diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8687 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 1235,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004048582995951417,
+      "grad_norm": 6.205234861448594,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.8098,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.439159589836823,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.8211,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.380589184814606,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.8285,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 5.894943990341649,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.7855,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 5.672995543445243,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.7973,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.57322730579203,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.7976,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.297306352672515,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.764,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 3.864611316638162,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.7473,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.2578249342305536,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.728,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.0439048105032027,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.7186,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 1.948210177955949,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.7145,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.5781398692343043,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.6945,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.100465049400763,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.7282,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.940338875160914,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.6982,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.931752120925353,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.7097,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 3.041603852663157,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.6678,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.679731645750539,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.6563,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.1729597218562495,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.6481,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.3906615130935143,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.6259,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.240970961935509,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.6139,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.303909966297366,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.6155,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.4144039566039117,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.6221,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.0688486622309799,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.5886,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 0.8255705511240683,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.5767,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.9234238547673832,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.5805,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.8175665290341697,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.5508,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 0.7181405131671256,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.5663,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.8311549300206752,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.5556,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.8584768271795313,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.5603,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.5703045220469919,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.5664,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.7576339489444829,
+      "learning_rate": 1e-05,
+      "loss": 0.563,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.7969562722577272,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.55,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.6341635931517555,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.5275,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.5626511032047168,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.5435,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.7215255001492022,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.5423,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.6114745270681707,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.5393,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6308388334209711,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.5326,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.5701773623909302,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.5022,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.5256738619084894,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.5385,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.6511201128778998,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.5325,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.5625470194872444,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.526,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.48918581431696806,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.5175,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.5565869110994346,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.5144,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.47354976685959066,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.5413,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.5838441270108868,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.4943,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.5095871012815254,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.5017,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.534415016986842,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.5055,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.6545483449368069,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.5018,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.4304937361779375,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.4898,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.6202954570403435,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.5018,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.5524918459836908,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.4898,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.4491788867376422,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.5144,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.698821271860418,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.5009,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.590987519976596,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.4976,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.547987526193415,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.4862,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.6015727700736162,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.4941,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.6262785129728523,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.5051,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.6908180912784855,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.4841,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.6998810052302603,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.4935,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.8238178044981528,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.4885,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.5988784604866073,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.4896,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.6118310322577062,
+      "learning_rate": 2e-05,
+      "loss": 0.4836,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.6725061805526886,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.4892,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.7313706193472439,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.4824,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.724819823862047,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.4865,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.7015614611753488,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.4828,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.7431029458150931,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.4742,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.6062732592272885,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.4688,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.6155633824848673,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.4726,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.8411674482274593,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.4785,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.8025199174450128,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.4742,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.6543626770229942,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.4896,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.8159048225490766,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.4874,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.7700468289790856,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.4746,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.6793490393959218,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.4832,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.9883224476020489,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.48,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 1.3460489858407663,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.4779,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.5388062957212449,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.4851,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 1.4287963349638593,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.47,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.7815321321671302,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.4804,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.8137800287728183,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.4708,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 1.1426378352164521,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.4735,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.8366434222669493,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.4632,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.7414301529630414,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.4754,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.8781936350236231,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.4686,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.9532623763769275,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.4639,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.8124631404193403,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.46,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.7489634006524941,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.4718,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 1.087626000061346,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.4734,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 1.2215005960917495,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.4677,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.8218943942049336,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.4762,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 1.0562897900492514,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.4606,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 1.1220945411616603,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.475,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.9588298043520763,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.4536,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 1.0177729646138223,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.4629,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.9534406333992326,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.4732,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.9970648849418211,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.4569,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 1.3440650788446324,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.4783,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.9706891585463167,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.4532,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.2507151791500588,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.4814,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 1.1850719046077187,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.4701,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 1.2264076758291518,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.4554,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 1.1908052997737473,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.4617,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 1.2383041554997873,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.4678,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 1.1177293731526716,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.4653,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 0.7816402592329814,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.4821,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 1.0839482733577726,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.4563,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.7684719210950609,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.4683,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.9985968940733589,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.4552,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 0.9009069075820254,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.4518,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 1.051117902643297,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.4706,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.9932916910252777,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.4581,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.0251754912524915,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.4625,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.8785656437066527,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.4663,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.9085253740691337,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.4573,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 0.8927138469768108,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.4652,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.8669076032631413,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.4507,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.7919278140686038,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.4586,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.7641871770562819,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.4545,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.9195136085880187,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.4507,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 1.5103607290123577,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.4472,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.6505032205406447,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.4664,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.9766777281479887,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.4428,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 1.352458100025805,
+      "learning_rate": 4e-05,
+      "loss": 0.4752,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.9941102740808624,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.4571,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 1.4304178520059871,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.4375,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.9624763346691738,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.4583,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.5324217220551277,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.4691,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 0.9466771314397988,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.4516,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 1.4551805184522377,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.4695,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 1.1149091050384115,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.4624,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 1.3395512966333256,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.456,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 1.162750894920116,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.4748,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.9659645863261146,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.4427,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 1.1091125664484944,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.4507,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.9912283819832903,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.4614,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 1.0776597904772773,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.4534,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.8139785069022181,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.452,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.9156401622384354,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.459,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.7832160570686941,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.4528,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.7989031004637507,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.4502,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 1.1424477376717754,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.4571,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.7591623646216822,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.4449,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.9295275064207563,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.4579,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.8375682633959708,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.4459,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.7738265568463651,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.4324,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.9279351320754683,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.4407,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.8979151216128812,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.4386,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.6920453673230823,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.45,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.7629972062718235,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.4472,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.92897924344384,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.4557,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 1.0335665028866827,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.4596,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 1.0193973250765165,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.4467,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 1.10098632498704,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.4677,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 1.0610725311495772,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.4509,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.9332532261861294,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.4368,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.6520358079969207,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.4414,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.7072210697080418,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.4531,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.5883950683773503,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.4207,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.530900094949884,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.4429,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.6473125944575453,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.4362,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.7508728140227405,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.4471,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.6758886708176138,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.4376,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.696059456865126,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.4549,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.6712665365251237,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.4415,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.816099897982888,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.4411,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.9144657810997687,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.4461,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 1.0208257684331865,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.4458,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.8768758114890868,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.4398,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.6327965626768028,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.4456,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.5125324587352701,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.4407,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.5400240130526412,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.4374,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.7497665941800526,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.4566,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.903002869119467,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.4324,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.8811229359982147,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.4508,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 1.0023192740359226,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.4503,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.8405441661856669,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.4366,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.5609389283795797,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.4492,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.6085823622058795,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.4295,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.9172915830944195,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.44,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.7356126221691202,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.4304,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.6917001653858973,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.4513,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.8002060882753199,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.4356,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.810112136211395,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.4377,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.7241833878074179,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.4293,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.6908166342678491,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.4341,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.7308301480744525,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.4245,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.7160905733554522,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.4398,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.8545996399262884,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.4523,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.5514583929016714,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.4378,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.6854182550962117,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.4477,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.8579951588579856,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.4371,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.8253135662458213,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.4361,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.6611441464044735,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.4371,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.649806524326919,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.4293,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.6814551821546702,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.4329,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.5534967105635986,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.4383,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.6395330026193269,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.442,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.7245922074527195,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.432,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.6099353010976422,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.438,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.6955672759984498,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.4245,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.7744805230593792,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.441,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.8195647452516337,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.4376,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.7425456048365862,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.4316,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.669865344798947,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.4323,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.7018728665315913,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.4323,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.6835643957884557,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.4385,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.7692539944777835,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.4161,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.6271776710387696,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.4261,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.7206368426784333,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.4115,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.5657914461707463,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.4346,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.5893712037687999,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.4291,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.5509115453576829,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.4223,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.6253026203724981,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.4252,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.6264561102505227,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.438,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.6143050294393535,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.433,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.7552404827242715,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.4402,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.7678400240757643,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.4299,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.6367878332461852,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.4287,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.9204592431980853,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.4251,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 1.0270948199479994,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.4451,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.9316267423064667,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.4413,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 1.0685958334562404,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.4426,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.7084591159257266,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.4294,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.7574651395261833,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.4445,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.7901172800985135,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.4081,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.7727794754416492,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.4183,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.7526273358493795,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.44,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.8847879624328233,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.4373,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.9820398091210053,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.4154,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.7108479959775019,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.4248,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.5578328055782079,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.4193,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.524423562782324,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.4362,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.7293016575450703,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.4382,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.8164050440544427,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.4342,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.7274383506014408,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.4337,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.7151162826018754,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.417,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.5616900868130706,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.4219,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.6620872871939919,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.4254,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.7001074019086237,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.4279,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.5639891906629695,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.4182,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.5477591133578158,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.4207,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.6758833240821043,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.4391,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.6633706497872046,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.4356,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.5946062373139375,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.4116,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.649742768222829,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.4245,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.594142910224814,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.4169,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.6814044132625747,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.4087,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.5857868826498102,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.402,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.6984729790331514,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.3869,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.6248111256031539,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.3901,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.589128293934781,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.4047,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.572095860423845,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.3886,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.40109857477638106,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.3882,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.6545428972277794,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.3838,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 1.0327759817191935,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.398,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.7990577847894725,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.3944,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.6293446646274732,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.3924,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.5454557669381036,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.3909,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.5985435163103952,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.4095,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.7265643990056517,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.3984,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.5931700895444992,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.3843,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.5195126077595559,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.396,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.7153725157992162,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.4029,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.7674482798755854,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.3881,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.5528688937016482,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.3985,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.6414457054919669,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.3974,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.5790619942737665,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.3832,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.5666861249706785,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.396,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.5687409856282354,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.3854,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.5397797105809105,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.3973,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.6646953620313993,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.3966,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.5365071378115381,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.3949,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.5721271645360055,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.3978,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.69205816019789,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.3915,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.47586771751918744,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.3798,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.7770566627414448,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.3957,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.6889965084141015,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.3929,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.597455033638451,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.4038,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.7637762989745905,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.3934,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.463556586960278,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.3869,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.6741159370030054,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.3886,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.6978956937591038,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.3941,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.40630416700799216,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.3948,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.6927267846393514,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.3877,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.6485712572866781,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.3901,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.6435077421316764,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.3944,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.6409813827730181,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.3848,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.513279964522509,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.3875,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.5663797537670312,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.3864,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.5046823813898209,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.3784,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.5745702210423957,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.3908,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.5854211571981596,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.3895,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.5987500270939421,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.4045,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.565657576399544,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.382,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.4093670799785138,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.3762,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.5219910768059558,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.3907,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.5924570939800625,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.3861,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.6281858784429692,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.3911,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.5876568081579141,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.4157,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.43457581658565453,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.3795,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.6648724269332081,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.3801,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.7138494674239412,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.3946,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.6359971949167302,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.3948,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.7032074084091181,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.3877,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.5802212646367035,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.3795,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.5612217433088732,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.3953,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.6488532483523958,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.3952,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.6757420850712045,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.3945,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.6512587685385935,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.3859,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.6162095638473365,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.3944,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.7542112390345727,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.3934,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.5578619878700577,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.3832,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.6578203876716276,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.3885,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.561493072123235,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.395,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.5107224273869957,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.4043,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.5949528177625124,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.3908,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.5339317908397138,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.3758,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.4099331150458228,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.3849,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.639193904681803,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.3905,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.5835583101167083,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.3856,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.5939749696922674,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.3961,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.5621127941390753,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.3941,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.6558421389565603,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.4051,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.552401449336659,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.3797,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.5812218788035067,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.3826,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.7010214783846468,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.3856,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.5113367680995546,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.3813,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.6775388663832761,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.381,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.5925911379740261,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.4017,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.5729746754070636,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.3904,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.6512270713550175,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.3768,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.6236936374671611,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.4037,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.7795094770738187,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.3856,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.6474705922533607,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.3919,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.48707478763143436,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.3949,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.6245432217351637,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.3921,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.42620130721175203,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.3872,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.5297650306082312,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.3884,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.5622391726867889,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.371,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.3986387496144838,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.3817,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.602351905663538,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.3918,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.5831469150516735,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.3832,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.41300502818860774,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.3878,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.6695708707541493,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.3809,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.48049382929015527,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.3911,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.7266430646944866,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.397,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.649903886875103,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.405,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.5734570744863803,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.3941,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.6475626967256354,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.3989,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.5907718546080165,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.3848,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.5753818550275187,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.3896,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.5712104146406147,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.399,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.558525489685879,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.3897,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.5738001110133827,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.3899,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.5392557360738763,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.3812,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.4869892485903321,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.3868,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.5163023550307676,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.3817,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.5667549519600803,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.3976,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.40644523057626547,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.3857,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.5060185620612704,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.403,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.48554006983655934,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.4009,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.5599200295993464,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.3878,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.43551367629116294,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.3868,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.5585506170863435,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.381,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.6117658643949843,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.3899,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.5022714741881177,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.3787,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.5108352872971004,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.3987,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.554049706181031,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.3765,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.6779277170155207,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.3889,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.5544790125754377,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.3933,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.5472001540260888,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.3842,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.5900445603376188,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.3843,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.5711612344718586,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.3928,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.5902848897319952,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.3989,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.8059885001285086,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.3841,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.6454962696687884,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.3921,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.4991822055212697,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.3886,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.46089790308063927,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.3843,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.531170696622426,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.3959,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.4326374610892509,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.3874,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.5511033941857586,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.3884,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.4520258837729466,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.3958,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.4756366051084123,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.3931,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.44897581589935404,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.3929,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.4023099760923694,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.3842,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.43522961370411883,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.3892,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.39066459565213635,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.3952,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.4479836387443633,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.3958,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.4128596594676463,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.3836,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.6107862691923381,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.3977,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.499960108852569,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.3905,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.4576581802538921,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.3874,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.5040798438428713,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.3965,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.553627283646747,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.3832,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.6042069872116371,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.39,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.5658857492830167,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.3926,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.722781810318122,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.3925,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.4568495115293902,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.3825,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.7324600846831593,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.3732,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.6430898154865641,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.3939,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.5415259320519601,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.3861,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.954945560724875,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.4033,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.6131872999092606,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.3945,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.4956628648339723,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.3761,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.5812883296044477,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.3731,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.49865912215457336,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.3767,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.4104280351094852,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.3936,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.5626189989885919,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.3713,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.47194886492858606,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.3966,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.47887310525684734,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.3749,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.4683440547374487,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.3856,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.47259360714113774,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.3739,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.43535756664053576,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.3847,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.4879599846186932,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.3872,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.38897427384292166,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.3797,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.4571230100790579,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.3912,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.48392263212645753,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.3747,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.4371332002454896,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.3785,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.39615403216540607,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.3667,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.44389213404472105,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.3741,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.45283192537268724,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.3794,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.5175483670925398,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.3891,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.4084455825018504,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.3926,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.5370684781030283,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.3974,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.4673089639023956,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.3878,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.5541482993689417,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.3847,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.44999660838257255,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.3982,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.4597103904124748,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.3811,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.4371241750084624,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.3817,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.42594656687825894,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.3801,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.5141533268222891,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.3954,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.44154834759086947,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.3966,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.49421946436396325,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.3808,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.428653177228696,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.4025,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.45518012808153346,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.387,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.5308477134282102,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.3857,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.41277197654078346,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.3856,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.5046814301680719,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.3854,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.5129446785091645,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.3798,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.390708140832756,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.3946,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.5163289626072375,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.3805,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.418041140104646,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.392,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.3731093809011662,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.3745,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.47331749472029216,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.3973,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.47122727626409167,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.3881,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.36917983791377473,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.3974,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.45473193449307187,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.387,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.38374840682626554,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.3755,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.4345648995146401,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.382,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.3989625928657803,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.3856,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.4035558593498326,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.3803,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.4760017110035709,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.3827,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.49827773675710735,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.3869,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.40509673862930884,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.3847,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.4814595664483894,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.3892,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.4285951634580541,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.3741,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.5211397891929354,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.3744,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.47744347466084963,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.3798,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.4806608369545864,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.3861,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.4552749954965176,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.3806,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.4630443400960386,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.3785,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.6635751137916376,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.3844,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.5517262146764772,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.3824,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.40001945531383837,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.3748,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.413863686804026,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.3858,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.5655387062238644,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.3784,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.4762547211727543,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.3832,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.4444705756211302,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.3901,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.39351985038966375,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.3782,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.4900010477282656,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.3774,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.44849861584294753,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.3783,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.506342999116641,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.3829,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.605409590598664,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.3905,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.3522092295004141,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.3844,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.4371194316772678,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.3719,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.42719640479125165,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.3854,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.3949565503430723,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.3894,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.3996712023366973,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.3815,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.35697506415399116,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.3869,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.4585311512215539,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.3841,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.44635415570852244,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.3794,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.41643373898182073,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.3758,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.5091088380600975,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.3794,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.3675162355048332,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.3784,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.4594350430820114,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.3834,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.4165416061662468,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.381,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.3973275015472934,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.3747,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.48207479533544023,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.3894,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.4133630361506054,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.3822,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.4139264314009238,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.3847,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.5305872597545637,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.3744,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.6798148313273555,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.3825,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.4844510029799461,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.3798,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.5116721325986757,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.3338,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.5491751804806969,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.3371,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.5706869923104404,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.3324,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.44563553763984537,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.3392,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.5232979767486394,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.3347,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.5364665096913053,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.3335,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.49455376817742447,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.3397,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.43285054265677275,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.3394,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.47469239599435764,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.3382,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.5647090642879392,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.34,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.5158792194752647,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.329,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.4796148475523577,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.3331,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.5650250515408259,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.3271,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.42928893337201524,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.3226,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.4555888970244126,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.3363,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.5381922621694787,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.3346,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.399943013559176,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.3294,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.4691463265250838,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.3318,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.49554685816017136,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.3317,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.4556581442711429,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.3243,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.45994000286555375,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.3364,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.4784543537626509,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.3356,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.4258163222826067,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.3345,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.4831143176960048,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.3352,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.5519059998063387,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.3266,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.4237497754332536,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.3396,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.4926971358410585,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.3325,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.46899757771182143,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.3345,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.43558268496533786,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.3392,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.3825833800091261,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.3333,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.4785738085984959,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.3465,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.47421599082236265,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.3315,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.4269620411885688,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.3341,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.5444022321114337,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.3318,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.49000414480790666,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.3296,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.38188885776043835,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.319,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.4633951956602257,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.3338,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.47155835644047916,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.3389,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.38250911245993025,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.3436,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.540765452028237,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.343,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.4359795250914236,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.3302,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.43753799705960256,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.3373,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.4334335372295121,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.335,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.5243068905922919,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.3391,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.4355376106138445,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.3213,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.4002883807116916,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.3382,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.4436337365492373,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.3318,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.36613358606685276,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.3362,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.346459442985385,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.3272,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.32517741063905714,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.3306,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.43916035038644813,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.3275,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.33178988495373124,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.3319,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.36725296956490655,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.3358,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.3458932413268092,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.337,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.3947678265877002,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.342,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.36007267732779263,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.3357,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.3756273603664928,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.3353,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.3521115191593858,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.3402,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.34940275084594,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.3354,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.30358365251832176,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.3394,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.31798223184262314,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.3338,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.31909015755517495,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.3363,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.35058384991758157,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.3399,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.3596680202559572,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.3378,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.3428233814315858,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.3316,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.354812030708218,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.3313,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.3116837174083235,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.3307,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.35009621193091517,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.3389,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.34972726777875474,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.3293,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.3172564095146072,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.3353,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.4055852045583838,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.3421,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.2895759865170754,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.3369,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.3468079663737713,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.3383,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.3348546856615597,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.3273,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.32958580224942285,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.3352,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.3600126044210305,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.3323,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.3285425461699581,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.3301,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.36307834345885037,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.3415,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.37845002984299925,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.3229,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.435645538519901,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.3363,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.33233001564888115,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.3275,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.35396447302964995,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.3457,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.3348622578838233,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.3349,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.3262602067891048,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.3407,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.31379287431465036,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.3347,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.2987974484978611,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.3348,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.3306078254014441,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.3278,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.33838786310371743,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.3262,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.3160284523572207,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.3247,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.3728126986877233,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.3299,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.37704644284918926,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.3315,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.3180826356240931,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.3486,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.33203080694470793,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.3226,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.32590276261912265,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.3392,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.36719656382945254,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.3457,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.3609871089799899,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.3339,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.33578740444951943,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.3428,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.39494055884549856,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.3451,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.44653609693488633,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.3281,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.33101206750259016,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.3462,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.427789739037875,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.3412,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.399960992434842,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.3364,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.3308035138985667,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.3259,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.41481225057589854,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.3373,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.3611310468841847,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.3502,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.339562439632638,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.337,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.36953122030238755,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.3343,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.31373892963134276,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.3275,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.3485585453100857,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.3245,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.32934469603543687,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.336,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.3399253826181739,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.33,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.341475620330196,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.3361,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.31727877060881593,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.3363,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.34263932909394984,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.3351,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.27466117372467946,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.3288,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.35526381630958015,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.3295,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.3538326695275714,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.3409,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.2947377260052619,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.3349,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.34333387301885565,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.321,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.3281775093541663,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.3329,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.354498870244311,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.339,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.33428450718680675,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.3319,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.3600501750637626,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.337,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.32286766329938654,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.327,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.3505958686412657,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.333,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.3555569197098744,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.3267,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.3681024839490119,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.3322,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.3316965656328951,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.3363,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.3109766326499792,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.3282,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.33499484212691494,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.3436,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.27201835857643747,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.3239,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.6645313368497602,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.3339,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.33633983514049887,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.3247,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.36757884413591085,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.3419,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.39503463308999326,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.3385,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.343789080021093,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.3255,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.41184839633981246,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.3347,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3067106219761176,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.332,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.46340668179135763,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.3299,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.3595592929921999,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.3318,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.32050797392555935,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.3339,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.33430650796557837,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.3397,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.36775833096103433,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.3288,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 1.1696009837894858,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.3583,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.4070849015529067,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.3483,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.37985452945229425,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.3304,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.3512119820791321,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.3354,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.2900572513064977,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.3364,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.3599118446571598,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.334,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.3335449071913083,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.3391,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.3310859062929236,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.3379,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.3564433452789242,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.3325,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.3139515873646076,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.3248,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.33432440811173325,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.3374,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.3037246281522331,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.3332,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.34194170215588365,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.3353,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.33729233197220004,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.3292,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.3344924613735196,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.3363,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.3045360474816711,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.3271,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.3400177483105601,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.3271,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.2682498349829441,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.3392,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.3182964139431847,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.3322,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.27851949395418285,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.338,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.31871187289360753,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.3413,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.3469817363843695,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.3412,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.31552288620869895,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.3402,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.38372369254171224,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.3381,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.3167630337673698,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.3443,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.3218919274851959,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.3349,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.32024001538803554,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.3305,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.2871682164948501,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.3382,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.3185823093784483,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.3309,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.31289091290341065,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.3328,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.2862809209565419,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.3357,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.3221973519386332,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.3313,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.25358095694233024,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.3361,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.3285578048540662,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.3208,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.2809641012293967,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.3441,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.34507353259315054,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.3263,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.3133311170288755,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.3245,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.35200693762916274,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.3353,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.2743235782323184,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.3284,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.35497782764256397,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.344,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.31479006970289564,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.3282,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.30368484025438164,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.34,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 9.650194619370367,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.3664,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.4824172909011846,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.3306,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.3295356806758759,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.335,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.4208169167785559,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.337,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.3637977255501184,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.3363,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.3720802309486925,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.3346,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.356859643651596,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.3242,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.3307377913066225,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.3291,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.3432707097134321,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.3305,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.3142425301841564,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.324,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.4030788917156947,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.347,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.30394273343969724,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.3364,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.3898445979901117,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.3432,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.3078494724238597,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.3292,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.3618263472777514,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.3374,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.31391988836335966,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.3407,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.3021613750249514,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.3337,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.3499974872673982,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.3346,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.2734777590181389,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.3266,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.3344173661023397,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.3409,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.3205238684058074,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.3347,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.31959143563882814,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.3314,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.36446688284242795,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.339,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.2790612704069962,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.335,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.30993296149506333,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.3346,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.2826626031394004,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.3314,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.2919709190023579,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.3396,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.2839717288499623,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.3307,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.29301252302325925,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.3349,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.277962491641055,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.3263,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.3147337175632768,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.3426,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.25865499945919573,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.3236,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.2887490111692345,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.3329,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.28857054007707766,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.3343,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.2917571615060971,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.331,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.31894445747810823,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.3321,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.3046870600771941,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.3396,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.2809980792178962,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.3283,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.33007727305181406,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.3415,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.29117425460275975,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.3294,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.27287644479354606,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.3319,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.2910080522076395,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.3364,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.285176891793022,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.3356,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.31903689902850263,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.3377,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.28267539520261126,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.329,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.34341210377471687,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.3354,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.27126913073341263,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.3336,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.2991917157623918,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.3291,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.2573455492629154,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.329,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.255823405670753,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.3233,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.27977687640101667,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.3302,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.25773046399596994,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.3331,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.29272622629864264,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.3374,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.2890618617306183,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.332,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.3168056711288643,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.3293,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.257956534478664,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.3227,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.27972187178508,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.3325,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.2819425347348191,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.3415,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.2658001699846072,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.3317,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.26973032717522905,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.3271,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.26414008170562425,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.3392,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.29402619210548087,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.3212,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.3877736779084821,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2917,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.30294776347687297,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.2816,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.47365193084155954,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.2869,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.3975459222317346,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.2853,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.37950852058660023,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.2794,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.3758280349691157,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.2866,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.39319358867091597,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.2853,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.3504821041822426,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.2861,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.3557303701434403,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.277,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.36221956108534514,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.2848,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.30595755242537254,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.2827,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.3264286985863475,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.2803,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.3613998825467719,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.2892,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.30856380528324123,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.2852,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.33031088841795136,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.2755,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.28890810582110327,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2831,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.3103503930811236,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.2816,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.31431192950062087,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.2813,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.2973055981837154,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.2859,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.29673965362677235,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.2806,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.26631879279230647,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.28,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.2694309352835352,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.2878,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.28075652141335133,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.2849,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.2850098627993749,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.2788,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.2512093969745408,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.2898,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.28204411339303803,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.2884,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.2374757831139644,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.2795,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.24210499705071634,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2848,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.25573804787207627,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.2819,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.25647892725026256,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.2809,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.21501647870425222,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.2809,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.2590314743910175,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.284,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.22624748998327732,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.2772,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.252455762784546,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.2837,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.24664408470637486,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.2712,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.25364106950020227,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.2773,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.26199081261506124,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.2923,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.24427172336850858,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.285,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.26258048667746825,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.293,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.2441175852934441,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2879,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.23125580589168523,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.2785,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.2295324085362962,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.2791,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.23697359581559782,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.2829,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.24427752461174784,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.2802,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.2469185752096676,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.2862,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.24332944222352068,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.277,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.2574345927238129,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.2875,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.2431347726402969,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.2699,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.2484137684020233,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.2839,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.23532783003776153,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.2752,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.24165557965224035,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.2828,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.257587899829399,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.2779,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.21855214565000405,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2763,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.24188402534894782,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.2812,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.23691287330504787,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.2879,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.24825967959954834,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.2893,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.22205597740476896,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.2832,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.24234525298388854,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2823,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.21415644280677545,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.279,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.24377166367851555,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2893,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.2409315327361712,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.2826,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.2438187407751783,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.2942,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.2371847212422516,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.2881,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.23358895574425997,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.2859,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.22656863178281714,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.2848,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.22919696426269107,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.2782,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.23933410229617105,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2863,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.2431075744743441,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.2832,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.22886564798834488,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.2796,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.2582816557253591,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.2941,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.23232525143077154,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2819,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.2501633489639947,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.2815,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.2318915832979581,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2819,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.2562886160628778,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.2803,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.2832171492117764,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.2826,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.23107498245786146,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.2829,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.2589264285272323,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2837,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.2211391139697957,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.2811,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.237142129575691,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.2794,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.24212874754688224,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2794,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.2354272484698857,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2826,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.23563412246713838,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.2899,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.24827798341082874,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.2856,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.2664631860186798,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.2822,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.22852196561339877,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.2842,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.2609363216842296,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.28,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.21155777907208548,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2925,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.24161882849573002,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.2869,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.24832902322602377,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.2883,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.2312887239158462,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.2859,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.2599243100368881,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.2786,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.21696446265541242,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.2772,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.24716842260183378,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2872,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.24635037987779443,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.2772,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.2483138000681244,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.2876,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.2562211146428564,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.2911,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.23944188559040136,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.2892,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.24134746811417213,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.2855,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.22156838846743013,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.2863,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.26374397512960057,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.2924,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.2200158668550904,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.2787,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.2186341807103241,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.2782,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.2352782829199123,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.2785,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.25485355231661105,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.2913,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.21761946782421546,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.2741,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.21566542912230557,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.2816,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.22123388793417376,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.2802,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.21029211482632162,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.2873,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.22228336323109757,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.2832,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.217890686209788,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.2806,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.21836805988147492,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.2811,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.24038235257018628,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.281,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.22030406749093023,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2818,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.21595781587182508,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.2891,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.22379837578625944,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.2905,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.2438732621053227,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.2784,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.22226578835270228,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.2858,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.25857151677285284,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.2832,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.2241917862032405,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.2835,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.21626992556137878,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.2766,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.27962704182169085,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2847,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.22661090427509625,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.2821,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.21831852780505412,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.2835,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.2449712033816466,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2801,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.2118284124325468,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.2846,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.21634320085487818,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2681,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.24853091585086678,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.2893,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.20508832277395325,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.2869,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.22622472508956243,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.28,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.2332907415945262,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.2857,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.20937802791526067,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.2865,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.2250126168222682,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.2794,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.2212863084616904,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.2776,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.2075596461677158,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2775,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.21219961347387012,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.2843,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.21472331629947655,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.2793,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.21977141514802995,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2915,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.2134148219130587,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.283,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.2141012532054685,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.2785,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.21572300200758432,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.2771,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.21268114542113953,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.295,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.22023365831489905,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.2873,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.23560358539810794,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.2861,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.21038426441836935,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.2794,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.2244277181401544,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.2849,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.19032283232851932,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.2792,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.21816952626397754,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.2841,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.20736118467558542,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.2735,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.19736180036521797,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.2764,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.20799415536291446,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.284,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.1976222981715159,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.2909,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.2076698806587942,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.2736,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.2163859001529791,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2905,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.20073897709756514,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.2886,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.20924918155895525,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.2873,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.2268302081725793,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.2837,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.19770461877252632,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.285,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.20872724129332013,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.2787,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.2191673803904504,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.2827,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.20299946201259333,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.2767,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.23103351622455826,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.2789,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.22220998667669614,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.2854,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.19273995997410254,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.2831,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.22612759693675127,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.278,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.22967717153337913,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2927,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.19843718645155023,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.2811,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.23587929401633084,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2774,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.21721720942685732,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.2806,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.20443758680136423,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.2833,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.2503749919373555,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.2773,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.22266140190447983,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2898,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.2064651134341291,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2822,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.22985885208392148,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.2723,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.2400084961846152,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.2834,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.22160416155154158,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.2819,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.25323932021865236,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.283,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.23921296935667252,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.2881,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.1997227246106119,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.2815,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.24325293331174874,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2871,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.24358324738130613,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.2879,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.20267105403296096,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.2716,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.2211609485076126,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.2769,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.21747771912146915,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2811,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.21735047102144273,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.2758,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.22485789021824545,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.2841,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.22083682322242176,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.2721,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.21229770077571772,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.286,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.20592584333027325,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.2838,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.2129050599771344,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.2754,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.20687290023017646,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.285,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.2097529873937568,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.2761,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.21674007689436553,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.2841,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.21043010685397742,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.2893,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.2006374104870828,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.2795,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.21150493557727504,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.2829,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.19333587430712756,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.2808,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.1984640596599664,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.2774,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.20869461388070049,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.2847,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.21276219662922177,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.2767,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.19886576562730207,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.2846,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.19550700530366136,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.2827,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.21087851575093383,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.2756,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.19978800396586038,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.2842,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.20847582206712645,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.2825,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.19197777297767069,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.2827,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.20497364217668493,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2741,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.22222599877006302,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.2892,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.1972444030144791,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.2863,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.19916948877495524,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.2874,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.21619952718048543,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.2888,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.19257699357882274,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.2787,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.19228161571594682,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.2801,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.2013576932835271,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.2788,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.2111980938323058,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2879,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.19481313728980137,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.2781,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.19950040755199208,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.2858,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.1952854544939184,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.2803,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.21128510758420493,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.2893,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.19929178789875707,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.2818,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.2044942555582018,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.2821,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.1874865288578094,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.278,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.18461109955609337,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.2816,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.19138799452176714,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.2789,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.19732053896288862,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.2866,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.19642941660008545,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.2798,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.19832985739839276,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.286,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.20061069908288018,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.2902,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.18583520608022852,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.2718,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.2032438109822769,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.2828,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.19750061493588147,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.2933,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.19213621338686643,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.2901,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.1885133906456868,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2805,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.19253675968046433,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.2792,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.20151225797585431,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.2875,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.18335324285747895,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.2799,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.1903664413570097,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.2815,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.18584430589368403,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2784,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.19512196772690446,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2884,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.18337216149653057,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.2775,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.1852538906580186,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.2868,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.22320033111263812,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2843,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.1820586745314744,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.2796,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.18131095987612963,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.2808,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.21273402077453818,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2834,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.18708163777446898,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.2782,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.18851044941085734,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.2784,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.18844609615225524,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.2757,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.31993592319206376,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.2449,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.2514877507220864,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.2458,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.20673227717406054,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.2459,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.2599421999294429,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.2475,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.3181023208376908,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.2494,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.26010713860703977,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.2528,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.25846117554351833,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.2436,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.27867492653573916,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.2536,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.2539569694825023,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.2501,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.22393029855126856,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.2434,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.2321250213171276,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.2438,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.22865791626916096,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.2463,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.20893097827627358,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.2411,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.2116388604329786,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.2431,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.24113648589812925,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.2464,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.22051909608890213,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.2529,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.20012672859097028,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.2488,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.19705958201474227,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.2455,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.20888357253726772,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.2486,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.19892383021849902,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.2407,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.21356616576596774,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.2464,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.1988990538802269,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.246,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.20163485879645746,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.2507,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.21165671673532574,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.2454,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.1995576429361156,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.2369,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.18653722993300728,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.244,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.18480911817580295,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.2407,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.1938004588293189,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.2475,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.18806640801852367,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.2466,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.19619469926693398,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.2488,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.19225730892847384,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.2502,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.19904357486661337,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.2571,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.19780686735089678,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.2427,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.18191813881034866,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.2462,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.1857200763819745,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.242,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.18949160989110048,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.2467,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.1915138775131474,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.2518,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.18234389114448207,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.2439,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.17587147698496658,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.2435,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.18867701549508642,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.2529,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.18086687613249733,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.2444,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.17892987942170718,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.2365,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.18811294220274494,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.244,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.189298217630801,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.2512,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.17227741010216982,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.2431,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.1787149212304926,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.2521,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.17682994526749052,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.2494,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.1794491588030196,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.2437,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.18185657666427854,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.248,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.1845065501358093,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.2422,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.1776158416276642,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.2457,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.17795009314234947,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.2422,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.1769487789956038,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.2458,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.18421360336589768,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.2388,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.17482522457216113,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.2379,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.18084319689391026,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.242,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.1777421311999751,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.2438,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.17396372722240797,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.24,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.17623575389120472,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.2497,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.1761011008151979,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.2409,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.18210775913636768,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.2437,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.17988936138900014,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.2424,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.18720530987543144,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.2503,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.18122422976967612,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.2478,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.19285975221000365,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.2391,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.17894635694246253,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.2401,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.17662489430521836,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.2511,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.18059237468965214,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.2462,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.17784578383076435,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.2483,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.18951408039650608,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.2384,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.18422981302327057,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.257,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.17738084992727723,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.2497,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.174245746040227,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.2367,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.18066713198667447,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.246,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.17410170901754068,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.2323,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.18017052005288361,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.2453,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.17973111065191474,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.244,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.17414236857862772,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.2436,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.1793709193959165,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.2367,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.18227645512304963,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.2445,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.17293798484902573,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.2422,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.18607048648758864,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.2437,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.17200649383514513,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.2435,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.1885196101161657,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.24,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.1862426921697455,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.2464,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.17897721152183524,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.2445,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.17901457309479724,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.2396,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.17556495865959248,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.2431,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.18321373058939977,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.2431,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.18605216525249982,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.2445,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.1831483519371274,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.2445,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.18356392748906256,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.239,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.18309169115110052,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.254,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.17378196362166004,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.2389,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.17718566357750176,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.2338,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.1835344507616818,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.2468,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.174652695510572,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.2315,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.18405769664918253,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.2475,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.17741668038562164,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.2429,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.17774354134772014,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.2426,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.1771982441134443,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.2435,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.17889189991527024,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.2455,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.17352813837423903,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.2439,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.17150471859908797,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.246,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.17600509433891096,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.2462,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.17462288186503672,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.2402,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.17456405044678025,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.2533,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.1744179567537626,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.246,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.17666527502234103,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.245,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.17464105358861867,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.2448,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.17537282819883798,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.238,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.1711218672999087,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.2423,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.17835049764018898,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.2475,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.17666462520711385,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.2439,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.1706676824474976,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.2352,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.19517210674534227,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.2488,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.1773955104085608,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.2519,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.17284442720673757,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.2506,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.17638330202721755,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.2458,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.17181705901550223,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.2424,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.1714192075160826,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.2426,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.16872214591415088,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.2401,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.173037314727984,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.2427,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.1758917513856566,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.24,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.17833900077541082,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.2465,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.1707192165743119,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.2425,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.1790688133322499,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.2519,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.17787638624240423,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.2516,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.17724728132444253,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.2499,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.17193946955972053,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.2505,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.16545073500100668,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.2428,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.17163664077114785,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.2411,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.187182782830994,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.2518,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.16804410266514422,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.2462,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.18057565486046387,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.256,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.1701324410994607,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.2441,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.17056791965662443,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.249,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.1796027213388459,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.2478,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.17665534348427897,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.2525,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.16785640338341248,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.2394,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.18109973408060945,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.2499,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.1661280863730415,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.2428,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.1710653399523274,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.2479,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.17719452037941702,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.2515,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.17725311122161086,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.2472,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.1783704210286494,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.2478,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.17608285592577283,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.2418,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.16719439008823933,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.2474,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.1794361553270752,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.2397,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.17287917505917788,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.2497,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.1765495694565263,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.2441,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.17256083395318694,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.2406,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.17091979117637704,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.2491,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.1667505309173422,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.2396,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.17274316550287544,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.2448,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.18312592090773688,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.2445,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.16633376297549168,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.2463,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.17093096524930373,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.2454,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.17076806535315037,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.2451,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.16749517444535353,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.2494,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.16976123186648653,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.2369,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.17969464124639115,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.2482,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.16950373743260055,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.2447,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.16962440418711147,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.2398,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.18154346637323493,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.2435,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.1664756500182125,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.2422,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.17092224683021293,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.2429,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.1727792714384811,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.234,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.17271426382245217,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.2479,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.17338898508809938,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.2455,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.16401941016524405,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.2481,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.17513998086941568,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.2522,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.1760015595997418,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.2401,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.16909483554782387,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.245,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.17039844517497932,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.2432,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.16658271169780306,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.24,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.16451841052807495,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.2445,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.17487905619073527,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.2477,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.17005941117893686,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.2428,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.1691962162252875,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.2443,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.16480302579883407,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.2465,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.1660834826347569,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.238,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.1736944508883059,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.2462,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.1735912993112853,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.244,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.1646420111630837,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.2449,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.1677264194670337,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.2442,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.17034305079965134,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.2487,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.1678894377029529,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.2449,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.1739305296618495,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.2479,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.1668727700571444,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.2493,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.17288272078226044,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.2522,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.17503958908531145,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.2503,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.1710850418728471,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.2446,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.17031188991209617,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.2471,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.16579576025890438,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.2461,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.1732982680019149,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.2527,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.16942613018483452,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.2448,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.17098125421156488,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.25,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.1689126664195718,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.2365,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.17028498955130714,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.246,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.1712666736296114,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.2458,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.16240122373744825,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.2483,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.16957147423811467,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.247,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.1659475241979864,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.2477,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.16963264942464057,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.2377,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.16284974582193593,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.2432,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.16873759309742217,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.2478,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.1710639869062214,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.2441,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.16913298941434232,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.2453,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.16064332006035628,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.2424,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.16550542223722375,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.246,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.1725629244542351,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.2452,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.17191222001366524,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.2424,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.16970252105669922,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.2476,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.16283575943605993,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.2421,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.168954789589942,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.246,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.16934315556276047,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.2451,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.17290624993342388,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.2504,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.16434164582466831,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.2409,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.16589964476576124,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.2407,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.17033951500614525,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.2407,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.16649312933821853,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.2445,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.16378587789617413,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.245,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.1697176217020348,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.2499,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.16402512456168342,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.2446,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.16681026486316336,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.2524,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.174595615830322,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.2403,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.16387840814956606,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.2448,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.17258525745618247,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.2478,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.17228191525493952,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.2484,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.1672375536292848,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.2448,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.17208763497732246,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.2515,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.17047478914463596,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.2422,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.1701029462010193,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.2488,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.16333665037772913,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.2391,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.16619245872430094,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.2448,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.17042613089944905,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.2484,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.17405940697099911,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.2478,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.1642377970412274,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.2323,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.17017004325427035,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.2417,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.1657834262012629,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.2446,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.1749780885631826,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.2437,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.17293785069448145,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.2467,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.16575945373155665,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.2462,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.16740525818041474,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.246,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.1674471738065622,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.2437,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.16511341927818182,
+      "learning_rate": 0.0,
+      "loss": 0.2384,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 5.052740113642029e+18,
+      "train_loss": 0.34700557460427767,
+      "train_runtime": 58005.5556,
+      "train_samples_per_second": 2.724,
+      "train_steps_per_second": 0.021
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1235,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.052740113642029e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}