loss": 0.7607,
+      "step": 326
+    },
+    {
+      "epoch": 1.8179291174426684,
+      "grad_norm": 0.4902729906062423,
+      "learning_rate": 6.407606384529168e-05,
+      "loss": 0.778,
+      "step": 327
+    },
+    {
+      "epoch": 1.8234885337039612,
+      "grad_norm": 0.46790228332539113,
+      "learning_rate": 6.39512208203769e-05,
+      "loss": 0.7701,
+      "step": 328
+    },
+    {
+      "epoch": 1.8290479499652537,
+      "grad_norm": 0.5469006439320506,
+      "learning_rate": 6.382601301204421e-05,
+      "loss": 0.7818,
+      "step": 329
+    },
+    {
+      "epoch": 1.834607366226546,
+      "grad_norm": 0.740833234029731,
+      "learning_rate": 6.370044232724158e-05,
+      "loss": 0.7603,
+      "step": 330
+    },
+    {
+      "epoch": 1.840166782487839,
+      "grad_norm": 0.9010614851907555,
+      "learning_rate": 6.35745106784437e-05,
+      "loss": 0.7731,
+      "step": 331
+    },
+    {
+      "epoch": 1.8457261987491314,
+      "grad_norm": 1.0238595680669471,
+      "learning_rate": 6.344821998362285e-05,
+      "loss": 0.7776,
+      "step": 332
+    },
+    {
+      "epoch": 1.8512856150104238,
+      "grad_norm": 1.0395463083151937,
+      "learning_rate": 6.332157216621964e-05,
+      "loss": 0.7729,
+      "step": 333
+    },
+    {
+      "epoch": 1.8568450312717166,
+      "grad_norm": 0.8370888480193454,
+      "learning_rate": 6.31945691551138e-05,
+      "loss": 0.7582,
+      "step": 334
+    },
+    {
+      "epoch": 1.862404447533009,
+      "grad_norm": 0.6973849424171054,
+      "learning_rate": 6.30672128845947e-05,
+      "loss": 0.7638,
+      "step": 335
+    },
+    {
+      "epoch": 1.8679638637943015,
+      "grad_norm": 0.5915083898425778,
+      "learning_rate": 6.293950529433199e-05,
+      "loss": 0.7754,
+      "step": 336
+    },
+    {
+      "epoch": 1.8735232800555943,
+      "grad_norm": 0.45025014596604773,
+      "learning_rate": 6.281144832934596e-05,
+      "loss": 0.7699,
+      "step": 337
+    },
+    {
+      "epoch": 1.8790826963168867,
+      "grad_norm": 0.38555129030834384,
+      "learning_rate": 6.268304393997806e-05,
+      "loss": 0.7759,
+      "step": 338
+    },
+    {
+      "epoch": 1.8846421125781792,
+      "grad_norm": 0.36084485280720024,
+      "learning_rate": 6.2554294081861e-05,
+      "loss": 0.784,
+      "step": 339
+    },
+    {
+      "epoch": 1.890201528839472,
+      "grad_norm": 0.405266578021245,
+      "learning_rate": 6.242520071588915e-05,
+      "loss": 0.7783,
+      "step": 340
+    },
+    {
+      "epoch": 1.8957609451007644,
+      "grad_norm": 0.4124119655208444,
+      "learning_rate": 6.229576580818852e-05,
+      "loss": 0.7672,
+      "step": 341
+    },
+    {
+      "epoch": 1.9013203613620568,
+      "grad_norm": 0.4488084951300913,
+      "learning_rate": 6.2165991330087e-05,
+      "loss": 0.7822,
+      "step": 342
+    },
+    {
+      "epoch": 1.9068797776233497,
+      "grad_norm": 0.4082635480033681,
+      "learning_rate": 6.20358792580841e-05,
+      "loss": 0.7693,
+      "step": 343
+    },
+    {
+      "epoch": 1.9124391938846421,
+      "grad_norm": 0.377339639242065,
+      "learning_rate": 6.190543157382107e-05,
+      "loss": 0.7654,
+      "step": 344
+    },
+    {
+      "epoch": 1.9179986101459345,
+      "grad_norm": 0.35237882278693,
+      "learning_rate": 6.177465026405058e-05,
+      "loss": 0.7607,
+      "step": 345
+    },
+    {
+      "epoch": 1.9235580264072274,
+      "grad_norm": 0.3671603093472658,
+      "learning_rate": 6.164353732060651e-05,
+      "loss": 0.7676,
+      "step": 346
+    },
+    {
+      "epoch": 1.9291174426685198,
+      "grad_norm": 0.38659296565653467,
+      "learning_rate": 6.151209474037363e-05,
+      "loss": 0.7558,
+      "step": 347
+    },
+    {
+      "epoch": 1.9346768589298122,
+      "grad_norm": 0.359729700225076,
+      "learning_rate": 6.13803245252571e-05,
+      "loss": 0.7706,
+      "step": 348
+    },
+    {
+      "epoch": 1.940236275191105,
+      "grad_norm": 0.3272217394439331,
+      "learning_rate": 6.124822868215213e-05,
+      "loss": 0.7674,
+      "step": 349
+    },
+    {
+      "epoch": 1.9457956914523975,
+      "grad_norm": 0.3587548812477975,
+      "learning_rate": 6.11158092229133e-05,
+      "loss": 0.7566,
+      "step": 350
+    },
+    {
+      "epoch": 1.95135510771369,
+      "grad_norm": 0.38927389559214304,
+      "learning_rate": 6.098306816432393e-05,
+      "loss": 0.7608,
+      "step": 351
+    },
+    {
+      "epoch": 1.9569145239749828,
+      "grad_norm": 0.345702831255413,
+      "learning_rate": 6.0850007528065385e-05,
+      "loss": 0.7642,
+      "step": 352
+    },
+    {
+      "epoch": 1.9624739402362752,
+      "grad_norm": 0.3699780588350165,
+      "learning_rate": 6.0716629340686314e-05,
+      "loss": 0.7585,
+      "step": 353
+    },
+    {
+      "epoch": 1.9680333564975676,
+      "grad_norm": 0.4077210246910142,
+      "learning_rate": 6.058293563357172e-05,
+      "loss": 0.7634,
+      "step": 354
+    },
+    {
+      "epoch": 1.9735927727588605,
+      "grad_norm": 0.42750949194399185,
+      "learning_rate": 6.0448928442912064e-05,
+      "loss": 0.7644,
+      "step": 355
+    },
+    {
+      "epoch": 1.9791521890201529,
+      "grad_norm": 0.4974639937813857,
+      "learning_rate": 6.031460980967225e-05,
+      "loss": 0.7736,
+      "step": 356
+    },
+    {
+      "epoch": 1.9847116052814453,
+      "grad_norm": 0.4864835186867503,
+      "learning_rate": 6.017998177956052e-05,
+      "loss": 0.7617,
+      "step": 357
+    },
+    {
+      "epoch": 1.9902710215427382,
+      "grad_norm": 0.35299484136118686,
+      "learning_rate": 6.004504640299736e-05,
+      "loss": 0.7675,
+      "step": 358
+    },
+    {
+      "epoch": 1.9958304378040306,
+      "grad_norm": 0.31827026637009653,
+      "learning_rate": 5.990980573508415e-05,
+      "loss": 0.7631,
+      "step": 359
+    },
+    {
+      "epoch": 2.001389854065323,
+      "grad_norm": 0.4024790298162928,
+      "learning_rate": 5.9774261835571996e-05,
+      "loss": 0.7498,
+      "step": 360
+    },
+    {
+      "epoch": 2.006949270326616,
+      "grad_norm": 0.47173300208049507,
+      "learning_rate": 5.9638416768830277e-05,
+      "loss": 0.7376,
+      "step": 361
+    },
+    {
+      "epoch": 2.0125086865879083,
+      "grad_norm": 0.6895792658770121,
+      "learning_rate": 5.950227260381522e-05,
+      "loss": 0.7353,
+      "step": 362
+    },
+    {
+      "epoch": 2.0180681028492007,
+      "grad_norm": 1.0070048914467056,
+      "learning_rate": 5.9365831414038424e-05,
+      "loss": 0.7444,
+      "step": 363
+    },
+    {
+      "epoch": 2.0236275191104935,
+      "grad_norm": 1.3775654286348442,
+      "learning_rate": 5.9229095277535236e-05,
+      "loss": 0.7357,
+      "step": 364
+    },
+    {
+      "epoch": 2.029186935371786,
+      "grad_norm": 0.5355386773201476,
+      "learning_rate": 5.909206627683313e-05,
+      "loss": 0.7403,
+      "step": 365
+    },
+    {
+      "epoch": 2.0347463516330784,
+      "grad_norm": 0.46568020048542935,
+      "learning_rate": 5.895474649891995e-05,
+      "loss": 0.7293,
+      "step": 366
+    },
+    {
+      "epoch": 2.0403057678943712,
+      "grad_norm": 0.8082533261595521,
+      "learning_rate": 5.88171380352122e-05,
+      "loss": 0.7468,
+      "step": 367
+    },
+    {
+      "epoch": 2.0458651841556637,
+      "grad_norm": 1.0373988341475624,
+      "learning_rate": 5.867924298152311e-05,
+      "loss": 0.732,
+      "step": 368
+    },
+    {
+      "epoch": 2.051424600416956,
+      "grad_norm": 0.7732534471549348,
+      "learning_rate": 5.854106343803075e-05,
+      "loss": 0.7358,
+      "step": 369
+    },
+    {
+      "epoch": 2.056984016678249,
+      "grad_norm": 0.6069046552269562,
+      "learning_rate": 5.840260150924609e-05,
+      "loss": 0.7096,
+      "step": 370
+    },
+    {
+      "epoch": 2.0625434329395413,
+      "grad_norm": 0.8205673012712087,
+      "learning_rate": 5.826385930398084e-05,
+      "loss": 0.7427,
+      "step": 371
+    },
+    {
+      "epoch": 2.0681028492008338,
+      "grad_norm": 0.8486167959591705,
+      "learning_rate": 5.812483893531543e-05,
+      "loss": 0.7307,
+      "step": 372
+    },
+    {
+      "epoch": 2.0736622654621266,
+      "grad_norm": 0.662144513978678,
+      "learning_rate": 5.7985542520566795e-05,
+      "loss": 0.734,
+      "step": 373
+    },
+    {
+      "epoch": 2.079221681723419,
+      "grad_norm": 0.457883479207113,
+      "learning_rate": 5.7845972181256104e-05,
+      "loss": 0.7421,
+      "step": 374
+    },
+    {
+      "epoch": 2.0847810979847115,
+      "grad_norm": 0.517800202826518,
+      "learning_rate": 5.770613004307648e-05,
+      "loss": 0.7278,
+      "step": 375
+    },
+    {
+      "epoch": 2.0903405142460043,
+      "grad_norm": 0.8598983388632617,
+      "learning_rate": 5.756601823586063e-05,
+      "loss": 0.7522,
+      "step": 376
+    },
+    {
+      "epoch": 2.0958999305072967,
+      "grad_norm": 0.4689738299674123,
+      "learning_rate": 5.742563889354837e-05,
+      "loss": 0.7358,
+      "step": 377
+    },
+    {
+      "epoch": 2.101459346768589,
+      "grad_norm": 0.6323865613873795,
+      "learning_rate": 5.728499415415415e-05,
+      "loss": 0.7223,
+      "step": 378
+    },
+    {
+      "epoch": 2.107018763029882,
+      "grad_norm": 3.6525531842750674,
+      "learning_rate": 5.71440861597345e-05,
+      "loss": 0.7474,
+      "step": 379
+    },
+    {
+      "epoch": 2.1125781792911744,
+      "grad_norm": 115.32668021205329,
+      "learning_rate": 5.700291705635538e-05,
+      "loss": 1.2108,
+      "step": 380
+    },
+    {
+      "epoch": 2.118137595552467,
+      "grad_norm": 3.1583730002039574,
+      "learning_rate": 5.68614889940595e-05,
+      "loss": 0.7749,
+      "step": 381
+    },
+    {
+      "epoch": 2.1236970118137597,
+      "grad_norm": 0.590660191944181,
+      "learning_rate": 5.671980412683363e-05,
+      "loss": 0.7389,
+      "step": 382
+    },
+    {
+      "epoch": 2.129256428075052,
+      "grad_norm": 1.8224073499970155,
+      "learning_rate": 5.657786461257567e-05,
+      "loss": 0.7478,
+      "step": 383
+    },
+    {
+      "epoch": 2.1348158443363445,
+      "grad_norm": 1.2951121366611997,
+      "learning_rate": 5.643567261306194e-05,
+      "loss": 0.7464,
+      "step": 384
+    },
+    {
+      "epoch": 2.1403752605976374,
+      "grad_norm": 1.5832979985529985,
+      "learning_rate": 5.629323029391411e-05,
+      "loss": 0.7421,
+      "step": 385
+    },
+    {
+      "epoch": 2.14593467685893,
+      "grad_norm": 1.2316696310155895,
+      "learning_rate": 5.6150539824566344e-05,
+      "loss": 0.7482,
+      "step": 386
+    },
+    {
+      "epoch": 2.151494093120222,
+      "grad_norm": 1.1353584045603453,
+      "learning_rate": 5.600760337823215e-05,
+      "loss": 0.7433,
+      "step": 387
+    },
+    {
+      "epoch": 2.157053509381515,
+      "grad_norm": 0.8270888985327509,
+      "learning_rate": 5.586442313187137e-05,
+      "loss": 0.7518,
+      "step": 388
+    },
+    {
+      "epoch": 2.1626129256428075,
+      "grad_norm": 0.9863689806177816,
+      "learning_rate": 5.572100126615695e-05,
+      "loss": 0.7464,
+      "step": 389
+    },
+    {
+      "epoch": 2.1681723419041,
+      "grad_norm": 0.4731001360415031,
+      "learning_rate": 5.557733996544179e-05,
+      "loss": 0.7346,
+      "step": 390
+    },
+    {
+      "epoch": 2.1737317581653928,
+      "grad_norm": 0.7801308944094809,
+      "learning_rate": 5.543344141772542e-05,
+      "loss": 0.7299,
+      "step": 391
+    },
+    {
+      "epoch": 2.179291174426685,
+      "grad_norm": 0.5816725512915829,
+      "learning_rate": 5.528930781462074e-05,
+      "loss": 0.7494,
+      "step": 392
+    },
+    {
+      "epoch": 2.1848505906879776,
+      "grad_norm": 1.0772848290573547,
+      "learning_rate": 5.514494135132058e-05,
+      "loss": 0.7408,
+      "step": 393
+    },
+    {
+      "epoch": 2.1904100069492705,
+      "grad_norm": 11.53337748265314,
+      "learning_rate": 5.500034422656429e-05,
+      "loss": 0.734,
+      "step": 394
+    },
+    {
+      "epoch": 2.195969423210563,
+      "grad_norm": 4.309994215264064,
+      "learning_rate": 5.4855518642604276e-05,
+      "loss": 0.7513,
+      "step": 395
+    },
+    {
+      "epoch": 2.2015288394718553,
+      "grad_norm": 1.3250161354260865,
+      "learning_rate": 5.4710466805172414e-05,
+      "loss": 0.7502,
+      "step": 396
+    },
+    {
+      "epoch": 2.207088255733148,
+      "grad_norm": 1.3777075745252818,
+      "learning_rate": 5.456519092344652e-05,
+      "loss": 0.7562,
+      "step": 397
+    },
+    {
+      "epoch": 2.2126476719944406,
+      "grad_norm": 1.0819205433328192,
+      "learning_rate": 5.4419693210016586e-05,
+      "loss": 0.7501,
+      "step": 398
+    },
+    {
+      "epoch": 2.218207088255733,
+      "grad_norm": 1.2833623880524272,
+      "learning_rate": 5.427397588085127e-05,
+      "loss": 0.7361,
+      "step": 399
+    },
+    {
+      "epoch": 2.223766504517026,
+      "grad_norm": 1.0451018448350549,
+      "learning_rate": 5.412804115526392e-05,
+      "loss": 0.7461,
+      "step": 400
+    },
+    {
+      "epoch": 2.2293259207783183,
+      "grad_norm": 0.8816531121474304,
+      "learning_rate": 5.3981891255878985e-05,
+      "loss": 0.7425,
+      "step": 401
+    },
+    {
+      "epoch": 2.2348853370396107,
+      "grad_norm": 0.7542953513475557,
+      "learning_rate": 5.383552840859799e-05,
+      "loss": 0.742,
+      "step": 402
+    },
+    {
+      "epoch": 2.2404447533009035,
+      "grad_norm": 0.6832304869722696,
+      "learning_rate": 5.368895484256578e-05,
+      "loss": 0.7399,
+      "step": 403
+    },
+    {
+      "epoch": 2.246004169562196,
+      "grad_norm": 0.7498644535493143,
+      "learning_rate": 5.3542172790136464e-05,
+      "loss": 0.7417,
+      "step": 404
+    },
+    {
+      "epoch": 2.2515635858234884,
+      "grad_norm": 0.5620794275795693,
+      "learning_rate": 5.339518448683945e-05,
+      "loss": 0.7437,
+      "step": 405
+    },
+    {
+      "epoch": 2.2571230020847812,
+      "grad_norm": 0.7280261530015376,
+      "learning_rate": 5.324799217134542e-05,
+      "loss": 0.7447,
+      "step": 406
+    },
+    {
+      "epoch": 2.2626824183460736,
+      "grad_norm": 0.5794346609888263,
+      "learning_rate": 5.310059808543221e-05,
+      "loss": 0.7428,
+      "step": 407
+    },
+    {
+      "epoch": 2.268241834607366,
+      "grad_norm": 0.594584254894041,
+      "learning_rate": 5.2953004473950676e-05,
+      "loss": 0.7456,
+      "step": 408
+    },
+    {
+      "epoch": 2.273801250868659,
+      "grad_norm": 0.4540497546609938,
+      "learning_rate": 5.2805213584790486e-05,
+      "loss": 0.7349,
+      "step": 409
+    },
+    {
+      "epoch": 2.2793606671299513,
+      "grad_norm": 0.48721903761938296,
+      "learning_rate": 5.26572276688459e-05,
+      "loss": 0.7473,
+      "step": 410
+    },
+    {
+      "epoch": 2.2849200833912438,
+      "grad_norm": 0.3735664300049461,
+      "learning_rate": 5.250904897998153e-05,
+      "loss": 0.7379,
+      "step": 411
+    },
+    {
+      "epoch": 2.2904794996525366,
+      "grad_norm": 0.42570356028831763,
+      "learning_rate": 5.23606797749979e-05,
+      "loss": 0.7423,
+      "step": 412
+    },
+    {
+      "epoch": 2.296038915913829,
+      "grad_norm": 0.3608144662151246,
+      "learning_rate": 5.2212122313597206e-05,
+      "loss": 0.725,
+      "step": 413
+    },
+    {
+      "epoch": 2.3015983321751214,
+      "grad_norm": 0.3972157863450482,
+      "learning_rate": 5.206337885834881e-05,
+      "loss": 0.7435,
+      "step": 414
+    },
+    {
+      "epoch": 2.3071577484364143,
+      "grad_norm": 0.37044100478568076,
+      "learning_rate": 5.191445167465481e-05,
+      "loss": 0.7305,
+      "step": 415
+    },
+    {
+      "epoch": 2.3127171646977067,
+      "grad_norm": 0.40372283544972726,
+      "learning_rate": 5.176534303071553e-05,
+      "loss": 0.7341,
+      "step": 416
+    },
+    {
+      "epoch": 2.318276580958999,
+      "grad_norm": 0.35291131427848343,
+      "learning_rate": 5.161605519749502e-05,
+      "loss": 0.7342,
+      "step": 417
+    },
+    {
+      "epoch": 2.323835997220292,
+      "grad_norm": 0.4106862294883081,
+      "learning_rate": 5.1466590448686375e-05,
+      "loss": 0.7441,
+      "step": 418
+    },
+    {
+      "epoch": 2.3293954134815844,
+      "grad_norm": 0.3315316376787157,
+      "learning_rate": 5.13169510606772e-05,
+      "loss": 0.7277,
+      "step": 419
+    },
+    {
+      "epoch": 2.334954829742877,
+      "grad_norm": 0.35212971202718274,
+      "learning_rate": 5.116713931251491e-05,
+      "loss": 0.7214,
+      "step": 420
+    },
+    {
+      "epoch": 2.3405142460041697,
+      "grad_norm": 0.3109496189357139,
+      "learning_rate": 5.101715748587195e-05,
+      "loss": 0.739,
+      "step": 421
+    },
+    {
+      "epoch": 2.346073662265462,
+      "grad_norm": 0.34317285629474853,
+      "learning_rate": 5.0867007865011186e-05,
+      "loss": 0.7362,
+      "step": 422
+    },
+    {
+      "epoch": 2.3516330785267545,
+      "grad_norm": 0.39953180128623716,
+      "learning_rate": 5.071669273675095e-05,
+      "loss": 0.7444,
+      "step": 423
+    },
+    {
+      "epoch": 2.3571924947880474,
+      "grad_norm": 0.29225775466241055,
+      "learning_rate": 5.0566214390430386e-05,
+      "loss": 0.7252,
+      "step": 424
+    },
+    {
+      "epoch": 2.36275191104934,
+      "grad_norm": 0.34147649076740483,
+      "learning_rate": 5.041557511787442e-05,
+      "loss": 0.7265,
+      "step": 425
+    },
+    {
+      "epoch": 2.368311327310632,
+      "grad_norm": 0.28209832198556345,
+      "learning_rate": 5.026477721335895e-05,
+      "loss": 0.7333,
+      "step": 426
+    },
+    {
+      "epoch": 2.373870743571925,
+      "grad_norm": 0.27112430335660087,
+      "learning_rate": 5.011382297357589e-05,
+      "loss": 0.7326,
+      "step": 427
+    },
+    {
+      "epoch": 2.3794301598332175,
+      "grad_norm": 0.2958995444714563,
+      "learning_rate": 4.9962714697598175e-05,
+      "loss": 0.7418,
+      "step": 428
+    },
+    {
+      "epoch": 2.38498957609451,
+      "grad_norm": 0.254705600992757,
+      "learning_rate": 4.9811454686844756e-05,
+      "loss": 0.7539,
+      "step": 429
+    },
+    {
+      "epoch": 2.3905489923558028,
+      "grad_norm": 0.2704842883747253,
+      "learning_rate": 4.966004524504552e-05,
+      "loss": 0.7384,
+      "step": 430
+    },
+    {
+      "epoch": 2.396108408617095,
+      "grad_norm": 0.27016709718038645,
+      "learning_rate": 4.950848867820628e-05,
+      "loss": 0.727,
+      "step": 431
+    },
+    {
+      "epoch": 2.4016678248783876,
+      "grad_norm": 0.3267030249113294,
+      "learning_rate": 4.935678729457355e-05,
+      "loss": 0.7288,
+      "step": 432
+    },
+    {
+      "epoch": 2.4072272411396805,
+      "grad_norm": 0.31940168140369773,
+      "learning_rate": 4.920494340459947e-05,
+      "loss": 0.7351,
+      "step": 433
+    },
+    {
+      "epoch": 2.412786657400973,
+      "grad_norm": 0.29347717180560046,
+      "learning_rate": 4.9052959320906564e-05,
+      "loss": 0.7269,
+      "step": 434
+    },
+    {
+      "epoch": 2.4183460736622653,
+      "grad_norm": 0.26336896811236926,
+      "learning_rate": 4.890083735825258e-05,
+      "loss": 0.7397,
+      "step": 435
+    },
+    {
+      "epoch": 2.423905489923558,
+      "grad_norm": 0.33942479552349747,
+      "learning_rate": 4.874857983349517e-05,
+      "loss": 0.7305,
+      "step": 436
+    },
+    {
+      "epoch": 2.4294649061848506,
+      "grad_norm": 0.3056551202672292,
+      "learning_rate": 4.859618906555667e-05,
+      "loss": 0.7445,
+      "step": 437
+    },
+    {
+      "epoch": 2.435024322446143,
+      "grad_norm": 0.255653122241217,
+      "learning_rate": 4.8443667375388686e-05,
+      "loss": 0.7435,
+      "step": 438
+    },
+    {
+      "epoch": 2.440583738707436,
+      "grad_norm": 0.27740072317554226,
+      "learning_rate": 4.8291017085936834e-05,
+      "loss": 0.7435,
+      "step": 439
+    },
+    {
+      "epoch": 2.4461431549687283,
+      "grad_norm": 0.22741019497934506,
+      "learning_rate": 4.8138240522105365e-05,
+      "loss": 0.7375,
+      "step": 440
+    },
+    {
+      "epoch": 2.4517025712300207,
+      "grad_norm": 0.26409421864038285,
+      "learning_rate": 4.7985340010721654e-05,
+      "loss": 0.737,
+      "step": 441
+    },
+    {
+      "epoch": 2.4572619874913135,
+      "grad_norm": 0.26315157997100586,
+      "learning_rate": 4.783231788050089e-05,
+      "loss": 0.7373,
+      "step": 442
+    },
+    {
+      "epoch": 2.462821403752606,
+      "grad_norm": 0.254673897089132,
+      "learning_rate": 4.767917646201051e-05,
+      "loss": 0.7241,
+      "step": 443
+    },
+    {
+      "epoch": 2.4683808200138984,
+      "grad_norm": 0.312897463352474,
+      "learning_rate": 4.7525918087634775e-05,
+      "loss": 0.7345,
+      "step": 444
+    },
+    {
+      "epoch": 2.473940236275191,
+      "grad_norm": 0.28397633530974886,
+      "learning_rate": 4.737254509153918e-05,
+      "loss": 0.7244,
+      "step": 445
+    },
+    {
+      "epoch": 2.4794996525364836,
+      "grad_norm": 0.413384182880905,
+      "learning_rate": 4.721905980963496e-05,
+      "loss": 0.738,
+      "step": 446
+    },
+    {
+      "epoch": 2.485059068797776,
+      "grad_norm": 0.5086958530399852,
+      "learning_rate": 4.706546457954351e-05,
+      "loss": 0.7421,
+      "step": 447
+    },
+    {
+      "epoch": 2.490618485059069,
+      "grad_norm": 0.7824855922432978,
+      "learning_rate": 4.691176174056071e-05,
+      "loss": 0.737,
+      "step": 448
+    },
+    {
+      "epoch": 2.4961779013203613,
+      "grad_norm": 0.4381170078896152,
+      "learning_rate": 4.675795363362142e-05,
+      "loss": 0.7312,
+      "step": 449
+    },
+    {
+      "epoch": 2.5017373175816537,
+      "grad_norm": 0.3248189318651948,
+      "learning_rate": 4.660404260126369e-05,
+      "loss": 0.7362,
+      "step": 450
+    },
+    {
+      "epoch": 2.5072967338429466,
+      "grad_norm": 0.22687788903538336,
+      "learning_rate": 4.6450030987593226e-05,
+      "loss": 0.7466,
+      "step": 451
+    },
+    {
+      "epoch": 2.512856150104239,
+      "grad_norm": 0.2484699828082822,
+      "learning_rate": 4.629592113824754e-05,
+      "loss": 0.7261,
+      "step": 452
+    },
+    {
+      "epoch": 2.5184155663655314,
+      "grad_norm": 0.26873441485423244,
+      "learning_rate": 4.614171540036037e-05,
+      "loss": 0.7274,
+      "step": 453
+    },
+    {
+      "epoch": 2.5239749826268243,
+      "grad_norm": 0.25584966455636315,
+      "learning_rate": 4.598741612252577e-05,
+      "loss": 0.7482,
+      "step": 454
+    },
+    {
+      "epoch": 2.5295343988881167,
+      "grad_norm": 0.25300161712467034,
+      "learning_rate": 4.5833025654762535e-05,
+      "loss": 0.7307,
+      "step": 455
+    },
+    {
+      "epoch": 2.535093815149409,
+      "grad_norm": 0.2556990789787588,
+      "learning_rate": 4.567854634847825e-05,
+      "loss": 0.7397,
+      "step": 456
+    },
+    {
+      "epoch": 2.540653231410702,
+      "grad_norm": 0.25012778975423555,
+      "learning_rate": 4.552398055643353e-05,
+      "loss": 0.746,
+      "step": 457
+    },
+    {
+      "epoch": 2.5462126476719944,
+      "grad_norm": 0.27118048168405157,
+      "learning_rate": 4.5369330632706223e-05,
+      "loss": 0.7419,
+      "step": 458
+    },
+    {
+      "epoch": 2.551772063933287,
+      "grad_norm": 0.28964960768196785,
+      "learning_rate": 4.521459893265548e-05,
+      "loss": 0.7346,
+      "step": 459
+    },
+    {
+      "epoch": 2.5573314801945797,
+      "grad_norm": 0.27364215502822814,
+      "learning_rate": 4.505978781288598e-05,
+      "loss": 0.7312,
+      "step": 460
+    },
+    {
+      "epoch": 2.562890896455872,
+      "grad_norm": 0.21770274914348633,
+      "learning_rate": 4.490489963121194e-05,
+      "loss": 0.7257,
+      "step": 461
+    },
+    {
+      "epoch": 2.5684503127171645,
+      "grad_norm": 0.25218759647538125,
+      "learning_rate": 4.474993674662128e-05,
+      "loss": 0.741,
+      "step": 462
+    },
+    {
+      "epoch": 2.5740097289784574,
+      "grad_norm": 0.2419318546982388,
+      "learning_rate": 4.4594901519239645e-05,
+      "loss": 0.7474,
+      "step": 463
+    },
+    {
+      "epoch": 2.57956914523975,
+      "grad_norm": 0.32215620862830957,
+      "learning_rate": 4.443979631029449e-05,
+      "loss": 0.723,
+      "step": 464
+    },
+    {
+      "epoch": 2.585128561501042,
+      "grad_norm": 0.29857838205094217,
+      "learning_rate": 4.428462348207911e-05,
+      "loss": 0.7309,
+      "step": 465
+    },
+    {
+      "epoch": 2.590687977762335,
+      "grad_norm": 0.2209060117495464,
+      "learning_rate": 4.412938539791665e-05,
+      "loss": 0.7359,
+      "step": 466
+    },
+    {
+      "epoch": 2.5962473940236275,
+      "grad_norm": 0.22472545767913193,
+      "learning_rate": 4.3974084422124134e-05,
+      "loss": 0.7324,
+      "step": 467
+    },
+    {
+      "epoch": 2.60180681028492,
+      "grad_norm": 0.1946564937314796,
+      "learning_rate": 4.381872291997641e-05,
+      "loss": 0.7362,
+      "step": 468
+    },
+    {
+      "epoch": 2.6073662265462128,
+      "grad_norm": 0.2359140723071903,
+      "learning_rate": 4.366330325767022e-05,
+      "loss": 0.7286,
+      "step": 469
+    },
+    {
+      "epoch": 2.612925642807505,
+      "grad_norm": 0.2129248467763739,
+      "learning_rate": 4.350782780228802e-05,
+      "loss": 0.7302,
+      "step": 470
+    },
+    {
+      "epoch": 2.6184850590687976,
+      "grad_norm": 0.24861193172162277,
+      "learning_rate": 4.33522989217621e-05,
+      "loss": 0.7203,
+      "step": 471
+    },
+    {
+      "epoch": 2.6240444753300904,
+      "grad_norm": 0.22260636691489233,
+      "learning_rate": 4.3196718984838345e-05,
+      "loss": 0.7328,
+      "step": 472
+    },
+    {
+      "epoch": 2.629603891591383,
+      "grad_norm": 0.21657344137925627,
+      "learning_rate": 4.304109036104029e-05,
+      "loss": 0.7369,
+      "step": 473
+    },
+    {
+      "epoch": 2.6351633078526753,
+      "grad_norm": 0.1829539329834595,
+      "learning_rate": 4.288541542063297e-05,
+      "loss": 0.7195,
+      "step": 474
+    },
+    {
+      "epoch": 2.640722724113968,
+      "grad_norm": 0.19991359500635122,
+      "learning_rate": 4.272969653458685e-05,
+      "loss": 0.7289,
+      "step": 475
+    },
+    {
+      "epoch": 2.6462821403752605,
+      "grad_norm": 0.19984516052864532,
+      "learning_rate": 4.257393607454167e-05,
+      "loss": 0.7266,
+      "step": 476
+    },
+    {
+      "epoch": 2.651841556636553,
+      "grad_norm": 0.18507080512216367,
+      "learning_rate": 4.241813641277036e-05,
+      "loss": 0.7402,
+      "step": 477
+    },
+    {
+      "epoch": 2.657400972897846,
+      "grad_norm": 0.22016264298352997,
+      "learning_rate": 4.226229992214293e-05,
+      "loss": 0.7302,
+      "step": 478
+    },
+    {
+      "epoch": 2.6629603891591382,
+      "grad_norm": 0.23800061174952958,
+      "learning_rate": 4.2106428976090286e-05,
+      "loss": 0.7336,
+      "step": 479
+    },
+    {
+      "epoch": 2.6685198054204307,
+      "grad_norm": 0.1989157186858635,
+      "learning_rate": 4.1950525948568073e-05,
+      "loss": 0.7338,
+      "step": 480
+    },
+    {
+      "epoch": 2.6740792216817235,
+      "grad_norm": 0.283763583201288,
+      "learning_rate": 4.17945932140206e-05,
+      "loss": 0.7266,
+      "step": 481
+    },
+    {
+      "epoch": 2.679638637943016,
+      "grad_norm": 0.321917277351061,
+      "learning_rate": 4.1638633147344575e-05,
+      "loss": 0.7288,
+      "step": 482
+    },
+    {
+      "epoch": 2.6851980542043083,
+      "grad_norm": 0.23459593411861587,
+      "learning_rate": 4.1482648123853e-05,
+      "loss": 0.727,
+      "step": 483
+    },
+    {
+      "epoch": 2.690757470465601,
+      "grad_norm": 0.24861513819835215,
+      "learning_rate": 4.132664051923897e-05,
+      "loss": 0.7392,
+      "step": 484
+    },
+    {
+      "epoch": 2.6963168867268936,
+      "grad_norm": 0.30446525462850177,
+      "learning_rate": 4.117061270953951e-05,
+      "loss": 0.7516,
+      "step": 485
+    },
+    {
+      "epoch": 2.701876302988186,
+      "grad_norm": 0.3155678535641023,
+      "learning_rate": 4.101456707109935e-05,
+      "loss": 0.7215,
+      "step": 486
+    },
+    {
+      "epoch": 2.707435719249479,
+      "grad_norm": 0.2323552605079562,
+      "learning_rate": 4.085850598053479e-05,
+      "loss": 0.7336,
+      "step": 487
+    },
+    {
+      "epoch": 2.7129951355107713,
+      "grad_norm": 0.2044343122866403,
+      "learning_rate": 4.0702431814697434e-05,
+      "loss": 0.7283,
+      "step": 488
+    },
+    {
+      "epoch": 2.7185545517720637,
+      "grad_norm": 0.28637190695954,
+      "learning_rate": 4.054634695063804e-05,
+      "loss": 0.7456,
+      "step": 489
+    },
+    {
+      "epoch": 2.7241139680333566,
+      "grad_norm": 0.3267253563070388,
+      "learning_rate": 4.039025376557033e-05,
+      "loss": 0.7441,
+      "step": 490
+    },
+    {
+      "epoch": 2.729673384294649,
+      "grad_norm": 0.3347609079486309,
+      "learning_rate": 4.023415463683472e-05,
+      "loss": 0.7504,
+      "step": 491
+    },
+    {
+      "epoch": 2.7352328005559414,
+      "grad_norm": 0.2919900086611737,
+      "learning_rate": 4.007805194186213e-05,
+      "loss": 0.7401,
+      "step": 492
+    },
+    {
+      "epoch": 2.7407922168172343,
+      "grad_norm": 0.20720695930174413,
+      "learning_rate": 3.9921948058137874e-05,
+      "loss": 0.7201,
+      "step": 493
+    },
+    {
+      "epoch": 2.7463516330785267,
+      "grad_norm": 0.3269135713536802,
+      "learning_rate": 3.97658453631653e-05,
+      "loss": 0.7424,
+      "step": 494
+    },
+    {
+      "epoch": 2.751911049339819,
+      "grad_norm": 0.4091768369958175,
+      "learning_rate": 3.960974623442968e-05,
+      "loss": 0.7323,
+      "step": 495
+    },
+    {
+      "epoch": 2.757470465601112,
+      "grad_norm": 0.36444186506436177,
+      "learning_rate": 3.9453653049361966e-05,
+      "loss": 0.7307,
+      "step": 496
+    },
+    {
+      "epoch": 2.7630298818624044,
+      "grad_norm": 0.2700660410690018,
+      "learning_rate": 3.929756818530258e-05,
+      "loss": 0.7459,
+      "step": 497
+    },
+    {
+      "epoch": 2.768589298123697,
+      "grad_norm": 0.1864682458158403,
+      "learning_rate": 3.914149401946522e-05,
+      "loss": 0.7331,
+      "step": 498
+    },
+    {
+      "epoch": 2.7741487143849897,
+      "grad_norm": 0.23385225361417114,
+      "learning_rate": 3.898543292890065e-05,
+      "loss": 0.7359,
+      "step": 499
+    },
+    {
+      "epoch": 2.779708130646282,
+      "grad_norm": 0.3193885099831001,
+      "learning_rate": 3.882938729046051e-05,
+      "loss": 0.7416,
+      "step": 500
+    },
+    {
+      "epoch": 2.7852675469075745,
+      "grad_norm": 0.3468433720321668,
+      "learning_rate": 3.867335948076104e-05,
+      "loss": 0.7203,
+      "step": 501
+    },
+    {
+      "epoch": 2.7908269631688674,
+      "grad_norm": 0.2336768102315487,
+      "learning_rate": 3.851735187614701e-05,
+      "loss": 0.7383,
+      "step": 502
+    },
+    {
+      "epoch": 2.7963863794301598,
+      "grad_norm": 0.20477576787186483,
+      "learning_rate": 3.836136685265543e-05,
+      "loss": 0.7375,
+      "step": 503
+    },
+    {
+      "epoch": 2.801945795691452,
+      "grad_norm": 0.3370531882508558,
+      "learning_rate": 3.820540678597942e-05,
+      "loss": 0.7329,
+      "step": 504
+    },
+    {
+      "epoch": 2.807505211952745,
+      "grad_norm": 0.3168959312457467,
+      "learning_rate": 3.804947405143193e-05,
+      "loss": 0.7278,
+      "step": 505
+    },
+    {
+      "epoch": 2.8130646282140375,
+      "grad_norm": 0.213213613251235,
+      "learning_rate": 3.789357102390973e-05,
+      "loss": 0.7294,
+      "step": 506
+    },
+    {
+      "epoch": 2.81862404447533,
+      "grad_norm": 0.19536547709601546,
+      "learning_rate": 3.7737700077857076e-05,
+      "loss": 0.7314,
+      "step": 507
+    },
+    {
+      "epoch": 2.8241834607366227,
+      "grad_norm": 0.22171103501129447,
+      "learning_rate": 3.7581863587229645e-05,
+      "loss": 0.7357,
+      "step": 508
+    },
+    {
+      "epoch": 2.829742876997915,
+      "grad_norm": 0.27575352788878604,
+      "learning_rate": 3.7426063925458355e-05,
+      "loss": 0.727,
+      "step": 509
+    },
+    {
+      "epoch": 2.8353022932592076,
+      "grad_norm": 0.3155479135697007,
+      "learning_rate": 3.727030346541317e-05,
+      "loss": 0.7349,
+      "step": 510
+    },
+    {
+      "epoch": 2.8408617095205004,
+      "grad_norm": 0.18619005189562574,
+      "learning_rate": 3.7114584579367034e-05,
+      "loss": 0.7285,
+      "step": 511
+    },
+    {
+      "epoch": 2.846421125781793,
+      "grad_norm": 0.296120411593163,
+      "learning_rate": 3.695890963895972e-05,
+      "loss": 0.7233,
+      "step": 512
+    },
+    {
+      "epoch": 2.8519805420430853,
+      "grad_norm": 0.26348492927406175,
+      "learning_rate": 3.6803281015161675e-05,
+      "loss": 0.7303,
+      "step": 513
+    },
+    {
+      "epoch": 2.857539958304378,
+      "grad_norm": 0.20058141840327803,
+      "learning_rate": 3.664770107823792e-05,
+      "loss": 0.7233,
+      "step": 514
+    },
+    {
+      "epoch": 2.8630993745656705,
+      "grad_norm": 0.254471235916167,
+      "learning_rate": 3.6492172197711984e-05,
+      "loss": 0.7265,
+      "step": 515
+    },
+    {
+      "epoch": 2.868658790826963,
+      "grad_norm": 0.18654197868120972,
+      "learning_rate": 3.633669674232979e-05,
+      "loss": 0.7341,
+      "step": 516
+    },
+    {
+      "epoch": 2.874218207088256,
+      "grad_norm": 0.20182500678634277,
+      "learning_rate": 3.61812770800236e-05,
+      "loss": 0.7339,
+      "step": 517
+    },
+    {
+      "epoch": 2.8797776233495482,
+      "grad_norm": 0.2344052637904768,
+      "learning_rate": 3.602591557787589e-05,
+      "loss": 0.7362,
+      "step": 518
+    },
+    {
+      "epoch": 2.8853370396108406,
+      "grad_norm": 0.2082291409375869,
+      "learning_rate": 3.5870614602083365e-05,
+      "loss": 0.729,
+      "step": 519
+    },
+    {
+      "epoch": 2.8908964558721335,
+      "grad_norm": 0.22360753610299702,
+      "learning_rate": 3.571537651792091e-05,
+      "loss": 0.7461,
+      "step": 520
+    },
+    {
+      "epoch": 2.896455872133426,
+      "grad_norm": 0.2603672706455522,
+      "learning_rate": 3.556020368970552e-05,
+      "loss": 0.7394,
+      "step": 521
+    },
+    {
+      "epoch": 2.9020152883947183,
+      "grad_norm": 0.22388119959520128,
+      "learning_rate": 3.540509848076037e-05,
+      "loss": 0.7246,
+      "step": 522
+    },
+    {
+      "epoch": 2.907574704656011,
+      "grad_norm": 0.27500339761238496,
+      "learning_rate": 3.525006325337873e-05,
+      "loss": 0.7371,
+      "step": 523
+    },
+    {
+      "epoch": 2.9131341209173036,
+      "grad_norm": 0.23292609157697053,
+      "learning_rate": 3.509510036878807e-05,
+      "loss": 0.7362,
+      "step": 524
+    },
+    {
+      "epoch": 2.918693537178596,
+      "grad_norm": 0.19290073858029927,
+      "learning_rate": 3.4940212187114024e-05,
+      "loss": 0.7229,
+      "step": 525
+    },
+    {
+      "epoch": 2.924252953439889,
+      "grad_norm": 0.24314351107870302,
+      "learning_rate": 3.478540106734452e-05,
+      "loss": 0.7367,
+      "step": 526
+    },
+    {
+      "epoch": 2.9298123697011813,
+      "grad_norm": 0.18761409664747392,
+      "learning_rate": 3.4630669367293797e-05,
+      "loss": 0.7331,
+      "step": 527
+    },
+    {
+      "epoch": 2.9353717859624737,
+      "grad_norm": 0.23665013272710855,
+      "learning_rate": 3.4476019443566474e-05,
+      "loss": 0.7478,
+      "step": 528
+    },
+    {
+      "epoch": 2.9409312022237666,
+      "grad_norm": 0.222031580657751,
+      "learning_rate": 3.4321453651521756e-05,
+      "loss": 0.7264,
+      "step": 529
+    },
+    {
+      "epoch": 2.946490618485059,
+      "grad_norm": 0.20604984922930578,
+      "learning_rate": 3.4166974345237465e-05,
+      "loss": 0.7203,
+      "step": 530
+    },
+    {
+      "epoch": 2.9520500347463514,
+      "grad_norm": 0.20100587036500991,
+      "learning_rate": 3.401258387747425e-05,
+      "loss": 0.7337,
+      "step": 531
+    },
+    {
+      "epoch": 2.9576094510076443,
+      "grad_norm": 0.1899794463734188,
+      "learning_rate": 3.385828459963966e-05,
+      "loss": 0.7441,
+      "step": 532
+    },
+    {
+      "epoch": 2.9631688672689367,
+      "grad_norm": 0.2206507855989311,
+      "learning_rate": 3.3704078861752463e-05,
+      "loss": 0.7268,
+      "step": 533
+    },
+    {
+      "epoch": 2.968728283530229,
+      "grad_norm": 0.1884387695061294,
+      "learning_rate": 3.354996901240678e-05,
+      "loss": 0.7289,
+      "step": 534
+    },
+    {
+      "epoch": 2.974287699791522,
+      "grad_norm": 0.20088164351300247,
+      "learning_rate": 3.339595739873631e-05,
+      "loss": 0.7301,
+      "step": 535
+    },
+    {
+      "epoch": 2.9798471160528144,
+      "grad_norm": 0.2190606208850353,
+      "learning_rate": 3.32420463663786e-05,
+      "loss": 0.7359,
+      "step": 536
+    },
+    {
+      "epoch": 2.985406532314107,
+      "grad_norm": 0.21071320916886715,
+      "learning_rate": 3.3088238259439304e-05,
+      "loss": 0.7287,
+      "step": 537
+    },
+    {
+      "epoch": 2.9909659485753997,
+      "grad_norm": 0.2503997265876079,
+      "learning_rate": 3.2934535420456506e-05,
+      "loss": 0.7202,
+      "step": 538
+    },
+    {
+      "epoch": 2.996525364836692,
+      "grad_norm": 0.22396626980732445,
+      "learning_rate": 3.2780940190365043e-05,
+      "loss": 0.7312,
+      "step": 539
+    },
+    {
+      "epoch": 3.0020847810979845,
+      "grad_norm": 0.2498542781596777,
+      "learning_rate": 3.262745490846084e-05,
+      "loss": 0.725,
+      "step": 540
+    },
+    {
+      "epoch": 3.0076441973592773,
+      "grad_norm": 0.2037678144034356,
+      "learning_rate": 3.2474081912365245e-05,
+      "loss": 0.6976,
+      "step": 541
+    },
+    {
+      "epoch": 3.0132036136205698,
+      "grad_norm": 0.27581558969736686,
+      "learning_rate": 3.23208235379895e-05,
+      "loss": 0.7119,
+      "step": 542
+    },
+    {
+      "epoch": 3.018763029881862,
+      "grad_norm": 0.30757106176107923,
+      "learning_rate": 3.216768211949912e-05,
+      "loss": 0.7182,
+      "step": 543
+    },
+    {
+      "epoch": 3.024322446143155,
+      "grad_norm": 0.3465027003205821,
+      "learning_rate": 3.201465998927835e-05,
+      "loss": 0.7012,
+      "step": 544
+    },
+    {
+      "epoch": 3.0298818624044475,
+      "grad_norm": 0.3205211872891199,
+      "learning_rate": 3.1861759477894656e-05,
+      "loss": 0.6971,
+      "step": 545
+    },
+    {
+      "epoch": 3.03544127866574,
+      "grad_norm": 0.26675757876064743,
+      "learning_rate": 3.170898291406317e-05,
+      "loss": 0.7005,
+      "step": 546
+    },
+    {
+      "epoch": 3.0410006949270327,
+      "grad_norm": 0.33157082769310375,
+      "learning_rate": 3.155633262461133e-05,
+      "loss": 0.692,
+      "step": 547
+    },
+    {
+      "epoch": 3.046560111188325,
+      "grad_norm": 0.3520237527349914,
+      "learning_rate": 3.1403810934443346e-05,
+      "loss": 0.6979,
+      "step": 548
+    },
+    {
+      "epoch": 3.052119527449618,
+      "grad_norm": 0.2806618105175904,
+      "learning_rate": 3.125142016650482e-05,
+      "loss": 0.6943,
+      "step": 549
+    },
+    {
+      "epoch": 3.0576789437109104,
+      "grad_norm": 0.26339431867257346,
+      "learning_rate": 3.109916264174743e-05,
+      "loss": 0.6992,
+      "step": 550
+    },
+    {
+      "epoch": 3.063238359972203,
+      "grad_norm": 0.29331652252245627,
+      "learning_rate": 3.094704067909345e-05,
+      "loss": 0.7074,
+      "step": 551
+    },
+    {
+      "epoch": 3.0687977762334953,
+      "grad_norm": 0.22842865907148296,
+      "learning_rate": 3.079505659540055e-05,
+      "loss": 0.695,
+      "step": 552
+    },
+    {
+      "epoch": 3.074357192494788,
+      "grad_norm": 0.22450151562077095,
+      "learning_rate": 3.064321270542646e-05,
+      "loss": 0.7091,
+      "step": 553
+    },
+    {
+      "epoch": 3.0799166087560805,
+      "grad_norm": 0.2901780200486473,
+      "learning_rate": 3.0491511321793737e-05,
+      "loss": 0.7176,
+      "step": 554
+    },
+    {
+      "epoch": 3.085476025017373,
+      "grad_norm": 0.2667950414899655,
+      "learning_rate": 3.0339954754954488e-05,
+      "loss": 0.7092,
+      "step": 555
+    },
+    {
+      "epoch": 3.091035441278666,
+      "grad_norm": 0.20249142476959187,
+      "learning_rate": 3.0188545313155257e-05,
+      "loss": 0.7044,
+      "step": 556
+    },
+    {
+      "epoch": 3.0965948575399582,
+      "grad_norm": 0.2417989895903253,
+      "learning_rate": 3.0037285302401828e-05,
+      "loss": 0.7107,
+      "step": 557
+    },
+    {
+      "epoch": 3.102154273801251,
+      "grad_norm": 0.3090304508627475,
+      "learning_rate": 2.9886177026424107e-05,
+      "loss": 0.7063,
+      "step": 558
+    },
+    {
+      "epoch": 3.1077136900625435,
+      "grad_norm": 0.23078910950196774,
+      "learning_rate": 2.9735222786641067e-05,
+      "loss": 0.6995,
+      "step": 559
+    },
+    {
+      "epoch": 3.113273106323836,
+      "grad_norm": 0.2160202375232252,
+      "learning_rate": 2.9584424882125593e-05,
+      "loss": 0.7149,
+      "step": 560
+    },
+    {
+      "epoch": 3.1188325225851288,
+      "grad_norm": 0.21966525904866915,
+      "learning_rate": 2.943378560956962e-05,
+      "loss": 0.7155,
+      "step": 561
+    },
+    {
+      "epoch": 3.124391938846421,
+      "grad_norm": 0.21942461047996667,
+      "learning_rate": 2.9283307263249048e-05,
+      "loss": 0.7024,
+      "step": 562
+    },
+    {
+      "epoch": 3.1299513551077136,
+      "grad_norm": 0.20165607592754078,
+      "learning_rate": 2.913299213498884e-05,
+      "loss": 0.6945,
+      "step": 563
+    },
+    {
+      "epoch": 3.135510771369006,
+      "grad_norm": 0.207087253518945,
+      "learning_rate": 2.8982842514128058e-05,
+      "loss": 0.7007,
+      "step": 564
+    },
+    {
+      "epoch": 3.141070187630299,
+      "grad_norm": 0.20062728334317925,
+      "learning_rate": 2.8832860687485107e-05,
+      "loss": 0.7113,
+      "step": 565
+    },
+    {
+      "epoch": 3.1466296038915913,
+      "grad_norm": 0.19818489838165365,
+      "learning_rate": 2.86830489393228e-05,
+      "loss": 0.679,
+      "step": 566
+    },
+    {
+      "epoch": 3.1521890201528837,
+      "grad_norm": 0.18399616142175082,
+      "learning_rate": 2.8533409551313628e-05,
+      "loss": 0.7127,
+      "step": 567
+    },
+    {
+      "epoch": 3.1577484364141766,
+      "grad_norm": 0.19589928841704413,
+      "learning_rate": 2.8383944802505003e-05,
+      "loss": 0.6942,
+      "step": 568
+    },
+    {
+      "epoch": 3.163307852675469,
+      "grad_norm": 0.18770585615805666,
+      "learning_rate": 2.8234656969284483e-05,
+      "loss": 0.6891,
+      "step": 569
+    },
+    {
+      "epoch": 3.1688672689367614,
+      "grad_norm": 0.23570832512589554,
+      "learning_rate": 2.8085548325345208e-05,
+      "loss": 0.6943,
+      "step": 570
+    },
+    {
+      "epoch": 3.1744266851980543,
+      "grad_norm": 0.1901096056694783,
+      "learning_rate": 2.7936621141651197e-05,
+      "loss": 0.7037,
+      "step": 571
+    },
+    {
+      "epoch": 3.1799861014593467,
+      "grad_norm": 0.21456549437287814,
+      "learning_rate": 2.7787877686402807e-05,
+      "loss": 0.6951,
+      "step": 572
+    },
+    {
+      "epoch": 3.185545517720639,
+      "grad_norm": 0.23553224751824287,
+      "learning_rate": 2.7639320225002108e-05,
+      "loss": 0.6918,
+      "step": 573
+    },
+    {
+      "epoch": 3.191104933981932,
+      "grad_norm": 0.18659378336889193,
+      "learning_rate": 2.7490951020018476e-05,
+      "loss": 0.6989,
+      "step": 574
+    },
+    {
+      "epoch": 3.1966643502432244,
+      "grad_norm": 0.21903803816553757,
+      "learning_rate": 2.7342772331154102e-05,
+      "loss": 0.6982,
+      "step": 575
+    },
+    {
+      "epoch": 3.2022237665045172,
+      "grad_norm": 0.2476302597618211,
+      "learning_rate": 2.7194786415209528e-05,
+      "loss": 0.6984,
+      "step": 576
+    },
+    {
+      "epoch": 3.2077831827658096,
+      "grad_norm": 0.17282018437992405,
+      "learning_rate": 2.7046995526049334e-05,
+      "loss": 0.7081,
+      "step": 577
+    },
+    {
+      "epoch": 3.213342599027102,
+      "grad_norm": 0.2142426744032126,
+      "learning_rate": 2.68994019145678e-05,
+      "loss": 0.7005,
+      "step": 578
+    },
+    {
+      "epoch": 3.218902015288395,
+      "grad_norm": 0.18982645909872903,
+      "learning_rate": 2.675200782865459e-05,
+      "loss": 0.7108,
+      "step": 579
+    },
+    {
+      "epoch": 3.2244614315496873,
+      "grad_norm": 0.1695279613152048,
+      "learning_rate": 2.6604815513160556e-05,
+      "loss": 0.704,
+      "step": 580
+    },
+    {
+      "epoch": 3.2300208478109798,
+      "grad_norm": 0.1868390211196699,
+      "learning_rate": 2.645782720986356e-05,
+      "loss": 0.712,
+      "step": 581
+    },
+    {
+      "epoch": 3.2355802640722726,
+      "grad_norm": 0.1800257784614353,
+      "learning_rate": 2.6311045157434235e-05,
+      "loss": 0.689,
+      "step": 582
+    },
+    {
+      "epoch": 3.241139680333565,
+      "grad_norm": 0.18122410935561592,
+      "learning_rate": 2.6164471591402016e-05,
+      "loss": 0.716,
+      "step": 583
+    },
+    {
+      "epoch": 3.2466990965948574,
+      "grad_norm": 0.17842363937321248,
+      "learning_rate": 2.6018108744121032e-05,
+      "loss": 0.7085,
+      "step": 584
+    },
+    {
+      "epoch": 3.25225851285615,
+      "grad_norm": 0.21715670011851243,
+      "learning_rate": 2.5871958844736083e-05,
+      "loss": 0.7079,
+      "step": 585
+    },
+    {
+      "epoch": 3.2578179291174427,
+      "grad_norm": 0.16191852665066875,
+      "learning_rate": 2.5726024119148757e-05,
+      "loss": 0.7156,
+      "step": 586
+    },
+    {
+      "epoch": 3.263377345378735,
+      "grad_norm": 0.18732524749733667,
+      "learning_rate": 2.5580306789983418e-05,
+      "loss": 0.7082,
+      "step": 587
+    },
+    {
+      "epoch": 3.2689367616400276,
+      "grad_norm": 0.19189813282737295,
+      "learning_rate": 2.54348090765535e-05,
+      "loss": 0.6969,
+      "step": 588
+    },
+    {
+      "epoch": 3.2744961779013204,
+      "grad_norm": 0.2040135809617877,
+      "learning_rate": 2.5289533194827586e-05,
+      "loss": 0.6864,
+      "step": 589
+    },
+    {
+      "epoch": 3.280055594162613,
+      "grad_norm": 0.20698619029904855,
+      "learning_rate": 2.514448135739574e-05,
+      "loss": 0.7149,
+      "step": 590
+    },
+    {
+      "epoch": 3.2856150104239052,
+      "grad_norm": 0.21355927166605512,
+      "learning_rate": 2.4999655773435726e-05,
+      "loss": 0.6965,
+      "step": 591
+    },
+    {
+      "epoch": 3.291174426685198,
+      "grad_norm": 0.19655275079030954,
+      "learning_rate": 2.4855058648679434e-05,
+      "loss": 0.709,
+      "step": 592
+    },
+    {
+      "epoch": 3.2967338429464905,
+      "grad_norm": 0.24361735832750547,
+      "learning_rate": 2.4710692185379264e-05,
+      "loss": 0.6981,
+      "step": 593
+    },
+    {
+      "epoch": 3.302293259207783,
+      "grad_norm": 0.2009267455727919,
+      "learning_rate": 2.4566558582274583e-05,
+      "loss": 0.7,
+      "step": 594
+    },
+    {
+      "epoch": 3.307852675469076,
+      "grad_norm": 0.25746317333166957,
+      "learning_rate": 2.442266003455823e-05,
+      "loss": 0.6986,
+      "step": 595
+    },
+    {
+      "epoch": 3.313412091730368,
+      "grad_norm": 0.26242513130569445,
+      "learning_rate": 2.427899873384306e-05,
+      "loss": 0.7138,
+      "step": 596
+    },
+    {
+      "epoch": 3.3189715079916606,
+      "grad_norm": 0.1882559013402892,
+      "learning_rate": 2.4135576868128636e-05,
+      "loss": 0.7013,
+      "step": 597
+    },
+    {
+      "epoch": 3.3245309242529535,
+      "grad_norm": 0.2889534969293074,
+      "learning_rate": 2.399239662176785e-05,
+      "loss": 0.7144,
+      "step": 598
+    },
+    {
+      "epoch": 3.330090340514246,
+      "grad_norm": 0.20123131689928303,
+      "learning_rate": 2.3849460175433673e-05,
+      "loss": 0.716,
+      "step": 599
+    },
+    {
+      "epoch": 3.3356497567755383,
+      "grad_norm": 0.23261788576448003,
+      "learning_rate": 2.3706769706085896e-05,
+      "loss": 0.6947,
+      "step": 600
+    },
+    {
+      "epoch": 3.341209173036831,
+      "grad_norm": 0.21185868547279016,
+      "learning_rate": 2.3564327386938075e-05,
+      "loss": 0.7044,
+      "step": 601
+    },
+    {
+      "epoch": 3.3467685892981236,
+      "grad_norm": 0.23101318712846342,
+      "learning_rate": 2.3422135387424342e-05,
+      "loss": 0.6948,
+      "step": 602
+    },
+    {
+      "epoch": 3.352328005559416,
+      "grad_norm": 0.17488832365954124,
+      "learning_rate": 2.3280195873166384e-05,
+      "loss": 0.7178,
+      "step": 603
+    },
+    {
+      "epoch": 3.357887421820709,
+      "grad_norm": 0.2277773250170869,
+      "learning_rate": 2.3138511005940502e-05,
+      "loss": 0.6903,
+      "step": 604
+    },
+    {
+      "epoch": 3.3634468380820013,
+      "grad_norm": 0.15592015395767056,
+      "learning_rate": 2.2997082943644635e-05,
+      "loss": 0.6961,
+      "step": 605
+    },
+    {
+      "epoch": 3.3690062543432937,
+      "grad_norm": 0.2772999957520557,
+      "learning_rate": 2.2855913840265504e-05,
+      "loss": 0.716,
+      "step": 606
+    },
+    {
+      "epoch": 3.3745656706045866,
+      "grad_norm": 0.1913976829716375,
+      "learning_rate": 2.271500584584585e-05,
+      "loss": 0.6921,
+      "step": 607
+    },
+    {
+      "epoch": 3.380125086865879,
+      "grad_norm": 0.17887054500690366,
+      "learning_rate": 2.2574361106451637e-05,
+      "loss": 0.6911,
+      "step": 608
+    },
+    {
+      "epoch": 3.385684503127172,
+      "grad_norm": 0.21750205862731467,
+      "learning_rate": 2.2433981764139383e-05,
+      "loss": 0.7096,
+      "step": 609
+    },
+    {
+      "epoch": 3.3912439193884643,
+      "grad_norm": 0.19593147325746155,
+      "learning_rate": 2.2293869956923517e-05,
+      "loss": 0.6982,
+      "step": 610
+    },
+    {
+      "epoch": 3.3968033356497567,
+      "grad_norm": 0.20784090087675822,
+      "learning_rate": 2.215402781874391e-05,
+      "loss": 0.701,
+      "step": 611
+    },
+    {
+      "epoch": 3.4023627519110495,
+      "grad_norm": 0.2094241373890822,
+      "learning_rate": 2.2014457479433204e-05,
+      "loss": 0.7006,
+      "step": 612
+    },
+    {
+      "epoch": 3.407922168172342,
+      "grad_norm": 0.19165733648750527,
+      "learning_rate": 2.1875161064684586e-05,
+      "loss": 0.7011,
+      "step": 613
+    },
+    {
+      "epoch": 3.4134815844336344,
+      "grad_norm": 0.28101320010779757,
+      "learning_rate": 2.1736140696019166e-05,
+      "loss": 0.6973,
+      "step": 614
+    },
+    {
+      "epoch": 3.4190410006949272,
+      "grad_norm": 0.18629632876510363,
+      "learning_rate": 2.1597398490753917e-05,
+      "loss": 0.7066,
+      "step": 615
+    },
+    {
+      "epoch": 3.4246004169562196,
+      "grad_norm": 0.22661440985457448,
+      "learning_rate": 2.1458936561969254e-05,
+      "loss": 0.7026,
+      "step": 616
+    },
+    {
+      "epoch": 3.430159833217512,
+      "grad_norm": 0.2865403117660965,
+      "learning_rate": 2.132075701847691e-05,
+      "loss": 0.708,
+      "step": 617
+    },
+    {
+      "epoch": 3.435719249478805,
+      "grad_norm": 0.15089884025551945,
+      "learning_rate": 2.1182861964787803e-05,
+      "loss": 0.7115,
+      "step": 618
+    },
+    {
+      "epoch": 3.4412786657400973,
+      "grad_norm": 0.2952596320891401,
+      "learning_rate": 2.1045253501080058e-05,
+      "loss": 0.6971,
+      "step": 619
+    },
+    {
+      "epoch": 3.4468380820013897,
+      "grad_norm": 0.2189167426361826,
+      "learning_rate": 2.0907933723166888e-05,
+      "loss": 0.7074,
+      "step": 620
+    },
+    {
+      "epoch": 3.4523974982626826,
+      "grad_norm": 0.21844536238960785,
+      "learning_rate": 2.0770904722464764e-05,
+      "loss": 0.6917,
+      "step": 621
+    },
+    {
+      "epoch": 3.457956914523975,
+      "grad_norm": 0.1843811278861263,
+      "learning_rate": 2.06341685859616e-05,
+      "loss": 0.7093,
+      "step": 622
+    },
+    {
+      "epoch": 3.4635163307852674,
+      "grad_norm": 0.2161993516067528,
+      "learning_rate": 2.049772739618479e-05,
+      "loss": 0.7032,
+      "step": 623
+    },
+    {
+      "epoch": 3.4690757470465603,
+      "grad_norm": 0.21374762977410636,
+      "learning_rate": 2.0361583231169747e-05,
+      "loss": 0.709,
+      "step": 624
+    },
+    {
+      "epoch": 3.4746351633078527,
+      "grad_norm": 0.2060717500194053,
+      "learning_rate": 2.022573816442801e-05,
+      "loss": 0.6936,
+      "step": 625
+    },
+    {
+      "epoch": 3.480194579569145,
+      "grad_norm": 0.19602587113180545,
+      "learning_rate": 2.0090194264915873e-05,
+      "loss": 0.7118,
+      "step": 626
+    },
+    {
+      "epoch": 3.485753995830438,
+      "grad_norm": 0.20086622178734398,
+      "learning_rate": 1.9954953597002654e-05,
+      "loss": 0.7105,
+      "step": 627
+    },
+    {
+      "epoch": 3.4913134120917304,
+      "grad_norm": 0.19511153222139876,
+      "learning_rate": 1.9820018220439487e-05,
+      "loss": 0.6957,
+      "step": 628
+    },
+    {
+      "epoch": 3.496872828353023,
+      "grad_norm": 0.19884256435263836,
+      "learning_rate": 1.9685390190327757e-05,
+      "loss": 0.7098,
+      "step": 629
+    },
+    {
+      "epoch": 3.5024322446143152,
+      "grad_norm": 0.15325133615321943,
+      "learning_rate": 1.955107155708795e-05,
+      "loss": 0.6971,
+      "step": 630
+    },
+    {
+      "epoch": 3.507991660875608,
+      "grad_norm": 0.16150214875067356,
+      "learning_rate": 1.9417064366428295e-05,
+      "loss": 0.7186,
+      "step": 631
+    },
+    {
+      "epoch": 3.5135510771369005,
+      "grad_norm": 0.1789338613581912,
+      "learning_rate": 1.92833706593137e-05,
+      "loss": 0.6964,
+      "step": 632
+    },
+    {
+      "epoch": 3.519110493398193,
+      "grad_norm": 0.20761650610037347,
+      "learning_rate": 1.9149992471934614e-05,
+      "loss": 0.6896,
+      "step": 633
+    },
+    {
+      "epoch": 3.524669909659486,
+      "grad_norm": 0.18407744122093334,
+      "learning_rate": 1.9016931835676078e-05,
+      "loss": 0.705,
+      "step": 634
+    },
+    {
+      "epoch": 3.530229325920778,
+      "grad_norm": 0.20855512031756548,
+      "learning_rate": 1.8884190777086705e-05,
+      "loss": 0.6974,
+      "step": 635
+    },
+    {
+      "epoch": 3.5357887421820706,
+      "grad_norm": 0.19336549621743102,
+      "learning_rate": 1.8751771317847876e-05,
+      "loss": 0.706,
+      "step": 636
+    },
+    {
+      "epoch": 3.5413481584433635,
+      "grad_norm": 0.21214610837166514,
+      "learning_rate": 1.8619675474742904e-05,
+      "loss": 0.7115,
+      "step": 637
+    },
+    {
+      "epoch": 3.546907574704656,
+      "grad_norm": 0.2022598865741006,
+      "learning_rate": 1.848790525962639e-05,
+      "loss": 0.7107,
+      "step": 638
+    },
+    {
+      "epoch": 3.5524669909659483,
+      "grad_norm": 0.21753789595081924,
+      "learning_rate": 1.8356462679393488e-05,
+      "loss": 0.7102,
+      "step": 639
+    },
+    {
+      "epoch": 3.558026407227241,
+      "grad_norm": 0.16764139285094648,
+      "learning_rate": 1.8225349735949435e-05,
+      "loss": 0.7051,
+      "step": 640
+    },
+    {
+      "epoch": 3.5635858234885336,
+      "grad_norm": 0.18750292991240983,
+      "learning_rate": 1.809456842617893e-05,
+      "loss": 0.7093,
+      "step": 641
+    },
+    {
+      "epoch": 3.569145239749826,
+      "grad_norm": 0.1684536912531375,
+      "learning_rate": 1.7964120741915905e-05,
+      "loss": 0.7072,
+      "step": 642
+    },
+    {
+      "epoch": 3.574704656011119,
+      "grad_norm": 0.15202938571970473,
+      "learning_rate": 1.7834008669913016e-05,
+      "loss": 0.7013,
+      "step": 643
+    },
+    {
+      "epoch": 3.5802640722724113,
+      "grad_norm": 0.15639709243101124,
+      "learning_rate": 1.770423419181147e-05,
+      "loss": 0.711,
+      "step": 644
+    },
+    {
+      "epoch": 3.5858234885337037,
+      "grad_norm": 0.16082010640489602,
+      "learning_rate": 1.7574799284110863e-05,
+      "loss": 0.6973,
+      "step": 645
+    },
+    {
+      "epoch": 3.5913829047949966,
+      "grad_norm": 0.1658067634382349,
+      "learning_rate": 1.744570591813901e-05,
+      "loss": 0.6946,
+      "step": 646
+    },
+    {
+      "epoch": 3.596942321056289,
+      "grad_norm": 0.16969803277351794,
+      "learning_rate": 1.7316956060021957e-05,
+      "loss": 0.6827,
+      "step": 647
+    },
+    {
+      "epoch": 3.6025017373175814,
+      "grad_norm": 0.16177207356044507,
+      "learning_rate": 1.718855167065404e-05,
+      "loss": 0.7036,
+      "step": 648
+    },
+    {
+      "epoch": 3.6080611535788742,
+      "grad_norm": 0.16664683404878003,
+      "learning_rate": 1.706049470566803e-05,
+      "loss": 0.6964,
+      "step": 649
+    },
+    {
+      "epoch": 3.6136205698401667,
+      "grad_norm": 0.19244596702153122,
+      "learning_rate": 1.6932787115405318e-05,
+      "loss": 0.703,
+      "step": 650
+    },
+    {
+      "epoch": 3.619179986101459,
+      "grad_norm": 0.14531076153595857,
+      "learning_rate": 1.6805430844886226e-05,
+      "loss": 0.7104,
+      "step": 651
+    },
+    {
+      "epoch": 3.624739402362752,
+      "grad_norm": 0.17188892378376985,
+      "learning_rate": 1.667842783378036e-05,
+      "loss": 0.7014,
+      "step": 652
+    },
+    {
+      "epoch": 3.6302988186240444,
+      "grad_norm": 0.14674805004070898,
+      "learning_rate": 1.655178001637716e-05,
+      "loss": 0.697,
+      "step": 653
+    },
+    {
+      "epoch": 3.6358582348853368,
+      "grad_norm": 0.18306188304250803,
+      "learning_rate": 1.6425489321556306e-05,
+      "loss": 0.7171,
+      "step": 654
+    },
+    {
+      "epoch": 3.6414176511466296,
+      "grad_norm": 0.1770643704313744,
+      "learning_rate": 1.6299557672758434e-05,
+      "loss": 0.7063,
+      "step": 655
+    },
+    {
+      "epoch": 3.646977067407922,
+      "grad_norm": 0.17018093462077719,
+      "learning_rate": 1.6173986987955794e-05,
+      "loss": 0.7014,
+      "step": 656
+    },
+    {
+      "epoch": 3.6525364836692145,
+      "grad_norm": 0.15144979941800404,
+      "learning_rate": 1.6048779179623098e-05,
+      "loss": 0.6987,
+      "step": 657
+    },
+    {
+      "epoch": 3.6580958999305073,
+      "grad_norm": 0.1306496439198933,
+      "learning_rate": 1.5923936154708332e-05,
+      "loss": 0.6992,
+      "step": 658
+    },
+    {
+      "epoch": 3.6636553161917997,
+      "grad_norm": 0.18827301402231672,
+      "learning_rate": 1.5799459814603716e-05,
+      "loss": 0.6939,
+      "step": 659
+    },
+    {
+      "epoch": 3.669214732453092,
+      "grad_norm": 0.1310253500615189,
+      "learning_rate": 1.5675352055116774e-05,
+      "loss": 0.6972,
+      "step": 660
+    },
+    {
+      "epoch": 3.674774148714385,
+      "grad_norm": 0.1583027749111766,
+      "learning_rate": 1.5551614766441483e-05,
+      "loss": 0.691,
+      "step": 661
+    },
+    {
+      "epoch": 3.6803335649756774,
+      "grad_norm": 0.1442216498175938,
+      "learning_rate": 1.54282498331294e-05,
+      "loss": 0.6931,
+      "step": 662
+    },
+    {
+      "epoch": 3.68589298123697,
+      "grad_norm": 0.16962082435760417,
+      "learning_rate": 1.5305259134061075e-05,
+      "loss": 0.6965,
+      "step": 663
+    },
+    {
+      "epoch": 3.6914523974982627,
+      "grad_norm": 0.15393789815894893,
+      "learning_rate": 1.5182644542417278e-05,
+      "loss": 0.7041,
+      "step": 664
+    },
+    {
+      "epoch": 3.697011813759555,
+      "grad_norm": 0.1407850186063632,
+      "learning_rate": 1.5060407925650662e-05,
+      "loss": 0.6907,
+      "step": 665
+    },
+    {
+      "epoch": 3.7025712300208475,
+      "grad_norm": 0.14524255117484877,
+      "learning_rate": 1.4938551145457152e-05,
+      "loss": 0.6991,
+      "step": 666
+    },
+    {
+      "epoch": 3.7081306462821404,
+      "grad_norm": 0.17278063379073116,
+      "learning_rate": 1.4817076057747728e-05,
+      "loss": 0.7089,
+      "step": 667
+    },
+    {
+      "epoch": 3.713690062543433,
+      "grad_norm": 0.14326689230859943,
+      "learning_rate": 1.4695984512619999e-05,
+      "loss": 0.7191,
+      "step": 668
+    },
+    {
+      "epoch": 3.7192494788047252,
+      "grad_norm": 0.1542646207279153,
+      "learning_rate": 1.4575278354330213e-05,
+      "loss": 0.6991,
+      "step": 669
+    },
+    {
+      "epoch": 3.724808895066018,
+      "grad_norm": 0.16046575585370446,
+      "learning_rate": 1.4454959421265029e-05,
+      "loss": 0.7049,
+      "step": 670
+    },
+    {
+      "epoch": 3.7303683113273105,
+      "grad_norm": 0.14105189080116268,
+      "learning_rate": 1.4335029545913556e-05,
+      "loss": 0.7065,
+      "step": 671
+    },
+    {
+      "epoch": 3.735927727588603,
+      "grad_norm": 0.17908408959110106,
+      "learning_rate": 1.4215490554839483e-05,
+      "loss": 0.6959,
+      "step": 672
+    },
+    {
+      "epoch": 3.741487143849896,
+      "grad_norm": 0.1411806231847332,
+      "learning_rate": 1.409634426865321e-05,
+      "loss": 0.6953,
+      "step": 673
+    },
+    {
+      "epoch": 3.747046560111188,
+      "grad_norm": 0.1376363802363308,
+      "learning_rate": 1.397759250198413e-05,
+      "loss": 0.7022,
+      "step": 674
+    },
+    {
+      "epoch": 3.752605976372481,
+      "grad_norm": 0.18325827062078764,
+      "learning_rate": 1.3859237063453002e-05,
+      "loss": 0.6977,
+      "step": 675
+    },
+    {
+      "epoch": 3.7581653926337735,
+      "grad_norm": 0.16076543857018247,
+      "learning_rate": 1.3741279755644427e-05,
+      "loss": 0.7089,
+      "step": 676
+    },
+    {
+      "epoch": 3.763724808895066,
+      "grad_norm": 0.15436087851311603,
+      "learning_rate": 1.3623722375079345e-05,
+      "loss": 0.6907,
+      "step": 677
+    },
+    {
+      "epoch": 3.7692842251563587,
+      "grad_norm": 0.17697113323045366,
+      "learning_rate": 1.3506566712187703e-05,
+      "loss": 0.6964,
+      "step": 678
+    },
+    {
+      "epoch": 3.774843641417651,
+      "grad_norm": 0.14034086311040858,
+      "learning_rate": 1.338981455128117e-05,
+      "loss": 0.7001,
+      "step": 679
+    },
+    {
+      "epoch": 3.7804030576789436,
+      "grad_norm": 0.1798120065285811,
+      "learning_rate": 1.3273467670526018e-05,
+      "loss": 0.7137,
+      "step": 680
+    },
+    {
+      "epoch": 3.7859624739402364,
+      "grad_norm": 0.14177752914344133,
+      "learning_rate": 1.3157527841915951e-05,
+      "loss": 0.6985,
+      "step": 681
+    },
+    {
+      "epoch": 3.791521890201529,
+      "grad_norm": 0.15387764621799602,
+      "learning_rate": 1.3041996831245172e-05,
+      "loss": 0.7127,
+      "step": 682
+    },
+    {
+      "epoch": 3.7970813064628213,
+      "grad_norm": 0.1383432826118217,
+      "learning_rate": 1.292687639808147e-05,
+      "loss": 0.6918,
+      "step": 683
+    },
+    {
+      "epoch": 3.802640722724114,
+      "grad_norm": 0.1322044138065007,
+      "learning_rate": 1.2812168295739466e-05,
+      "loss": 0.6926,
+      "step": 684
+    },
+    {
+      "epoch": 3.8082001389854065,
+      "grad_norm": 0.1459059689077956,
+      "learning_rate": 1.2697874271253844e-05,
+      "loss": 0.6878,
+      "step": 685
+    },
+    {
+      "epoch": 3.813759555246699,
+      "grad_norm": 0.13579645760751588,
+      "learning_rate": 1.258399606535277e-05,
+      "loss": 0.7,
+      "step": 686
+    },
+    {
+      "epoch": 3.819318971507992,
+      "grad_norm": 0.1355030336005884,
+      "learning_rate": 1.2470535412431387e-05,
+      "loss": 0.6918,
+      "step": 687
+    },
+    {
+      "epoch": 3.8248783877692842,
+      "grad_norm": 0.12460724345510901,
+      "learning_rate": 1.2357494040525416e-05,
+      "loss": 0.709,
+      "step": 688
+    },
+    {
+      "epoch": 3.8304378040305767,
+      "grad_norm": 0.1401072983412368,
+      "learning_rate": 1.2244873671284788e-05,
+      "loss": 0.7081,
+      "step": 689
+    },
+    {
+      "epoch": 3.8359972202918695,
+      "grad_norm": 0.1410007273902405,
+      "learning_rate": 1.2132676019947493e-05,
+      "loss": 0.7073,
+      "step": 690
+    },
+    {
+      "epoch": 3.841556636553162,
+      "grad_norm": 0.1347022225437132,
+      "learning_rate": 1.2020902795313352e-05,
+      "loss": 0.6991,
+      "step": 691
+    },
+    {
+      "epoch": 3.8471160528144543,
+      "grad_norm": 0.16412649743147234,
+      "learning_rate": 1.1909555699718118e-05,
+      "loss": 0.6948,
+      "step": 692
+    },
+    {
+      "epoch": 3.852675469075747,
+      "grad_norm": 0.1304705458644605,
+      "learning_rate": 1.179863642900744e-05,
+      "loss": 0.7176,
+      "step": 693
+    },
+    {
+      "epoch": 3.8582348853370396,
+      "grad_norm": 0.201843338518629,
+      "learning_rate": 1.1688146672511143e-05,
+      "loss": 0.7149,
+      "step": 694
+    },
+    {
+      "epoch": 3.863794301598332,
+      "grad_norm": 0.13997227423899886,
+      "learning_rate": 1.1578088113017337e-05,
+      "loss": 0.7015,
+      "step": 695
+    },
+    {
+      "epoch": 3.869353717859625,
+      "grad_norm": 0.16701470491046372,
+      "learning_rate": 1.1468462426746988e-05,
+      "loss": 0.6942,
+      "step": 696
+    },
+    {
+      "epoch": 3.8749131341209173,
+      "grad_norm": 0.1504863826631891,
+      "learning_rate": 1.1359271283328224e-05,
+      "loss": 0.7117,
+      "step": 697
+    },
+    {
+      "epoch": 3.8804725503822097,
+      "grad_norm": 0.15305924367225743,
+      "learning_rate": 1.1250516345771016e-05,
+      "loss": 0.7046,
+      "step": 698
+    },
+    {
+      "epoch": 3.8860319666435026,
+      "grad_norm": 0.14203590464816526,
+      "learning_rate": 1.1142199270441774e-05,
+      "loss": 0.6949,
+      "step": 699
+    },
+    {
+      "epoch": 3.891591382904795,
+      "grad_norm": 0.14514210360106045,
+      "learning_rate": 1.1034321707038153e-05,
+      "loss": 0.7134,
+      "step": 700
+    },
+    {
+      "epoch": 3.8971507991660874,
+      "grad_norm": 0.143692077723024,
+      "learning_rate": 1.092688529856393e-05,
+      "loss": 0.6996,
+      "step": 701
+    },
+    {
+      "epoch": 3.9027102154273803,
+      "grad_norm": 0.12831987823898527,
+      "learning_rate": 1.0819891681303973e-05,
+      "loss": 0.712,
+      "step": 702
+    },
+    {
+      "epoch": 3.9082696316886727,
+      "grad_norm": 0.15655538061615665,
+      "learning_rate": 1.0713342484799342e-05,
+      "loss": 0.7124,
+      "step": 703
+    },
+    {
+      "epoch": 3.913829047949965,
+      "grad_norm": 0.13774726018900033,
+      "learning_rate": 1.060723933182243e-05,
+      "loss": 0.6946,
+      "step": 704
+    },
+    {
+      "epoch": 3.919388464211258,
+      "grad_norm": 0.15042270167507696,
+      "learning_rate": 1.0501583838352269e-05,
+      "loss": 0.7101,
+      "step": 705
+    },
+    {
+      "epoch": 3.9249478804725504,
+      "grad_norm": 0.14701679045150232,
+      "learning_rate": 1.0396377613549924e-05,
+      "loss": 0.6975,
+      "step": 706
+    },
+    {
+      "epoch": 3.930507296733843,
+      "grad_norm": 0.1348037007073666,
+      "learning_rate": 1.0291622259733995e-05,
+      "loss": 0.6898,
+      "step": 707
+    },
+    {
+      "epoch": 3.9360667129951357,
+      "grad_norm": 0.14628526905379458,
+      "learning_rate": 1.0187319372356175e-05,
+      "loss": 0.6942,
+      "step": 708
+    },
+    {
+      "epoch": 3.941626129256428,
+      "grad_norm": 0.13615945281914504,
+      "learning_rate": 1.0083470539976986e-05,
+      "loss": 0.7064,
+      "step": 709
+    },
+    {
+      "epoch": 3.9471855455177205,
+      "grad_norm": 0.14466139658327062,
+      "learning_rate": 9.980077344241547e-06,
+      "loss": 0.7026,
+      "step": 710
+    },
+    {
+      "epoch": 3.9527449617790134,
+      "grad_norm": 0.13153285118835809,
+      "learning_rate": 9.877141359855567e-06,
+      "loss": 0.7,
+      "step": 711
+    },
+    {
+      "epoch": 3.9583043780403058,
+      "grad_norm": 0.1419211581773803,
+      "learning_rate": 9.774664154561249e-06,
+      "loss": 0.6884,
+      "step": 712
+    },
+    {
+      "epoch": 3.963863794301598,
+      "grad_norm": 0.1301976817383647,
+      "learning_rate": 9.672647289113497e-06,
+      "loss": 0.698,
+      "step": 713
+    },
+    {
+      "epoch": 3.969423210562891,
+      "grad_norm": 0.12100613516338608,
+      "learning_rate": 9.5710923172561e-06,
+      "loss": 0.6899,
+      "step": 714
+    },
+    {
+      "epoch": 3.9749826268241835,
+      "grad_norm": 0.12619604031849105,
+      "learning_rate": 9.470000785698117e-06,
+      "loss": 0.7149,
+      "step": 715
+    },
+    {
+      "epoch": 3.980542043085476,
+      "grad_norm": 0.13766300137361,
+      "learning_rate": 9.369374234090248e-06,
+      "loss": 0.6991,
+      "step": 716
+    },
+    {
+      "epoch": 3.9861014593467687,
+      "grad_norm": 0.13324844897778138,
+      "learning_rate": 9.269214195001482e-06,
+      "loss": 0.6986,
+      "step": 717
+    },
+    {
+      "epoch": 3.991660875608061,
+      "grad_norm": 0.13768942764899267,
+      "learning_rate": 9.169522193895614e-06,
+      "loss": 0.7133,
+      "step": 718
+    },
+    {
+      "epoch": 3.9972202918693536,
+      "grad_norm": 0.15395342971750262,
+      "learning_rate": 9.070299749108184e-06,
+      "loss": 0.6978,
+      "step": 719
+    },
+    {
+      "epoch": 4.0222376650451706,
+      "grad_norm": 0.19291822566974853,
+      "learning_rate": 8.971548371823205e-06,
+      "loss": 0.6835,
+      "step": 720
+    },
+    {
+      "epoch": 4.0277970813064625,
+      "grad_norm": 0.16867681043248,
+      "learning_rate": 8.873269566050262e-06,
+      "loss": 0.6883,
+      "step": 721
+    },
+    {
+      "epoch": 4.033356497567755,
+      "grad_norm": 0.16326964961944673,
+      "learning_rate": 8.775464828601477e-06,
+      "loss": 0.674,
+      "step": 722
+    },
+    {
+      "epoch": 4.038915913829048,
+      "grad_norm": 0.16270178485101133,
+      "learning_rate": 8.678135649068862e-06,
+      "loss": 0.677,
+      "step": 723
+    },
+    {
+      "epoch": 4.04447533009034,
+      "grad_norm": 0.1917386211633971,
+      "learning_rate": 8.581283509801501e-06,
+      "loss": 0.6761,
+      "step": 724
+    },
+    {
+      "epoch": 4.050034746351633,
+      "grad_norm": 0.1811245778611519,
+      "learning_rate": 8.484909885883073e-06,
+      "loss": 0.6744,
+      "step": 725
+    },
+    {
+      "epoch": 4.055594162612926,
+      "grad_norm": 0.18514509335495727,
+      "learning_rate": 8.389016245109305e-06,
+      "loss": 0.6703,
+      "step": 726
+    },
+    {
+      "epoch": 4.061153578874218,
+      "grad_norm": 0.18858746875253857,
+      "learning_rate": 8.293604047965677e-06,
+      "loss": 0.6671,
+      "step": 727
+    },
+    {
+      "epoch": 4.066712995135511,
+      "grad_norm": 0.1850999009534838,
+      "learning_rate": 8.19867474760514e-06,
+      "loss": 0.679,
+      "step": 728
+    },
+    {
+      "epoch": 4.072272411396804,
+      "grad_norm": 0.17959193780762223,
+      "learning_rate": 8.104229789825994e-06,
+      "loss": 0.6812,
+      "step": 729
+    },
+    {
+      "epoch": 4.077831827658096,
+      "grad_norm": 0.16993387051766873,
+      "learning_rate": 8.010270613049918e-06,
+      "loss": 0.6646,
+      "step": 730
+    },
+    {
+      "epoch": 4.0833912439193885,
+      "grad_norm": 0.18321014233141378,
+      "learning_rate": 7.916798648299964e-06,
+      "loss": 0.672,
+      "step": 731
+    },
+    {
+      "epoch": 4.088950660180681,
+      "grad_norm": 0.1760023318041356,
+      "learning_rate": 7.823815319178844e-06,
+      "loss": 0.6757,
+      "step": 732
+    },
+    {
+      "epoch": 4.094510076441973,
+      "grad_norm": 0.16517402304449183,
+      "learning_rate": 7.731322041847207e-06,
+      "loss": 0.6966,
+      "step": 733
+    },
+    {
+      "epoch": 4.100069492703266,
+      "grad_norm": 0.1534852983464337,
+      "learning_rate": 7.639320225002106e-06,
+      "loss": 0.6849,
+      "step": 734
+    },
+    {
+      "epoch": 4.105628908964559,
+      "grad_norm": 0.17441227331833753,
+      "learning_rate": 7.547811269855501e-06,
+      "loss": 0.6922,
+      "step": 735
+    },
+    {
+      "epoch": 4.111188325225851,
+      "grad_norm": 0.14483951588357777,
+      "learning_rate": 7.456796570112939e-06,
+      "loss": 0.6896,
+      "step": 736
+    },
+    {
+      "epoch": 4.116747741487144,
+      "grad_norm": 0.12347091194252453,
+      "learning_rate": 7.366277511952326e-06,
+      "loss": 0.6781,
+      "step": 737
+    },
+    {
+      "epoch": 4.122307157748437,
+      "grad_norm": 0.18498892500395253,
+      "learning_rate": 7.2762554740028445e-06,
+      "loss": 0.6781,
+      "step": 738
+    },
+    {
+      "epoch": 4.127866574009729,
+      "grad_norm": 0.253100855454218,
+      "learning_rate": 7.186731827323883e-06,
+      "loss": 0.6783,
+      "step": 739
+    },
+    {
+      "epoch": 4.1334259902710215,
+      "grad_norm": 0.12547901205381473,
+      "learning_rate": 7.097707935384232e-06,
+      "loss": 0.6726,
+      "step": 740
+    },
+    {
+      "epoch": 4.138985406532314,
+      "grad_norm": 0.13373732748809097,
+      "learning_rate": 7.009185154041258e-06,
+      "loss": 0.6751,
+      "step": 741
+    },
+    {
+      "epoch": 4.144544822793606,
+      "grad_norm": 0.1362307104037655,
+      "learning_rate": 6.921164831520322e-06,
+      "loss": 0.6777,
+      "step": 742
+    },
+    {
+      "epoch": 4.150104239054899,
+      "grad_norm": 0.11860525182508581,
+      "learning_rate": 6.833648308394156e-06,
+      "loss": 0.6838,
+      "step": 743
+    },
+    {
+      "epoch": 4.155663655316192,
+      "grad_norm": 0.13403843504100116,
+      "learning_rate": 6.7466369175625475e-06,
+      "loss": 0.6807,
+      "step": 744
+    },
+    {
+      "epoch": 4.161223071577484,
+      "grad_norm": 0.1306791214455748,
+      "learning_rate": 6.660131984231917e-06,
+      "loss": 0.6852,
+      "step": 745
+    },
+    {
+      "epoch": 4.166782487838777,
+      "grad_norm": 0.1234499283506164,
+      "learning_rate": 6.574134825895276e-06,
+      "loss": 0.6936,
+      "step": 746
+    },
+    {
+      "epoch": 4.17234190410007,
+      "grad_norm": 0.11590058368187668,
+      "learning_rate": 6.488646752312031e-06,
+      "loss": 0.6875,
+      "step": 747
+    },
+    {
+      "epoch": 4.177901320361362,
+      "grad_norm": 0.13263371190427117,
+      "learning_rate": 6.403669065488141e-06,
+      "loss": 0.6722,
+      "step": 748
+    },
+    {
+      "epoch": 4.183460736622655,
+      "grad_norm": 0.10833522836358132,
+      "learning_rate": 6.3192030596562e-06,
+      "loss": 0.6703,
+      "step": 749
+    },
+    {
+      "epoch": 4.1890201528839475,
+      "grad_norm": 0.11948792048123115,
+      "learning_rate": 6.235250021255788e-06,
+      "loss": 0.6804,
+      "step": 750
+    },
+    {
+      "epoch": 4.194579569145239,
+      "grad_norm": 0.11643796212883935,
+      "learning_rate": 6.15181122891384e-06,
+      "loss": 0.6835,
+      "step": 751
+    },
+    {
+      "epoch": 4.200138985406532,
+      "grad_norm": 0.12784086684218327,
+      "learning_rate": 6.068887953425213e-06,
+      "loss": 0.7022,
+      "step": 752
+    },
+    {
+      "epoch": 4.205698401667825,
+      "grad_norm": 0.11701035763266034,
+      "learning_rate": 5.986481457733284e-06,
+      "loss": 0.6986,
+      "step": 753
+    },
+    {
+      "epoch": 4.211257817929117,
+      "grad_norm": 0.11133015565758361,
+      "learning_rate": 5.904592996910739e-06,
+      "loss": 0.6767,
+      "step": 754
+    },
+    {
+      "epoch": 4.21681723419041,
+      "grad_norm": 0.1172582650820155,
+      "learning_rate": 5.823223818140458e-06,
+      "loss": 0.6804,
+      "step": 755
+    },
+    {
+      "epoch": 4.222376650451703,
+      "grad_norm": 0.1108105150189601,
+      "learning_rate": 5.7423751606965296e-06,
+      "loss": 0.6763,
+      "step": 756
+    },
+    {
+      "epoch": 4.227936066712995,
+      "grad_norm": 0.10612497316301034,
+      "learning_rate": 5.662048255925357e-06,
+      "loss": 0.6726,
+      "step": 757
+    },
+    {
+      "epoch": 4.233495482974288,
+      "grad_norm": 0.10674977611900638,
+      "learning_rate": 5.582244327226907e-06,
+      "loss": 0.6785,
+      "step": 758
+    },
+    {
+      "epoch": 4.2390548992355805,
+      "grad_norm": 0.11506717156249478,
+      "learning_rate": 5.502964590036093e-06,
+      "loss": 0.6682,
+      "step": 759
+    },
+    {
+      "epoch": 4.2446143154968725,
+      "grad_norm": 0.10977340853370719,
+      "learning_rate": 5.424210251804227e-06,
+      "loss": 0.6891,
+      "step": 760
+    },
+    {
+      "epoch": 4.250173731758165,
+      "grad_norm": 0.10897381874854041,
+      "learning_rate": 5.345982511980707e-06,
+      "loss": 0.6927,
+      "step": 761
+    },
+    {
+      "epoch": 4.255733148019458,
+      "grad_norm": 0.1068366534868816,
+      "learning_rate": 5.268282561994658e-06,
+      "loss": 0.6831,
+      "step": 762
+    },
+    {
+      "epoch": 4.26129256428075,
+      "grad_norm": 0.11704849472765765,
+      "learning_rate": 5.191111585236828e-06,
+      "loss": 0.6925,
+      "step": 763
+    },
+    {
+      "epoch": 4.266851980542043,
+      "grad_norm": 0.1082316966515285,
+      "learning_rate": 5.114470757041572e-06,
+      "loss": 0.6845,
+      "step": 764
+    },
+    {
+      "epoch": 4.272411396803336,
+      "grad_norm": 0.11684449251733979,
+      "learning_rate": 5.038361244668952e-06,
+      "loss": 0.6915,
+      "step": 765
+    },
+    {
+      "epoch": 4.277970813064628,
+      "grad_norm": 0.1076252680934964,
+      "learning_rate": 4.96278420728693e-06,
+      "loss": 0.6757,
+      "step": 766
+    },
+    {
+      "epoch": 4.283530229325921,
+      "grad_norm": 0.11124587762106175,
+      "learning_rate": 4.887740795953764e-06,
+      "loss": 0.6922,
+      "step": 767
+    },
+    {
+      "epoch": 4.289089645587214,
+      "grad_norm": 0.11190635320264125,
+      "learning_rate": 4.813232153600393e-06,
+      "loss": 0.6847,
+      "step": 768
+    },
+    {
+      "epoch": 4.294649061848506,
+      "grad_norm": 0.1114491984468364,
+      "learning_rate": 4.739259415013138e-06,
+      "loss": 0.6761,
+      "step": 769
+    },
+    {
+      "epoch": 4.3002084781097984,
+      "grad_norm": 0.10315313503375144,
+      "learning_rate": 4.6658237068163285e-06,
+      "loss": 0.6729,
+      "step": 770
+    },
+    {
+      "epoch": 4.305767894371091,
+      "grad_norm": 0.11278503790768903,
+      "learning_rate": 4.5929261474552164e-06,
+      "loss": 0.6765,
+      "step": 771
+    },
+    {
+      "epoch": 4.311327310632383,
+      "grad_norm": 0.11395222910887802,
+      "learning_rate": 4.520567847178847e-06,
+      "loss": 0.6922,
+      "step": 772
+    },
+    {
+      "epoch": 4.316886726893676,
+      "grad_norm": 0.11402328760203634,
+      "learning_rate": 4.4487499080232685e-06,
+      "loss": 0.6757,
+      "step": 773
+    },
+    {
+      "epoch": 4.322446143154969,
+      "grad_norm": 0.11252026920394785,
+      "learning_rate": 4.377473423794643e-06,
+      "loss": 0.6827,
+      "step": 774
+    },
+    {
+      "epoch": 4.328005559416261,
+      "grad_norm": 0.10596243645246563,
+      "learning_rate": 4.30673948005266e-06,
+      "loss": 0.6798,
+      "step": 775
+    },
+    {
+      "epoch": 4.333564975677554,
+      "grad_norm": 0.11097354795039509,
+      "learning_rate": 4.236549154093954e-06,
+      "loss": 0.6639,
+      "step": 776
+    },
+    {
+      "epoch": 4.339124391938847,
+      "grad_norm": 0.11133822797610729,
+      "learning_rate": 4.1669035149357255e-06,
+      "loss": 0.6843,
+      "step": 777
+    },
+    {
+      "epoch": 4.344683808200139,
+      "grad_norm": 0.10566152940069157,
+      "learning_rate": 4.097803623299443e-06,
+      "loss": 0.68,
+      "step": 778
+    },
+    {
+      "epoch": 4.3502432244614315,
+      "grad_norm": 0.11005589437352328,
+      "learning_rate": 4.029250531594722e-06,
+      "loss": 0.6808,
+      "step": 779
+    },
+    {
+      "epoch": 4.355802640722724,
+      "grad_norm": 0.11436092029090196,
+      "learning_rate": 3.961245283903239e-06,
+      "loss": 0.6937,
+      "step": 780
+    },
+    {
+      "epoch": 4.361362056984016,
+      "grad_norm": 0.1084924543952515,
+      "learning_rate": 3.893788915962873e-06,
+      "loss": 0.6851,
+      "step": 781
+    },
+    {
+      "epoch": 4.366921473245309,
+      "grad_norm": 0.10292395998913514,
+      "learning_rate": 3.8268824551519214e-06,
+      "loss": 0.6791,
+      "step": 782
+    },
+    {
+      "epoch": 4.372480889506602,
+      "grad_norm": 0.11533814757630266,
+      "learning_rate": 3.760526920473462e-06,
+      "loss": 0.6833,
+      "step": 783
+    },
+    {
+      "epoch": 4.378040305767894,
+      "grad_norm": 0.10749393726522205,
+      "learning_rate": 3.6947233225397993e-06,
+      "loss": 0.6778,
+      "step": 784
+    },
+    {
+      "epoch": 4.383599722029187,
+      "grad_norm": 0.10782596391834809,
+      "learning_rate": 3.62947266355711e-06,
+      "loss": 0.678,
+      "step": 785
+    },
+    {
+      "epoch": 4.38915913829048,
+      "grad_norm": 0.10920407631291443,
+      "learning_rate": 3.5647759373101585e-06,
+      "loss": 0.6756,
+      "step": 786
+    },
+    {
+      "epoch": 4.394718554551772,
+      "grad_norm": 0.10267228546675215,
+      "learning_rate": 3.500634129147167e-06,
+      "loss": 0.6741,
+      "step": 787
+    },
+    {
+      "epoch": 4.400277970813065,
+      "grad_norm": 0.11750192758379473,
+      "learning_rate": 3.437048215964827e-06,
+      "loss": 0.6983,
+      "step": 788
+    },
+    {
+      "epoch": 4.4058373870743575,
+      "grad_norm": 0.09915445619844299,
+      "learning_rate": 3.3740191661933764e-06,
+      "loss": 0.6847,
+      "step": 789
+    },
+    {
+      "epoch": 4.411396803335649,
+      "grad_norm": 0.10665441100795098,
+      "learning_rate": 3.311547939781887e-06,
+      "loss": 0.6731,
+      "step": 790
+    },
+    {
+      "epoch": 4.416956219596942,
+      "grad_norm": 0.11341264840729572,
+      "learning_rate": 3.2496354881836268e-06,
+      "loss": 0.69,
+      "step": 791
+    },
+    {
+      "epoch": 4.422515635858235,
+      "grad_norm": 0.10742879505946441,
+      "learning_rate": 3.188282754341603e-06,
+      "loss": 0.6722,
+      "step": 792
+    },
+    {
+      "epoch": 4.428075052119527,
+      "grad_norm": 0.09857512757834257,
+      "learning_rate": 3.1274906726741317e-06,
+      "loss": 0.6868,
+      "step": 793
+    },
+    {
+      "epoch": 4.43363446838082,
+      "grad_norm": 0.09914583832141616,
+      "learning_rate": 3.067260169060697e-06,
+      "loss": 0.6919,
+      "step": 794
+    },
+    {
+      "epoch": 4.439193884642113,
+      "grad_norm": 0.1053410371476582,
+      "learning_rate": 3.0075921608277415e-06,
+      "loss": 0.6728,
+      "step": 795
+    },
+    {
+      "epoch": 4.444753300903405,
+      "grad_norm": 0.10244450162235408,
+      "learning_rate": 2.9484875567348117e-06,
+      "loss": 0.6733,
+      "step": 796
+    },
+    {
+      "epoch": 4.450312717164698,
+      "grad_norm": 0.10827946870195214,
+      "learning_rate": 2.8899472569606166e-06,
+      "loss": 0.6817,
+      "step": 797
+    },
+    {
+      "epoch": 4.4558721334259905,
+      "grad_norm": 0.11096751325342262,
+      "learning_rate": 2.8319721530894084e-06,
+      "loss": 0.6833,
+      "step": 798
+    },
+    {
+      "epoch": 4.4614315496872825,
+      "grad_norm": 0.10884842138622,
+      "learning_rate": 2.7745631280973005e-06,
+      "loss": 0.6857,
+      "step": 799
+    },
+    {
+      "epoch": 4.466990965948575,
+      "grad_norm": 0.1061505303127893,
+      "learning_rate": 2.7177210563389178e-06,
+      "loss": 0.6643,
+      "step": 800
+    },
+    {
+      "epoch": 4.472550382209868,
+      "grad_norm": 0.10184960600698661,
+      "learning_rate": 2.6614468035340137e-06,
+      "loss": 0.6838,
+      "step": 801
+    },
+    {
+      "epoch": 4.47810979847116,
+      "grad_norm": 0.10986135159522677,
+      "learning_rate": 2.6057412267543347e-06,
+      "loss": 0.6642,
+      "step": 802
+    },
+    {
+      "epoch": 4.483669214732453,
+      "grad_norm": 0.11858338514040914,
+      "learning_rate": 2.550605174410512e-06,
+      "loss": 0.6896,
+      "step": 803
+    },
+    {
+      "epoch": 4.489228630993746,
+      "grad_norm": 0.11205998179561413,
+      "learning_rate": 2.49603948623919e-06,
+      "loss": 0.6706,
+      "step": 804
+    },
+    {
+      "epoch": 4.494788047255038,
+      "grad_norm": 0.10980379918802778,
+      "learning_rate": 2.442044993290198e-06,
+      "loss": 0.6693,
+      "step": 805
+    },
+    {
+      "epoch": 4.500347463516331,
+      "grad_norm": 0.10176209096587764,
+      "learning_rate": 2.388622517913937e-06,
+      "loss": 0.6951,
+      "step": 806
+    },
+    {
+      "epoch": 4.505906879777624,
+      "grad_norm": 0.11498795165110608,
+      "learning_rate": 2.335772873748807e-06,
+      "loss": 0.6738,
+      "step": 807
+    },
+    {
+      "epoch": 4.511466296038916,
+      "grad_norm": 0.11488527586994095,
+      "learning_rate": 2.283496865708843e-06,
+      "loss": 0.667,
+      "step": 808
+    },
+    {
+      "epoch": 4.517025712300208,
+      "grad_norm": 0.10001308436336462,
+      "learning_rate": 2.231795289971448e-06,
+      "loss": 0.6727,
+      "step": 809
+    },
+    {
+      "epoch": 4.522585128561501,
+      "grad_norm": 0.10142949842798026,
+      "learning_rate": 2.1806689339652864e-06,
+      "loss": 0.6861,
+      "step": 810
+    },
+    {
+      "epoch": 4.528144544822793,
+      "grad_norm": 0.10831719019410518,
+      "learning_rate": 2.130118576358249e-06,
+      "loss": 0.6901,
+      "step": 811
+    },
+    {
+      "epoch": 4.533703961084086,
+      "grad_norm": 0.10533792235643635,
+      "learning_rate": 2.0801449870456424e-06,
+      "loss": 0.6914,
+      "step": 812
+    },
+    {
+      "epoch": 4.539263377345379,
+      "grad_norm": 0.11015380431460803,
+      "learning_rate": 2.0307489271384107e-06,
+      "loss": 0.6742,
+      "step": 813
+    },
+    {
+      "epoch": 4.544822793606671,
+      "grad_norm": 0.10267772411742519,
+      "learning_rate": 1.9819311489516122e-06,
+      "loss": 0.6788,
+      "step": 814
+    },
+    {
+      "epoch": 4.550382209867964,
+      "grad_norm": 0.10352184892836523,
+      "learning_rate": 1.9336923959928855e-06,
+      "loss": 0.6753,
+      "step": 815
+    },
+    {
+      "epoch": 4.555941626129257,
+      "grad_norm": 0.10255910515017448,
+      "learning_rate": 1.8860334029511795e-06,
+      "loss": 0.6888,
+      "step": 816
+    },
+    {
+      "epoch": 4.561501042390549,
+      "grad_norm": 0.1101267679218166,
+      "learning_rate": 1.8389548956855474e-06,
+      "loss": 0.6676,
+      "step": 817
+    },
+    {
+      "epoch": 4.5670604586518415,
+      "grad_norm": 0.09919355138446458,
+      "learning_rate": 1.7924575912140785e-06,
+      "loss": 0.6816,
+      "step": 818
+    },
+    {
+      "epoch": 4.572619874913134,
+      "grad_norm": 0.10048348598330001,
+      "learning_rate": 1.7465421977030094e-06,
+      "loss": 0.668,
+      "step": 819
+    },
+    {
+      "epoch": 4.578179291174426,
+      "grad_norm": 0.09864741465772012,
+      "learning_rate": 1.7012094144558932e-06,
+      "loss": 0.6906,
+      "step": 820
+    },
+    {
+      "epoch": 4.583738707435719,
+      "grad_norm": 0.09852629373371855,
+      "learning_rate": 1.656459931903016e-06,
+      "loss": 0.671,
+      "step": 821
+    },
+    {
+      "epoch": 4.589298123697012,
+      "grad_norm": 0.10250147345118527,
+      "learning_rate": 1.6122944315907841e-06,
+      "loss": 0.6904,
+      "step": 822
+    },
+    {
+      "epoch": 4.594857539958304,
+      "grad_norm": 0.10233888167494866,
+      "learning_rate": 1.568713586171451e-06,
+      "loss": 0.6864,
+      "step": 823
+    },
+    {
+      "epoch": 4.600416956219597,
+      "grad_norm": 0.1018547320747838,
+      "learning_rate": 1.5257180593927845e-06,
+      "loss": 0.6739,
+      "step": 824
+    },
+    {
+      "epoch": 4.60597637248089,
+      "grad_norm": 0.10391800179774276,
+      "learning_rate": 1.4833085060880349e-06,
+      "loss": 0.69,
+      "step": 825
+    },
+    {
+      "epoch": 4.611535788742182,
+      "grad_norm": 0.10061052758881242,
+      "learning_rate": 1.4414855721658705e-06,
+      "loss": 0.6771,
+      "step": 826
+    },
+    {
+      "epoch": 4.617095205003475,
+      "grad_norm": 0.09769748536321253,
+      "learning_rate": 1.400249894600636e-06,
+      "loss": 0.6873,
+      "step": 827
+    },
+    {
+      "epoch": 4.6226546212647674,
+      "grad_norm": 0.10073570885444108,
+      "learning_rate": 1.3596021014225858e-06,
+      "loss": 0.6769,
+      "step": 828
+    },
+    {
+      "epoch": 4.628214037526059,
+      "grad_norm": 0.10144089064892822,
+      "learning_rate": 1.3195428117083453e-06,
+      "loss": 0.6814,
+      "step": 829
+    },
+    {
+      "epoch": 4.633773453787352,
+      "grad_norm": 0.09732733650725929,
+      "learning_rate": 1.2800726355714743e-06,
+      "loss": 0.6728,
+      "step": 830
+    },
+    {
+      "epoch": 4.639332870048645,
+      "grad_norm": 0.10872983620981282,
+      "learning_rate": 1.2411921741531807e-06,
+      "loss": 0.6945,
+      "step": 831
+    },
+    {
+      "epoch": 4.644892286309937,
+      "grad_norm": 0.12488470295046429,
+      "learning_rate": 1.2029020196131502e-06,
+      "loss": 0.7003,
+      "step": 832
+    },
+    {
+      "epoch": 4.65045170257123,
+      "grad_norm": 0.11938725767925232,
+      "learning_rate": 1.165202755120558e-06,
+      "loss": 0.6682,
+      "step": 833
+    },
+    {
+      "epoch": 4.656011118832523,
+      "grad_norm": 0.09568930125694097,
+      "learning_rate": 1.1280949548451559e-06,
+      "loss": 0.6832,
+      "step": 834
+    },
+    {
+      "epoch": 4.661570535093815,
+      "grad_norm": 0.0907090050821951,
+      "learning_rate": 1.0915791839485412e-06,
+      "loss": 0.6839,
+      "step": 835
+    },
+    {
+      "epoch": 4.667129951355108,
+      "grad_norm": 0.09108448182810729,
+      "learning_rate": 1.0556559985755377e-06,
+      "loss": 0.6813,
+      "step": 836
+    },
+    {
+      "epoch": 4.6726893676164005,
+      "grad_norm": 0.09467659368857642,
+      "learning_rate": 1.020325945845757e-06,
+      "loss": 0.6887,
+      "step": 837
+    },
+    {
+      "epoch": 4.6782487838776925,
+      "grad_norm": 0.09322239875663856,
+      "learning_rate": 9.855895638452284e-07,
+      "loss": 0.6751,
+      "step": 838
+    },
+    {
+      "epoch": 4.683808200138985,
+      "grad_norm": 0.09515328712397934,
+      "learning_rate": 9.514473816182179e-07,
+      "loss": 0.6704,
+      "step": 839
+    },
+    {
+      "epoch": 4.689367616400278,
+      "grad_norm": 0.10020370379321845,
+      "learning_rate": 9.17899919159182e-07,
+      "loss": 0.6874,
+      "step": 840
+    },
+    {
+      "epoch": 4.69492703266157,
+      "grad_norm": 0.09790464816845096,
+      "learning_rate": 8.849476874048313e-07,
+      "loss": 0.6948,
+      "step": 841
+    },
+    {
+      "epoch": 4.700486448922863,
+      "grad_norm": 0.10255487590164812,
+      "learning_rate": 8.525911882263593e-07,
+      "loss": 0.6719,
+      "step": 842
+    },
+    {
+      "epoch": 4.706045865184156,
+      "grad_norm": 0.10203961568031111,
+      "learning_rate": 8.208309144217907e-07,
+      "loss": 0.6735,
+      "step": 843
+    },
+    {
+      "epoch": 4.711605281445448,
+      "grad_norm": 0.09324266068650292,
+      "learning_rate": 7.896673497084806e-07,
+      "loss": 0.6829,
+      "step": 844
+    },
+    {
+      "epoch": 4.717164697706741,
+      "grad_norm": 0.09193097095894487,
+      "learning_rate": 7.591009687157513e-07,
+      "loss": 0.6727,
+      "step": 845
+    },
+    {
+      "epoch": 4.722724113968034,
+      "grad_norm": 0.09523637486286199,
+      "learning_rate": 7.291322369776587e-07,
+      "loss": 0.6903,
+      "step": 846
+    },
+    {
+      "epoch": 4.728283530229326,
+      "grad_norm": 0.09029936441258628,
+      "learning_rate": 6.997616109258953e-07,
+      "loss": 0.6814,
+      "step": 847
+    },
+    {
+      "epoch": 4.733842946490618,
+      "grad_norm": 0.09621068677262763,
+      "learning_rate": 6.709895378828624e-07,
+      "loss": 0.6812,
+      "step": 848
+    },
+    {
+      "epoch": 4.739402362751911,
+      "grad_norm": 0.10052019299112373,
+      "learning_rate": 6.428164560548134e-07,
+      "loss": 0.6712,
+      "step": 849
+    },
+    {
+      "epoch": 4.744961779013203,
+      "grad_norm": 0.09083034386967587,
+      "learning_rate": 6.152427945252415e-07,
+      "loss": 0.678,
+      "step": 850
+    },
+    {
+      "epoch": 4.750521195274496,
+      "grad_norm": 0.09879226071076684,
+      "learning_rate": 5.882689732482849e-07,
+      "loss": 0.6757,
+      "step": 851
+    },
+    {
+      "epoch": 4.756080611535789,
+      "grad_norm": 0.09553488895974188,
+      "learning_rate": 5.618954030423629e-07,
+      "loss": 0.6752,
+      "step": 852
+    },
+    {
+      "epoch": 4.761640027797081,
+      "grad_norm": 0.09424384994567137,
+      "learning_rate": 5.361224855839231e-07,
+      "loss": 0.683,
+      "step": 853
+    },
+    {
+      "epoch": 4.767199444058374,
+      "grad_norm": 0.09152077250908355,
+      "learning_rate": 5.109506134012954e-07,
+      "loss": 0.6859,
+      "step": 854
+    },
+    {
+      "epoch": 4.772758860319667,
+      "grad_norm": 0.09865066740590055,
+      "learning_rate": 4.863801698687409e-07,
+      "loss": 0.6862,
+      "step": 855
+    },
+    {
+      "epoch": 4.778318276580959,
+      "grad_norm": 0.09925592595855842,
+      "learning_rate": 4.6241152920060817e-07,
+      "loss": 0.6738,
+      "step": 856
+    },
+    {
+      "epoch": 4.7838776928422515,
+      "grad_norm": 0.09697390059397973,
+      "learning_rate": 4.390450564456172e-07,
+      "loss": 0.6896,
+      "step": 857
+    },
+    {
+      "epoch": 4.789437109103544,
+      "grad_norm": 0.0967282255695822,
+      "learning_rate": 4.1628110748132664e-07,
+      "loss": 0.6782,
+      "step": 858
+    },
+    {
+      "epoch": 4.794996525364836,
+      "grad_norm": 0.09442707370223684,
+      "learning_rate": 3.941200290086844e-07,
+      "loss": 0.6816,
+      "step": 859
+    },
+    {
+      "epoch": 4.800555941626129,
+      "grad_norm": 0.09576785298097937,
+      "learning_rate": 3.725621585467698e-07,
+      "loss": 0.6705,
+      "step": 860
+    },
+    {
+      "epoch": 4.806115357887422,
+      "grad_norm": 0.09695945225291408,
+      "learning_rate": 3.5160782442764216e-07,
+      "loss": 0.6899,
+      "step": 861
+    },
+    {
+      "epoch": 4.811674774148714,
+      "grad_norm": 0.09577760816364103,
+      "learning_rate": 3.3125734579134484e-07,
+      "loss": 0.6777,
+      "step": 862
+    },
+    {
+      "epoch": 4.817234190410007,
+      "grad_norm": 0.09338488218320969,
+      "learning_rate": 3.115110325810422e-07,
+      "loss": 0.6854,
+      "step": 863
+    },
+    {
+      "epoch": 4.8227936066713,
+      "grad_norm": 0.0960229378366019,
+      "learning_rate": 2.9236918553829486e-07,
+      "loss": 0.6763,
+      "step": 864
+    },
+    {
+      "epoch": 4.828353022932592,
+      "grad_norm": 0.09329244119901567,
+      "learning_rate": 2.738320961984897e-07,
+      "loss": 0.6897,
+      "step": 865
+    },
+    {
+      "epoch": 4.833912439193885,
+      "grad_norm": 0.09515237568102272,
+      "learning_rate": 2.559000468863948e-07,
+      "loss": 0.6762,
+      "step": 866
+    },
+    {
+      "epoch": 4.839471855455177,
+      "grad_norm": 0.09217529962550147,
+      "learning_rate": 2.3857331071185595e-07,
+      "loss": 0.6844,
+      "step": 867
+    },
+    {
+      "epoch": 4.845031271716469,
+      "grad_norm": 0.09436065825798744,
+      "learning_rate": 2.2185215156563578e-07,
+      "loss": 0.6807,
+      "step": 868
+    },
+    {
+      "epoch": 4.850590687977762,
+      "grad_norm": 0.09523828279039967,
+      "learning_rate": 2.057368241154123e-07,
+      "loss": 0.6864,
+      "step": 869
+    },
+    {
+      "epoch": 4.856150104239055,
+      "grad_norm": 0.09536282527606363,
+      "learning_rate": 1.9022757380186662e-07,
+      "loss": 0.6746,
+      "step": 870
+    },
+    {
+      "epoch": 4.861709520500347,
+      "grad_norm": 0.09068920548317291,
+      "learning_rate": 1.7532463683499258e-07,
+      "loss": 0.6953,
+      "step": 871
+    },
+    {
+      "epoch": 4.86726893676164,
+      "grad_norm": 0.09687334992464874,
+      "learning_rate": 1.6102824019043728e-07,
+      "loss": 0.6804,
+      "step": 872
+    },
+    {
+      "epoch": 4.872828353022933,
+      "grad_norm": 0.1059025814156545,
+      "learning_rate": 1.473386016061129e-07,
+      "loss": 0.6888,
+      "step": 873
+    },
+    {
+      "epoch": 4.878387769284225,
+      "grad_norm": 0.09122337586792513,
+      "learning_rate": 1.3425592957881707e-07,
+      "loss": 0.6764,
+      "step": 874
+    },
+    {
+      "epoch": 4.883947185545518,
+      "grad_norm": 0.08918864835295429,
+      "learning_rate": 1.2178042336111084e-07,
+      "loss": 0.6928,
+      "step": 875
+    },
+    {
+      "epoch": 4.8895066018068105,
+      "grad_norm": 0.09078091143268772,
+      "learning_rate": 1.0991227295824136e-07,
+      "loss": 0.6773,
+      "step": 876
+    },
+    {
+      "epoch": 4.8950660180681025,
+      "grad_norm": 0.10134111972909693,
+      "learning_rate": 9.865165912527286e-08,
+      "loss": 0.689,
+      "step": 877
+    },
+    {
+      "epoch": 4.900625434329395,
+      "grad_norm": 0.09866237453180664,
+      "learning_rate": 8.799875336433339e-08,
+      "loss": 0.674,
+      "step": 878
+    },
+    {
+      "epoch": 4.906184850590688,
+      "grad_norm": 0.09296303972334871,
+      "learning_rate": 7.795371792198136e-08,
+      "loss": 0.6724,
+      "step": 879
+    },
+    {
+      "epoch": 4.91174426685198,
+      "grad_norm": 0.08889541850184303,
+      "learning_rate": 6.851670578677195e-08,
+      "loss": 0.6667,
+      "step": 880
+    },
+    {
+      "epoch": 4.917303683113273,
+      "grad_norm": 0.09207938757906006,
+      "learning_rate": 5.968786068688115e-08,
+      "loss": 0.6853,
+      "step": 881
+    },
+    {
+      "epoch": 4.922863099374566,
+      "grad_norm": 0.09370939668317128,
+      "learning_rate": 5.146731708795649e-08,
+      "loss": 0.6806,
+      "step": 882
+    },
+    {
+      "epoch": 4.928422515635858,
+      "grad_norm": 0.09199672461751096,
+      "learning_rate": 4.3855200191043056e-08,
+      "loss": 0.6859,
+      "step": 883
+    },
+    {
+      "epoch": 4.933981931897151,
+      "grad_norm": 0.09359409102339064,
+      "learning_rate": 3.685162593070057e-08,
+      "loss": 0.6827,
+      "step": 884
+    },
+    {
+      "epoch": 4.939541348158444,
+      "grad_norm": 0.09622079928972942,
+      "learning_rate": 3.0456700973204854e-08,
+      "loss": 0.6887,
+      "step": 885
+    },
+    {
+      "epoch": 4.945100764419736,
+      "grad_norm": 0.08901882738431295,
+      "learning_rate": 2.4670522714957955e-08,
+      "loss": 0.6689,
+      "step": 886
+    },
+    {
+      "epoch": 4.950660180681028,
+      "grad_norm": 0.09115816975859309,
+      "learning_rate": 1.949317928097827e-08,
+      "loss": 0.6838,
+      "step": 887
+    },
+    {
+      "epoch": 4.956219596942321,
+      "grad_norm": 0.08967241218428385,
+      "learning_rate": 1.4924749523572703e-08,
+      "loss": 0.6764,
+      "step": 888
+    },
+    {
+      "epoch": 4.961779013203613,
+      "grad_norm": 0.09584309542019002,
+      "learning_rate": 1.0965303021142071e-08,
+      "loss": 0.6877,
+      "step": 889
+    },
+    {
+      "epoch": 4.967338429464906,
+      "grad_norm": 0.09452425078700585,
+      "learning_rate": 7.614900077093091e-09,
+      "loss": 0.6903,
+      "step": 890
+    },
+    {
+      "epoch": 4.972897845726199,
+      "grad_norm": 0.09503080666295127,
+      "learning_rate": 4.873591718945747e-09,
+      "loss": 0.685,
+      "step": 891
+    },
+    {
+      "epoch": 4.978457261987491,
+      "grad_norm": 0.09096217519727556,
+      "learning_rate": 2.741419697560588e-09,
+      "loss": 0.6779,
+      "step": 892
+    },
+    {
+      "epoch": 4.984016678248784,
+      "grad_norm": 0.08920592010999287,
+      "learning_rate": 1.2184164864725845e-09,
+      "loss": 0.6931,
+      "step": 893
+    },
+    {
+      "epoch": 4.989576094510077,
+      "grad_norm": 0.09316414190683223,
+      "learning_rate": 3.0460528142040033e-10,
+      "loss": 0.6791,
+      "step": 894
+    },
+    {
+      "epoch": 4.995135510771369,
+      "grad_norm": 0.09052975457523055,
+      "learning_rate": 0.0,
+      "loss": 0.6829,
+      "step": 895
+    },
+    {
+      "epoch": 4.995135510771369,
+      "step": 895,
+      "total_flos": 2.3044737561377374e+19,
+      "train_loss": 0.13391224918418757,
+      "train_runtime": 17729.1155,
+      "train_samples_per_second": 25.968,
+      "train_steps_per_second": 0.05
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 895,
+  "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": 2.3044737561377374e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}