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": 6.740385834896804,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.8901,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.920246429742201,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.8927,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.756328521759113,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.8948,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.400797072915948,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.8513,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.2362429705203235,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.8543,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 6.452808219060324,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.8511,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.91629182688498,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.8289,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.713172557174165,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.8117,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.624197956265344,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.7541,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.6434303496395892,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.7389,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.3958597495756515,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.7766,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.553578097859101,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.7264,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 3.7810984140766934,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.7227,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.7468990332182437,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.7007,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.4424304512798627,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.6903,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 2.9756804574618405,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.6696,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.8321709621017455,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.661,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.195382211970952,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.6146,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.6943684273055846,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.6161,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.5647131028291448,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.6091,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.687484195875063,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.6199,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.791713554873221,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.588,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.7488055577825385,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.5559,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.5290802623476083,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.5781,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.9670756399737532,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.5893,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.062370892788722,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.5695,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 1.1777067418582223,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.5664,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.9213228773832565,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.5271,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.7169865312366667,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.5447,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.8197539306060699,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.5606,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.8750397611574102,
+      "learning_rate": 1e-05,
+      "loss": 0.5681,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.7169843865865001,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.5209,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.563909395170388,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.5478,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.7135919489723639,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.5363,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.7797300252000677,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.5302,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.69394368512141,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.5415,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.6049725164769383,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.5479,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.792440634544916,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.5292,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.6270907304369094,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.5254,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.5768620445427862,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.5018,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.7375694037662699,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.5299,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.5544848118130357,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.5081,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.6530400781942998,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.5201,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.641965991901472,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.5594,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.7185129959401229,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.5157,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 1.1748382851414736,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.5586,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.6021164115199034,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.5443,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.6543659413337735,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.5317,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.5722707976127267,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.4767,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.6268055583972744,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.5092,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.5525667147086882,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.5048,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.6856797906792099,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.5027,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.6341424532639707,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.5014,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.5490713096724242,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.5267,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.5814893262482217,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.5118,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.5039886161582569,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.5099,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.5757569701964345,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.4924,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.49072568091036584,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.5081,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.5615997215850098,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.4805,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.5394375374469637,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.5083,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.6406435053337161,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.5208,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.5585616168671116,
+      "learning_rate": 2e-05,
+      "loss": 0.5182,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.5585368328647685,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.4944,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.5608454830970301,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.4853,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.5234440122056468,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.5033,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.5972587983694941,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.4996,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.5494161315230617,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.4984,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.6447366390885845,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.5079,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.6234337776221902,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.518,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.6275858748034182,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.4932,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.6682595402804966,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.5207,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.572948576901077,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.4869,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.6777434940759066,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.494,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.7111020895303044,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.4739,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.7333942220562994,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.4891,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.6514272928888254,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.482,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.8504011240052327,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.5128,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.5618711990913071,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.4722,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.7605649911326771,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.4708,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.6905889878496513,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.5156,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.729697001193802,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.4833,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.59365151891791,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.4921,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.882848334930967,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.48,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.6323172765175958,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.4859,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.7492710920322998,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.4843,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.6322941599265259,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.4724,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.7282494475501473,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.493,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.7685511409952741,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.4933,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.8619080293448267,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.4973,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.712300119439316,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.4711,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.8292481945370408,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.4821,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.7683754435339795,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.4701,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.6340966416347625,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.4952,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.8019008105465946,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.4811,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.6746926526107159,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.4731,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.8175622887172311,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.4781,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.7693829311431553,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.484,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.7798761326272695,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.4877,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.8449294919209344,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.4825,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.349862802832352,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.4898,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.8547833542692935,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.4894,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.6994928583130127,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.4815,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.8722592836060485,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.5093,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.7063119515090477,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.4963,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.7580068019550265,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.5053,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 0.6151986766655261,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.4639,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 1.0064251401152187,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.4854,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.6773401530182601,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.486,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.7976678489013066,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.5023,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 0.6522148993548305,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.4856,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.8134110528610328,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.487,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.9409753529333306,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.4694,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 0.7834860968964302,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.4666,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.6653962546180308,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.4778,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.6971331087857664,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.4829,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 0.5317618375918276,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.4923,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.7474454357986482,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.4862,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.6669880314911852,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.4758,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.7137321532768925,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.481,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.7394969126293961,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.4968,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.6212868571976881,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.4939,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.6219386330231694,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.4579,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.6323549468404666,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.5015,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.6201420818421494,
+      "learning_rate": 4e-05,
+      "loss": 0.4997,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.5797487303860601,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.4852,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 0.580881984306495,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.502,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.6625827085782464,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.4644,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.6119269589251701,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.4924,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 1.2272191600486646,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.4934,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.513520556120108,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.4633,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.6670485191372171,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.4886,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.663096632718563,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.4939,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.6252481203955033,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.4882,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.5490612978186047,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.4841,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.6384110262150869,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.4587,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.5775602805207963,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.4704,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.5106663735374594,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.4941,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.5955721178861322,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.4798,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.5728123790060237,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.4843,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.5670918191954331,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.4812,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.4690596062582823,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.4858,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.5337860264949407,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.4665,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.5116080139935769,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.4994,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.5433525256053604,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.4796,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.5360557046973295,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.4812,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.49464184024019026,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.4737,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.5835032595155188,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.4797,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.512842648612992,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.4753,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.6478012850788085,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.5214,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.50664428296494,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.4775,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.4927779653592698,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.4551,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.557372415341607,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.4865,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.6120452379467499,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.4919,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.6098499193998929,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.4644,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.5595876851040312,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.4658,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.5140613557923895,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.4748,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.590218448461427,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.4803,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.5540075106090432,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.4979,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.5698194620899528,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.4805,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.5655903759888442,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.4922,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.711502870047008,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.4837,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.5572058949059686,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.4782,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.6610068254359778,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.5021,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.5732142557049419,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.4838,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.6208325233429195,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.4664,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.46416671443957913,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.4751,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.5660732183249676,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.5037,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.5949461959045867,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.4693,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.4571156330175842,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.4774,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.5028724347555893,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.4659,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.5262761041237974,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.4895,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.5412240631787528,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.4783,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.5350924942754249,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.4645,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.4962484142927729,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.4904,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.4677711698287888,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.4592,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.5180377950932957,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.4822,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.636447596986452,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.4951,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.5243002052487874,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.476,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.5027795110439414,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.4973,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.5235175024100606,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.4722,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.45858454773802804,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.4829,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.4685319713853591,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.4782,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.596459080142442,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.4851,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.4685967941115949,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.4699,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.5837554740494245,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.4653,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.49055621801246263,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.4766,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.48782513292286833,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.474,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.5479379652613823,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.4919,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.4341879895557755,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.4809,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.461838614778309,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.4707,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.4899136399231559,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.4814,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.47548702883463206,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.4735,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.5087496767714864,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.4631,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.6118406830120656,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.4777,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.5569495289031484,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.4936,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.5285756225265875,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.4655,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.4318586438095664,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.4764,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.4433801390556757,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.4664,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.4228592921346308,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.4709,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.4135470324357291,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.4591,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.49951200516988065,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.4681,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.5265951461152462,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.4479,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.5222971407468757,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.468,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.5902606492028387,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.4717,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.451367023677449,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.4864,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.5893513787550103,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.4695,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.43869800690622024,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.4706,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.5135342248256751,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.4498,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.4734342355107856,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.4859,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.5912128937664437,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.4548,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.49942498842128785,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.4662,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.5277576024489117,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.471,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.48618569812971923,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.4664,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.4898391800167409,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.4698,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.4194896114224963,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.4547,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.4799230967676678,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.4519,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.4983792991328246,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.4776,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.5411803008912188,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.4637,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.48796202964469104,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.4588,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.48375789575567385,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.473,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.48095768490670804,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.4631,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.4780720164724375,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.5041,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.4589040590240593,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.4649,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.5380120876920643,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.4524,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.48342382306115805,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.4787,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.45517380979687927,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.4755,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.4610199460115016,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.4764,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.4774349480773483,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.461,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.45328509246739157,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.4697,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.49805480272286756,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.4673,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.4820422227315985,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.4948,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.4746659901650543,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.4653,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.4729792170974164,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.4702,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.5104847552639836,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.476,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.41287795640535935,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.4751,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.4894385194322158,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.4742,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.4577022514604468,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.4813,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.4557958711667173,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.4625,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.4785169802093376,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.4702,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.4540744734340732,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.4619,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.4434212707754281,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.4723,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.4390908234541278,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.4665,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.48465520782179067,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.4612,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.4441652154059999,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.4869,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.49994074005338385,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.4636,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.4192780165736327,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.432,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5896989835822443,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.4779,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.4874394895855727,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.3785,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.5024123152544872,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.401,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.4912056976397693,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.3942,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.5160147568140342,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.3884,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.5897261116682698,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.3951,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.48509171881070623,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.3761,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.5076118687507739,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.4138,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.5181786870858002,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.4012,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.5430844985855875,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.3728,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.4791662869191057,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.3862,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.4916322331169058,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.356,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.44061329132060373,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.3715,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.5069269950800566,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.3813,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.45378023865021505,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.3642,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.5154068048264729,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.3671,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.4454000771621132,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.3895,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.4162798953985799,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.3906,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.4791830686027796,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.372,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.4857310700984934,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.3971,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.4998983464045228,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.3551,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.4571512696413754,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.3718,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.5510074536657567,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.3765,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.4483724732789123,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.404,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.544898320786372,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.4013,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.4923173297745832,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.3729,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.5531273957813273,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.3798,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.5475048759113466,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.36,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.4848155951496587,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.372,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.5078159486584235,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.3518,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.4935622325986127,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.4049,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.5031376387269626,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.4112,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.5336692983889876,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.3787,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.43131311509652825,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.3928,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.5265821860364409,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.3863,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.47332094612577286,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.431,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.520927260559151,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.3704,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.4979501829941292,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.3644,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.4650390066753633,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.3681,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.4939538334137132,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.3738,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.513229393212166,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.382,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.4517232325553022,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.387,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.5308674564337309,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.383,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.5168236965577124,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.3901,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.5112270829787677,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.3753,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.4807978412059491,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.3756,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.47731059403158393,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.3899,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.538586540854821,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.3942,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.41262963159774907,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.401,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.4949635843699596,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.4016,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.42282492402310884,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.418,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.424570747186379,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.3822,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.3840629320447825,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.4006,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.4478412792663612,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.3861,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.4095732965082129,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.3709,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.40225881065127644,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.3903,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.4235690794547888,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.3952,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.38098363689642184,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.3594,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.47888180905213434,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.3606,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.4991217993915264,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.394,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.446747667645466,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.3882,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.4010701281777994,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.3716,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.44437373191757495,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.3726,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.4219059332273884,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.375,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.39340770006512565,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.3778,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.4304274863910016,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.3825,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.3881520201277579,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.3664,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.44277139001041255,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.3714,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.4241356154657148,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.3841,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.4728889031829457,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.3795,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.43072828111715755,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.3686,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.4706782618093202,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.362,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.43367079679527964,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.3634,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.45837544258983687,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.3776,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.46414812821930135,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.3771,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.45419682416105367,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.384,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.46893056674599265,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.4004,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.4936318087930411,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.3872,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.48230377841849326,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.3905,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.38561421099692605,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.3537,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.6867052750748319,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.3994,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.3760778424080266,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.3963,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.6301323751303246,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.3867,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.484521648678965,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.3762,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.5732975340748635,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.4046,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.5068708230944138,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.373,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.5462525142689983,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.3839,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.48105984440434296,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.3999,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.4768833954324231,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.4043,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.471615254042351,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.3923,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.47387682044518525,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.384,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.4915154851138282,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.3567,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.4296490699032737,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.3702,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.6640391484747535,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.3712,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.3704392707201703,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.3803,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.49326943399775086,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.3688,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.4647872150641248,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.4099,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.4918421836460209,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.4048,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.5085994808885695,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.399,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.4168268704329702,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.3854,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.47020655958074214,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.3678,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.9506743273258478,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.3996,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.4043661357313021,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.367,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.4091717213486011,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.4074,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.4317661168179089,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.3908,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.3905274690872735,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.3808,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.443974289731305,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.3832,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.443535435277106,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.3507,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.47979003783354723,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.3839,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.4435181811025084,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.3781,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.41145124610188355,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.3928,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.4087631885369669,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.3797,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.4429094888171724,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.3786,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.4098633012180486,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.3855,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.38941450645531817,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.3712,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.4171063119169325,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.3737,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.4846159197073734,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.3958,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.40621882463696457,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.3978,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.5202493340299374,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.391,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.4060562835928887,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.3868,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.5075674549506289,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.3782,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.4444778036187676,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.369,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.4953774266535264,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.3955,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.39294852441634187,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.3821,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.4892352702354414,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.3626,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.38475057537088697,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.3718,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.4633946740841796,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.388,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.3864025108559231,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.3738,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.468361138685071,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.3911,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.4071276126408481,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.4041,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.5356692609178367,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.3983,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.397547555736217,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.4146,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.4340927639258484,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.3751,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.41747900223033674,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.3805,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.35955835130319347,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.3771,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.4668243942996388,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.3802,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.38011971774214487,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.4118,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.4142685722714953,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.3781,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.4376920656385786,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.3887,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.3797868708111411,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.4081,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.6089063891152744,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.404,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.3941870422632155,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.3655,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.41519392841349856,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.3764,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.4305695876288022,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.3667,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.38762093410315773,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.3832,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.4772336321659672,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.3893,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.364341687518406,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.3908,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.4259887094684019,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.3745,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.370398539614128,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.3866,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.4273340409301899,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.3973,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.3623790786614363,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.3904,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.44245580153924724,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.3837,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.38144508096311974,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.3856,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.43981765290238817,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.3805,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.4359103218471024,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.3789,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.41115423409776136,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.3993,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.43686096584922257,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.3799,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.4005271064320402,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.3681,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.43429249652318347,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.3874,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.3964526556377579,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.3772,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.4127783057272486,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.3646,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.3785100640395628,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.3624,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.37516119730782976,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.4038,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.37277434000250015,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.3637,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.4094912285573221,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.3956,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.4248775936663197,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.3842,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.4206330841618615,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.3775,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.3715706550204222,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.3532,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.3836868626780221,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.3666,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.41749134958770556,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.3855,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.3735957903395914,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.3819,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.4139057089379323,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.3958,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.41160785029701263,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.3733,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.40414165530628543,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.3836,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.38127945561333537,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.3849,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.3855164720178694,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.3841,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.4826943447353877,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.3804,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.38759388425697433,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.3813,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.4908246882236006,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.4098,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.37406686979483045,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.4049,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.43126331024628906,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.3675,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.41876145233540785,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.401,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.38627564490500027,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.3699,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.4532550698782523,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.3639,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.36811858298655564,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.3783,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.517646673378861,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.3681,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.38078558816247404,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.3901,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.3697187523757208,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.3864,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.3693547641520804,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.3825,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.34634697576773443,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.3573,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.34825845172427505,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.3926,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.34630831189148986,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.3845,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.36811952538124787,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.3779,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.3654278752843546,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.3734,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.3881918957530412,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.4119,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.3687685570284086,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.3612,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.3807707319346911,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.3954,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.38183348085827756,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.3722,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.38995201541071295,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.3828,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.38108107462121255,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.3758,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.36964225972254267,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.3873,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.378480839281557,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.4055,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.3659280126382641,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.405,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.38078858556464457,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.3682,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.3470920749101487,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.3592,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.38240253157292864,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.4046,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.3858691206393752,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.3812,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.39771373416652916,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.4218,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.45919103971259473,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.3908,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.3930477263483372,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.3654,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.40406628794288496,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.3926,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.3935469886695049,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.3948,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.48721970800355,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.3916,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.39205381526490674,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.3769,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.4030328035720233,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.3459,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.39000636790453685,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.3821,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.3705700374796685,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.3869,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.42014174397975157,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.3747,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.36103420329360236,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.3946,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.40399802868882667,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.3871,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.3856589514277633,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.3942,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.39258296887650074,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.3826,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.42605077768585375,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.3819,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.3740960969361372,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.382,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.42572034796565694,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.3923,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.3671399346801679,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.3578,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.3596779384512789,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.376,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.473129390308768,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.4034,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.3394594996885203,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.3588,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.41050795825061087,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.3544,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.4357616802715729,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.4167,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.3752190159794142,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.4016,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.41872248936254564,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.3792,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.3548433511266204,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.4098,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.39184140093392844,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.3657,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.4358037829079541,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.3974,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.39197976058024764,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.3986,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.43742343817654816,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.3807,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.3800995895417488,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.3617,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.4056492221219262,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.3996,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.380328580394458,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.3539,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.4197669593788721,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.3681,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.3824151516162376,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.368,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.3828401601254583,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.3699,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.389572181870289,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.3872,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.3676409019811678,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.3978,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.37681638490771924,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.3864,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.37502515172478734,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.3806,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.5176243001895248,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.3092,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.43926080560264025,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.2771,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.6367751426134431,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.2803,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.48426944770709796,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.243,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.4519800547956986,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.273,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.42349827034301735,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.2863,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.4028294909221444,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.2854,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.4716619849325911,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.2659,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.41151867300985834,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.2869,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.43678003075939137,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.2699,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.4101278181537702,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.2642,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.5302842842920188,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.2901,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.36677160611122206,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.281,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.4965504948576975,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.3229,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.3627340124418424,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.2526,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.47012741458092683,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.3014,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.3717176207661459,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.2563,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.44802939896094135,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.2653,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.4152067548593731,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.301,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.4536257301022609,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.2795,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.42390593868270393,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.281,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.4150968478429568,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.2606,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.40993138796176487,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.2703,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.3832089938486429,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.2625,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.379852241434031,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.2821,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.45535380588366,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.2648,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.370799492344548,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.2814,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.41043135824536064,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.2776,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.4038812781200423,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.2962,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.4015320032288122,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.2774,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.3979855507792342,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.2863,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.3857878027816672,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.2908,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.36243681820611523,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.2881,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.37450039912030914,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.2618,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.40911950121718144,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.2762,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.42780304102508226,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.2648,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.3925775107191047,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.3124,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.40449903034250884,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.2455,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.40315605096795654,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.2989,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.43816885015419954,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.3025,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.37092965953876533,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.2596,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.37529594514093584,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.2584,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.36208520122878113,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.2647,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.3755883878582342,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.2812,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.37290725667543834,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.2833,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.36242128153143066,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.2959,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.3514675711179837,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.2858,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.37060075523923997,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.2548,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.41237005224504214,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.2669,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.37572635491176315,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.2775,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.3494910475503836,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.3,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.3415705345229295,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.2427,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.43177423455008834,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.2571,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.37921174655519974,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.2787,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.3901866328862818,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.2918,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.4189373219225008,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.2775,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.37454405621283904,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.2781,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.3926237632077254,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.2896,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.3634049740970153,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.2815,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.36558338710369914,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.289,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.3824797589525182,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.2885,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.35652462228329074,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.2554,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.4312130518515649,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.2829,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.3671191310994884,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.282,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.40951386427800757,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.2867,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.36407371091133106,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.2738,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.4102518457362343,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.2912,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.38872608973419515,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.2966,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.46756146165875184,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.2912,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.4021853311397074,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.2779,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.36456280683008074,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.2889,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.4039303664884462,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.2687,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.37228245074979793,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.2883,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.44007320939375527,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.2747,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.3871366404399088,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.295,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.42760712855303173,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.2786,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.40696286065525694,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.279,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.3767997281357431,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.2746,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.34605991657279483,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.2844,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.3504540492116704,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.2685,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.3432223919168418,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.2759,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.36638515692869106,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.2757,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.3771812865536888,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.261,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.4387119236342223,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.2925,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.3881924636341501,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.2936,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.4113897858819768,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.2839,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.3882441658387403,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.3153,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.364988685359374,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.2619,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.37496441060504865,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.2706,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.40092793744136185,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.2734,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.39270756236268917,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.2773,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.41591601623664887,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.2667,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.3774207859822856,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.2737,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.3702484062897029,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.2691,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.36176943894499275,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.3148,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.4105287366313257,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.275,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.395788248651059,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.2856,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3995386900319535,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.2866,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.42799690255989775,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.276,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.40192359610258466,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.2813,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.3945938988411901,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.3245,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.3876505872883844,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.2839,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.3837204563105158,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.2796,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.3369434774090093,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.2879,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.3987671271970287,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.2849,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.334835717129818,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.2772,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.4360002841424393,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.2608,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.34711178530307796,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.2797,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.3815806386544806,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.2662,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.35330518367082153,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.276,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.3727148386631637,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.279,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.375399385660405,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.2913,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.3459674498168844,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.2526,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.40907166884263857,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.2694,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.33429090419936447,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.2588,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.3970080568996152,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.2624,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.3191700364983603,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.2795,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.3588720657497929,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.2787,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.3469137858391588,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.2709,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.3486190383684771,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.2889,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.34945402027513234,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.2787,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.3346509673793292,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.3063,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.34130829037355015,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.2823,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.33424511826845466,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.2918,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.3610604888655325,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.2908,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.3517756806244098,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.2952,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.32545515964992583,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.2607,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.37774407649613184,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.2994,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.33142798684013985,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.2551,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.3609765075319366,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.2819,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.36688575579427407,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.2711,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.38539808726082064,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.2756,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.34707204390647595,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.2976,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.367392297311753,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.2886,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.365988662015454,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.304,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.3809002915901562,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.3036,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.3528432292795748,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.2845,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3556417049491436,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.2537,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.3390068504549609,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.2896,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.36050621772394553,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.2786,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.3434561695800572,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.2694,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.36005244317988083,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.2777,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.38177115676286266,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.2763,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.3670783402568686,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.2694,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.3714208736771641,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.2782,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.37085581791853733,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.2652,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.37041011358807063,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.3057,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.36029443714074183,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.3096,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.3453469091238674,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.261,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.3522018827653238,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.2782,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.3724673414075293,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.2967,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.3690475978814498,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.2531,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.3375029297845298,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.2653,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.4127806187717277,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.301,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.3325278234997997,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.2808,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.35817846323742714,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.2777,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.36271413743015235,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.3033,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.3544917902267329,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.3141,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.3866653275322747,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.2855,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.3336996612953003,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.2717,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.3869858782661253,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.2769,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.3414072887667954,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.2594,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.3591409189706269,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.2886,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.40296618699488435,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.2704,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.37859847973893695,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.3053,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.3522055215762954,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.2723,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.3396700364213576,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.2949,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.3613825715169157,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.2677,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.35352552698088996,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.297,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.3793517980583883,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.2918,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.352980451375437,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.3,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.363231466423601,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.2855,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.3418661266004588,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.3169,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.36572077399396535,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.281,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.35274769708788534,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.2917,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.31599313156552317,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.2532,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.35011070356557894,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.2623,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.35144998188961163,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.2891,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.321848083179381,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.2919,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.3555129162454493,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.294,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.356136252665241,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.2489,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.31600951555266016,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.2866,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.37029633786788335,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.2771,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.38986970916543484,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.2676,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.32598705035548936,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.3048,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.33051688211740066,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.2905,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.4319392120988048,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.2717,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.32391572880374103,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.2826,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.3586279491656753,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.2969,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.3720701948410337,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.2799,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.33098926493191133,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.3121,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.37139870960346927,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.2695,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.33917783927623346,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.2763,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.36372853869127203,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.267,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.35453417575348334,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.2733,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.3400084340855158,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.2844,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.3753473050932974,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.2907,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.31349371031052975,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.2667,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 1.0198831702786206,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.3056,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.33423780317589047,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.275,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.32645593449724664,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.2943,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.3402370676530391,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.2807,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 1.7788878921909068,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.2723,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.36231754144610395,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.2669,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.32164158872065196,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.2594,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.32400946069016495,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.2861,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.347821841121233,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.2523,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.34360657859125937,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.2653,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.3805541397542839,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.2837,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.33483410649688156,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.2867,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.3555197090818476,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.298,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.3529655080037139,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.278,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.3485266349418984,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.2655,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.33777158825473075,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.2792,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.3180987039683282,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.2476,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.36684534393481594,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.2627,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.3257608602709413,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.2841,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.35379803131276744,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.2613,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.339011568074374,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.2647,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.3471889734990589,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.277,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.325776309954113,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.2765,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.3202257868582969,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.2982,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.353393063800112,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.2597,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.3137459323040916,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.2716,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.523391911697847,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.2745,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.34785532818583204,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.2822,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.31887070090053626,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.2948,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.3275413571606758,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.2586,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.34437348224492215,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.2717,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.32940180482233145,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.2806,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.3159880959997788,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.2527,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.34526003689337875,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.2825,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.3507040375319101,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.2977,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.3197506488237526,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.2793,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.3177578954373676,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.2806,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.32397532176231125,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.278,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.331618219144697,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.2727,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.30651483589664896,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.2692,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.31884573540881705,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.2608,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.30804795351489844,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.2754,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.31557151623748486,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.2822,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.35152306036888625,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.2592,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.32695202421282266,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.2836,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.33379914605592775,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.2953,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.34579348846876273,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.271,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.3160828631931422,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.2788,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.3264887239588382,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.2598,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.49890525781808914,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2242,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.3072601686643192,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.1972,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.6474646152519135,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.1917,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.5039363944146941,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.1875,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.349916089113549,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.1984,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.4045606798275686,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.1918,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.42760275751460747,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.1842,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.3620788069440049,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.1862,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.324837903135638,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.183,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.40871129971239223,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.1799,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.40162806416076424,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.2032,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.3491688928311875,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.2203,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.35960009200628806,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.2005,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.3718024479184929,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.2017,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.3435309395131916,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.1976,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.3664036403144208,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2164,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.3414769133128347,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.1934,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.31508200670477327,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.1986,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.3279133557108216,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.2002,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.29809872586354125,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.2077,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.2917379552069454,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2179,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.31727431980605025,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.1854,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.29825120598442034,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.2016,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.3065753750856843,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.2047,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.3445600715212902,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.1973,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.3004695116415208,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.1866,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.30613921554329526,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.1889,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.3186033141553785,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2071,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.3223058911607382,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.1918,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.30346355279114395,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.2206,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.2899146203423682,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.2025,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 2.6206885877789405,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.2394,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.34364890200090914,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.1767,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.29493450508534025,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.1897,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.2789252165521922,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.201,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.30755923014957254,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.1979,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.3055959169069634,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.1826,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.2930635749350004,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.2056,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.3110436463641971,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.186,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.2964849478395447,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2148,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.2795015964336447,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.191,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.2990825192239617,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.2214,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.30095928400623123,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.1791,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.29564153632930573,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.2102,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.2789377015391875,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.2156,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.3047277657345958,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.1779,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.2890794163624452,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.1894,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.3074158929583563,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.1954,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.2935848147696264,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.2161,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.29611389587459164,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.17,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.30759385965434083,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.1894,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.31212245827231344,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.194,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.30022751097368583,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2167,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.2815755035792649,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.1968,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.3045620206489717,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.2049,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.3186378557925368,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.1988,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.29323222817737543,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.192,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.3027594760839441,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2107,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.2919000121402253,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.1946,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.3071372999372556,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.1796,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.2998029371723068,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.1814,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.28803164625203564,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.1908,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.3137589585855464,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.2062,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.2924115629165695,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.1962,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.32378273809993896,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.2111,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.30701029931382673,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.1903,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.2987258526037619,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2037,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.31501070072518217,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.1921,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.31471865163450746,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.2062,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.28683538168524386,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.1819,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.29541989540569386,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2298,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.29664250746060344,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.1872,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.28553140236642865,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.207,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.29522295443995644,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.1833,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.2893896549524068,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.1882,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.287318774957385,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.1874,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.29754801997221436,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.1963,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.2894721486226119,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.1838,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.298892474410285,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.199,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.30065252620128424,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2172,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.2828006797789871,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.1867,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.2767599528660379,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.1965,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.27872468269526746,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.1834,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.2905115435316346,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.178,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.2864581921490991,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.173,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.299537921811291,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.2062,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.2862347907139773,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2135,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.27503278958244726,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.1893,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.28055440675023163,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.1968,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.26981734884903735,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.2108,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.293973677626135,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.2076,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.30980349462134105,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.1832,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.29420833607785696,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.1945,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.2975062662805042,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.1893,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.30172705186863313,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.1802,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.283496409229382,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.2078,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.2899804522325341,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.2187,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.29264034462780947,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.1893,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.31344457032653966,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.202,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.28077025895777674,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.212,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.2842524072040799,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.1627,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.33277319752825923,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.1868,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.28745127849359564,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.1787,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.28818793572904544,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.1761,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.2974843623449345,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.2196,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.28160045928738187,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.1806,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.2841753289709473,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.2034,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.2846924964733624,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.2079,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.5132332335895863,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.1799,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.3073376977068637,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.1904,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.2974321767869341,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.213,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.2998864291241181,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.1794,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.29458351436093544,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2005,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.2863424427459185,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.2035,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.2751246553760637,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.1846,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.30755421466383875,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.199,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.29394064814057547,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.2093,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.29511874026593266,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.201,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.28923599235122915,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.1925,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.2941083894414968,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.2071,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.2869338563454239,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2159,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.3039616918961605,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.1586,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.3036888267491569,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.1891,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.30387407035742564,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2073,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.2892821022656449,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.196,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.2932756744117501,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.1749,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.3041181467430715,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.2011,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.2792032959361263,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.1876,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.2831822826606107,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.2123,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.2954487135898219,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.2047,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.28553600860594425,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.188,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.2762114448411096,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.1868,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.26745061167476075,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.2018,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.28322655524417484,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.1837,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.2764064517292345,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.1767,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.2696339628810312,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.1808,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.48861264757617534,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2072,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.278605506260484,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.2025,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.27810744475984145,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.1869,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.2892109835506956,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.1898,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.2856649947396643,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.1909,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.26040734913991337,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.2032,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.26389685819518255,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.1817,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.2853836932595229,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.1985,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.28104106253314587,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.2034,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.29504540263809437,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.181,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.28289232943141435,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.1798,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.2803013184239837,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.1996,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.2753238451807648,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.166,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.29726298427422154,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.1932,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.27881002675707245,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.2191,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.2651867085397501,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.2031,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.2947317318275717,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.1703,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.29971911439889765,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.2007,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.26468323536681554,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.1912,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.3013026692845797,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.2042,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.28621579400146185,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.1848,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.27078925114789526,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.1757,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.31884019906117206,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.1939,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.28832762626713615,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.189,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.28451855888092714,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.1963,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.26674069423841884,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.1667,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.31009388324760895,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.2123,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.3019484734450629,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.1981,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.2832214215523304,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2023,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.2605240616710156,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.2108,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.2811737372425827,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2012,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.28774538124308585,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.1935,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.28640396843098453,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.1828,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.2765367224517338,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.1866,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.2843706084049911,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2125,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.27006807527751736,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.1765,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.27712388338433,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.1975,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.2737604905603402,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.1975,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.284100610863378,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.1922,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.2810212404965406,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.2051,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.27836659550093196,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.1933,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.26280001000515113,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.1923,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.27071750893447444,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.1981,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.27913934311545785,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.1858,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.286825163585507,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.1806,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.276613248740108,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.1892,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.26946092154365664,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2188,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.28512549707637236,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.2053,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.29052710055800485,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.1995,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.2928786243295691,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.2094,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.2786032336882576,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.211,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.2856557598361653,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.1684,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.2885020904803489,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.183,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.27449232513042376,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.1647,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.2916425892712198,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.1761,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.2747379764126054,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.1875,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.2767127703143383,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.193,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.2778648907187701,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.1887,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.2694997396915752,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.1679,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.2766423652969236,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.1884,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.26920760786791587,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.2064,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.2735189458059663,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.2057,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.2794634422560465,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.1914,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.26526045252134217,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.1794,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.2648553262501651,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.2065,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.2757823782247332,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.1627,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.2838027656851305,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.1922,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.28634106992698116,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.1801,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.27695338051102514,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.2046,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.28222147292223937,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2001,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.2773992209001866,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.1961,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.2699388728698982,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.1753,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.279875853051315,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.1821,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.276985241846293,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.1887,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.2765046659664868,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.1961,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.2722489538288802,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.1837,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.2597794658225895,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.1839,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.28138066795058997,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.1974,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.2864148939467885,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.1786,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.2837889083641184,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.1791,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.2641277716027119,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.1738,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.25892428541012114,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.2123,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.26765852151723574,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.1969,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.28766939845061323,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.1994,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.26618903714137426,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.1868,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.2824038032122547,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.1684,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.2621044538401367,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.1929,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.27955699036814985,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.19,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.27352816571066646,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.194,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.261704123519139,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.1828,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.27632184428288636,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.1924,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.26203113736352024,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.207,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.2641732361549359,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.203,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.2636525515739288,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.1727,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.27090795359432124,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.1913,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.276466493528456,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.206,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.26938144517455803,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.1894,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.26577829359775273,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.1817,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.25966765652457674,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.1752,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.26887844055866067,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.2019,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.2821677154457374,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2023,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.2773605749664554,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.1911,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.2774214690308135,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.1663,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.2781795248084092,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.1932,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.27055472649531465,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.1813,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.26321624917605024,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.1947,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.2635558788341588,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.2002,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.2747300998155333,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2007,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.26815035186693204,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.1937,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.27301036764063297,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.2027,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.26190367229341033,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.2035,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.3659663830947102,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.1658,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.32402811287416633,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.1185,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.24925919402213498,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.1507,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.25243288149200227,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.1486,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.3292437698159164,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.1589,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.32705329771610153,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.1502,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.3733390507219955,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.172,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.26170758236008396,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.1362,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.24433498663187,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.1319,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.29651077483078725,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.147,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.32570744271292834,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.1541,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.2898044142604171,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.1479,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.24030940394335729,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.1683,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.23753565416304284,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.1534,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.24973481786302448,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.1468,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.29609737034270794,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.1444,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.2738082426583033,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.1515,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.2410127898233117,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.1494,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.24039567555546226,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.1404,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.23165114023122046,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.1525,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.247001826261612,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.1338,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.23834387611404542,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.1539,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.24343458349979788,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.1424,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.22033064203172575,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.161,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.2436122468546149,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.1441,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.25080563192028904,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.1476,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.23298607066857008,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.1432,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.23577487368572345,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.1568,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.21928892844990178,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.1333,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.23046353595582159,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.1502,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.23105451086506934,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.1462,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.23611010074918337,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.1411,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.221142370066441,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.1395,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.21667613846356382,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.139,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.21201602570079908,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.1734,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.21933652628074973,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.153,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.21657087548250037,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.1567,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.2633537458535954,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.1433,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.23360404810876692,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.152,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.22245973627282697,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.1474,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.22050108790188455,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.1615,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.22955272684684166,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.1346,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.2122679990045934,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.1094,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.21090220974668528,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.1388,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.2067589614313047,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.1223,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.21793330782099612,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.1715,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.22237108853085516,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.1474,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.22422022240509962,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.1332,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.22428402338913792,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.1397,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.21964352997518344,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.1299,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.21632018265744377,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1691,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.22004429468205866,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.158,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.22313754297536056,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.1428,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.20514890634588814,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.1234,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.22489874842431085,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.1676,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.20997886391789125,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.131,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.22567111740598525,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.1426,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.22982559728373045,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.1412,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.2192240713753221,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.1446,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.21054172184767392,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.1533,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.22151529358980812,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.15,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.2132983502152742,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.1532,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.22217847590261836,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.1362,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.2155764314002926,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.138,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.2184754483968967,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.1313,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.21823619253333926,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.1448,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.21108175166568413,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.1689,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.2279246859290953,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.1663,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.2240395510953398,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.1336,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.23328150749691753,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.1487,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.22337070344716578,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.147,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.22328276331640545,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.1574,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.20197236065625346,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.1338,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.2189734277780874,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.1472,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.22038400446636458,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.1448,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.21805179548530015,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.1635,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.22358392778754504,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.1486,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.22263183871432818,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.1659,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.21754779319040615,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.1631,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.21221245894125246,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.133,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.21207552985111777,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.1453,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.21240525935187604,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.1431,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.23128959988112371,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.1407,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.21801940184770333,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.1443,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.21727128611688867,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.1339,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.20993062263726675,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.1291,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.22414576360504557,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.1222,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.21758324567406465,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.1469,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.21455077765601963,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.1603,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.22092384225741038,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.1585,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.21730658700240682,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.1281,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.2197091815814211,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.1441,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.22548770279269637,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.1726,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.23749160934664731,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.1257,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.23036549626846411,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.1637,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.21829676648164487,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.1541,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.2240650514359236,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.147,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.23694493479129372,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.1338,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.22278174013347415,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.1429,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.2074768053824781,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.1407,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.2247332301495699,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.1528,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.22020018964576732,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.1595,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.2145064290580566,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.1442,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.2217430923854829,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.1565,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.21505363085051807,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.1774,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.2139387675554987,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.1598,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.22095081519865153,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.1312,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.21648863792719736,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.1508,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.21654133683102436,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.1515,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.2090956554938466,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.1463,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.21369415842826184,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.1408,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.22238503568365184,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.1586,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 1.3257910416384273,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.1679,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.20944757993491245,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.1622,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.2250499180737349,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.1472,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.2190643903828286,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.1539,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.21282771685403232,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.1168,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.2151969193938169,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.1556,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.21438467410900583,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.1345,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.20967485170247316,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.1475,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.20691776546514648,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.1575,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.21101230955183722,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.0989,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.22182548508963082,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.1308,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.21806871591832683,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.138,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.21875989838721643,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.1403,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.21608575845665742,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.1446,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.22193022390902956,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.1545,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.21923329514132114,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.1362,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.2149490122950632,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.1544,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.22384714149511192,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.1701,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.23349877397569943,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.1527,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.20926274446024357,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.1544,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.20686691263131649,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.1081,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.21968440974745335,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.1622,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 1.1504491469595914,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.1457,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.21878684963657602,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.1784,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.21399956206401172,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.1494,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.21744168573361483,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.1636,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.2134028409356121,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.1553,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.20182515565947604,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.1414,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.3148078201173959,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.1548,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.22112225888271292,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.14,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.21085372804291108,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.1646,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.22384564601737036,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.1456,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.20770977579704553,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.1403,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.2227982577114367,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.1445,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.21155246924558152,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.1537,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.22208602708341824,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.1121,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.21500080463860413,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.1734,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.21722051418865473,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.1374,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.21846064298073975,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.148,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.21154565274883053,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.1373,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.20678749617332243,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.137,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.21870002038213007,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.1618,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.21771365734676582,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.1561,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.22808557429642345,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.1382,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.21797713237543187,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.1228,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.2128266820564796,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.1678,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.20804708074656011,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.1343,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.21070435709385804,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.1393,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.23437521776778675,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.1293,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.21052053198916318,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.1292,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.2343818954520919,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.1354,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.21545047991132568,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.163,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.21709612996791197,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.1689,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.21260036028334678,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.1523,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.2147472793986078,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.1328,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.20920909434610352,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.1391,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.20897476829620548,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.126,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.2065768581662248,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.1411,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.21467264816604745,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.1365,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.2141275195313549,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.1469,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.22151221084921607,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.133,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.21443294431653404,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1485,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.2015349344958383,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.134,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.22230283685869948,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.1231,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.22161784328972875,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.1657,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.21566930603425222,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.1504,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.2149000582062493,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.1379,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.2193489807558439,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.1462,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.21297370760255174,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.1546,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.21084667138965363,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.1318,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.24262565115959966,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.139,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.21647516640543946,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.1254,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.22673249959262418,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.1517,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.19828910582664871,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.1358,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.2183077214096187,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.139,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.2142404041456037,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.1613,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.21156054972514474,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.1736,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.22036659886691976,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.1679,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.21234987385584908,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.1551,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.22386471405318398,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.1546,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.20397107919903132,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.1191,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.21165265469851932,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.1368,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.21146427216248537,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.1402,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.21720113393539073,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.1276,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.21891520974139006,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.1498,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.21486010244410958,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.1355,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.22304579023279575,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.1393,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.2120044470044638,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.1335,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.21918609686587776,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.147,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.2069639098160405,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.1436,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.21073281429632057,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.1319,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.23919555049328628,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.1648,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.21265820532356125,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.1312,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.21496554043469576,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.1282,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.2263024426598275,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.1632,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.2198928612185592,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.1441,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.21420916840511228,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.1313,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.21909426202352777,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.1397,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.21471933823883288,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.1325,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.21548708553535942,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.1304,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.2246022271769227,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.1649,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.21325530343442547,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.1383,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.2178145083385332,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.1395,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.2220556627245492,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.1198,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.20503158143802447,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.1367,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.21419153456890355,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.1597,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.21567074342806,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.1659,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.21844059071048255,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.1427,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.22158102871185703,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.1467,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.21766022493092005,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.1487,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.22193509276605541,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.1253,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.21710885366109967,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.1157,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.21011510010710105,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.128,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.22081544826657604,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.1597,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.22773526372220854,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.1327,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.21155258579843356,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.1297,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.21981624005398917,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.1573,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.22188981890062653,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.1505,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.2158847150263269,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.1711,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.21409255387614307,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.1403,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.2202016517883113,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.1363,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.21755429617890332,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.1533,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.22605264631278918,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.168,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.2244882537291937,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.1466,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.2236018812759928,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.1616,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.21765369751912522,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.1284,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.20589612945427302,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.1435,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.23638954666048237,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.1507,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.23261446676617017,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.1221,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.2134189852824687,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.1205,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.2353428163400785,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.1222,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.20355128995366056,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.147,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.21689152908285508,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.1358,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.21262415551852695,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.1377,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.22210269505854566,
+      "learning_rate": 0.0,
+      "loss": 0.1682,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 1.962803781229871e+18,
+      "train_loss": 0.30276708501553246,
+      "train_runtime": 23532.6537,
+      "train_samples_per_second": 6.714,
+      "train_steps_per_second": 0.052
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1235,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.962803781229871e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}