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": 6.0209985757066,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.7833,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.140660121309178,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.7939,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.072756867335653,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.784,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.137887933758524,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.8165,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 5.599942479075599,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.7751,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.5518999693284705,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.7972,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.231536627130041,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.7519,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 3.9189693765771314,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.753,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.2424610744892495,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.7252,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 1.984454873601974,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.7138,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 1.864457967104818,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.7221,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.5543025284573586,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.7027,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 3.8221843572346668,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.6908,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.9603371818664512,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.7044,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.9494558087083353,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.7081,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 3.0064789512977725,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.6657,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.571814130068918,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.6348,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.046357557525771,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.6357,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.4216052305258073,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.6443,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.202892068262959,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.6179,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.246769133918462,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.6136,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.2580445808243486,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.6081,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 0.9910900744819711,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.5864,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 0.8008105276417916,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.5977,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.8758803062111592,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.5808,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.7326263369016149,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.5873,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 0.6745363629257599,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.5587,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.7963592897271838,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.5663,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.7669351275978341,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.5675,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.49034120442015316,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.5333,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.7088463794361682,
+      "learning_rate": 1e-05,
+      "loss": 0.5455,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.7177365020369466,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.5448,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.6219376275793546,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.5343,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.544914923792428,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.5469,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.6013692829294335,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.5255,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.5360825826301626,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.5383,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.5899561775772828,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.5354,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.5433869302758779,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.5271,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.46920341748940647,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.534,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.6230756025882064,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.5161,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.5411921555326302,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.5148,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.46754864550906095,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.4985,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.5473102186350659,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.4984,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.47860376318582176,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.5211,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.5426824449396559,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.5123,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.48462493929264283,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.5224,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.5159269081325798,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.5113,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.5431114554532477,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.5021,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.40812607898141146,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.4937,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.4837437041441507,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.5159,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.4831853412715895,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.4888,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.44510451792486916,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.4974,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.5139641446812477,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.4821,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.4063072305357434,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.5031,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.43382373585551803,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.4856,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.4218348357209089,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.4909,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.42657851670276226,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.5159,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.6231661197167063,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.4866,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.6551661716089583,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.491,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.5774740899966747,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.47,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.5587438244256441,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.4779,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.5358218758014036,
+      "learning_rate": 2e-05,
+      "loss": 0.5026,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.5409266738990139,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.4838,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.6958502026130866,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.4831,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.842846664968342,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.4931,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.8824700158603827,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.4795,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.8820635854691208,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.4648,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.6389441326504619,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.4998,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.63254785326373,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.4909,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.932019720443268,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.4881,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.8692493573857664,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.4884,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.998367169587847,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.48,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 1.1108737727114586,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.4821,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.7633175033732822,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.4832,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.7677866728643374,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.4945,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 1.2592809758353982,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.4698,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.7950763436682357,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.4712,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.7776215351736183,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.4679,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 1.2538967522114544,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.4748,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.6587976074277176,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.4769,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 1.008027269328153,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.4687,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 1.2967315483573514,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.459,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.7420860387536625,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.4737,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 1.3897310689985578,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.4653,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.7451102546457747,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.4681,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 1.3905437764460429,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.4654,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.7503681493858384,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.4691,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 1.2011999998427274,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.4656,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.7752810573495393,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.4675,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 1.0820619005390244,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.4807,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.8046947567346853,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.4729,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 1.1615038229098937,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.4667,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.7274944113282349,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.4648,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.911586738855292,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.4889,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.8346777016171704,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.4758,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 1.2502890632689463,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.4736,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.7397511379505902,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.4547,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.7876441025799468,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.4494,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.8829689394071494,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.4564,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 0.7450416684002913,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.4708,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.9808820020868677,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.4753,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 1.307118640403883,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.4642,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.9220960040922769,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.4676,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.9977040273438086,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.4658,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 1.0738883978327918,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.4662,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.100190470533654,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.4708,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 1.429847980816082,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.4647,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 1.0016237445607659,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.4612,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 1.4371447013220455,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.468,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 1.3282647242627876,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.4556,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 1.1035779928036245,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.4433,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 1.1221389424185353,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.464,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.2591407264274406,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.4629,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 1.0465678336655513,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.4675,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 1.4415625995822492,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.4664,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 0.8982787373912933,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.4507,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 1.5072022188226764,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.452,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 1.0868986940916785,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.4541,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 1.1820292340508223,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.4451,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 1.0416643552281175,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.4483,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.9315998909588796,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.468,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 1.7138193210706267,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.4528,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.8713871563872041,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.4505,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 1.5088541977178818,
+      "learning_rate": 4e-05,
+      "loss": 0.4701,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 1.0048754415525112,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.456,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 1.5186018597941249,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.4653,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 1.244184005239641,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.4568,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.0021688172248575,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.4508,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 1.4871958112429875,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.4588,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.9884605133829857,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.4383,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 1.3123599202485823,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.4584,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 1.15739715455915,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.4573,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 1.200361042839525,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.4603,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.982557009450253,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.4558,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 1.4240478690904788,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.4605,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.9040581419661444,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.4572,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 1.2262490081978172,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.4389,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.7517742691140388,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.4574,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 1.0118970422582587,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.4413,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.9359277994647996,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.4539,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.8548190760529903,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.4562,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.9095202401235661,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.4524,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 1.0043996459368898,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.4658,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 1.150825567127224,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.4586,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.9101813120520105,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.4514,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.8290199074272425,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.4576,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.97797211099642,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.4446,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 1.1670715917585588,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.4453,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 1.0109893271461565,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.4482,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 1.091362982246299,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.4496,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.7384917880072617,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.4491,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.7761977091520266,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.4526,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.9031342533095368,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.4426,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 1.190247905428938,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.4617,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.8941447738000764,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.4398,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.8147487813458288,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.4327,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.6957006420974314,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.4505,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.8295770768683954,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.4367,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 1.0387061530135724,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.4399,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.8688360733959256,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.433,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.534392079897795,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.4415,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.6307924607173281,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.4375,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.7204750609249482,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.4555,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 1.0998350598143873,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.4393,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.8731393836278047,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.4452,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.6107914392193622,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.4447,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.6186218117903691,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.4317,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.7331538694909225,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.4475,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.7780022542059469,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.4426,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.578808307135567,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.4446,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.5277384491303917,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.4433,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.5658743139545046,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.4389,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.624814366912005,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.4427,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.6021788724327709,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.4452,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.6199981338665804,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.4382,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.5277208728762361,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.4511,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.6533424226839859,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.4382,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.7184279261404535,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.4389,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.8041288977673464,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.4515,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.7929137613757264,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.4453,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.6588356247772326,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.4536,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.7810650533982932,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.4501,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.7337525376215297,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.44,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.5137280687336808,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.4397,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.9065682028095097,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.4456,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.7575759925764095,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.4316,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.5803512728368538,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.4417,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.9074441967041665,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.439,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 1.009879235305068,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.4388,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.8915311835214333,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.4414,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.7321331583445425,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.4459,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.6302830869717269,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.4468,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.5771524530792954,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.4427,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.668616765976795,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.4333,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.6276196699080668,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.4429,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.6180525048307305,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.4374,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.5720049595270201,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.4379,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.512576193013642,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.4495,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.5238064285879453,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.4369,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.6107642434024007,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.4348,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.4924983086091909,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.4291,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.6505557963451652,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.4339,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.7202032677548859,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.4291,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.8071053794919038,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.44,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 1.0286996240811666,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.453,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.8259277060034556,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.4336,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 1.093049860225589,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.4301,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.7992974309759008,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.4448,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.7140849046053198,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.45,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.7310164911101095,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.4416,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.6109009045356021,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.4349,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.7506305588025658,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.439,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.8354918195106555,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.4421,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.9138960954896738,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.4268,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.9336134575523228,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.4277,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.6818449610309424,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.4376,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.9036462263262338,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.4349,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 1.0356390681374037,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.4359,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.9147707387170884,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.4288,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.8364403450039624,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.44,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.6490752152099419,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.4312,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.5663565369494368,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.4369,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.663783833602109,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.4444,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.6575228276806868,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.4331,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.4913440919889319,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.4246,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.49898682535069855,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.4437,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.6341178808037145,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.4337,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.6490325768616015,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.435,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.5266517365615568,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.4247,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.6635986081128127,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.4282,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.7096318976690807,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.4352,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.5959448812735991,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.4251,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.5377228772477493,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.4264,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.7422988720617584,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.4358,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.7592792269841849,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.4359,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.715836827711735,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.4274,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.8421655825301569,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.4265,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.581150020673114,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.4397,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.739438327485817,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.4195,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.7408148558275824,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.4317,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.6007503287772233,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.4343,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.819244009096363,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.4268,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.674723170207535,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.4228,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.5891590119739555,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.426,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.53097129807281,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.4254,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.5021655970929085,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.4291,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.47284664540101734,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.4288,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.4877013631616044,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.3938,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.5492065872167522,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.3955,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.7021764773422999,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.3946,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.7896691478493846,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.3819,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.7377427763238812,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.3879,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.5465448336287927,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.3906,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.6198554107326206,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.4019,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.813759455460567,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.4008,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.9827995129475839,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.3978,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.8777342498867204,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.4,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.5806879838006048,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.3833,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.7397665534127399,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.4031,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.937988204433178,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.3952,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.900628127028175,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.397,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.7165326736508755,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.3981,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.893513157060121,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.4016,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 1.100431420312907,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.3982,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.7170790878747889,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.391,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.9985274956437098,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.3972,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.867121886690608,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.4059,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.774704422661643,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.3936,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.6140239985283513,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.3899,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.6428768503277095,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.3986,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.6699110244740665,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.3944,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.5796575138562663,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.3832,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.6861584029641044,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.3926,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.5742119303263356,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.3911,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.4915252545302171,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.4041,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.4363964778527775,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.3913,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.48168550278638755,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.3922,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.5438940821660355,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.3927,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.3989983955851633,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.3907,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.5023800401377421,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.3914,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.588375844082356,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.3854,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.7819244857871227,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.3961,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.6890335317343781,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.3879,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.5685227397487015,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.3988,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.5884017278366755,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.3956,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.4956640235638424,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.3932,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.5504276578528046,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.4011,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.5126873400935409,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.3978,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.5240163324095931,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.387,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.5501998202782281,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.3855,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.47518705310286324,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.3809,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.5148594288900207,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.3849,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.5472852047979864,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.3874,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.6006157166647186,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.392,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.48400101944406226,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.4013,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.5210247094885213,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.3888,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.6426118936520138,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.3855,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.612793064650296,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.3841,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.6888855812130795,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.3969,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.5856376120181688,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.3927,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.5767874108525055,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.3894,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.621174046052376,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.3989,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.7132428406684461,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.3944,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.6437923986512445,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.4082,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.6670493987050836,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.3901,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.5824208581907999,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.3835,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.5261442500376323,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.3884,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.5944205060603868,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.3959,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.5564557772849685,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.3954,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.5980630281780356,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.3972,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.6916509466414591,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.3987,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.5916655095842713,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.3906,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.450516209671426,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.3892,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.46982338966867243,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.397,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.5152395618014035,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.3895,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.630190114045453,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.3887,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.6201415999903612,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.4077,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.5428045298363967,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.3821,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.5216449952168027,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.3929,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.4888792252840507,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.3917,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.4371025132306773,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.3907,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.4398745397513851,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.4,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.4764056429692493,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.3996,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.42147375884571775,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.386,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.47956478176636347,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.389,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.4131285449969575,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.3795,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.4401285452716876,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.3953,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.6700337857368063,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.391,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.7228957895021331,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.3891,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.589011513793146,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.3895,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.49705195590041035,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.3846,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.5495748638943557,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.3923,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.7048061364907288,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.3924,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.8356516492238854,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.3849,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.6150348956872572,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.3978,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.4142034260091864,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.388,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.6542767034350907,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.404,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.6385936005973254,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.3956,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.5625259302524528,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.3961,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.5351419887303357,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.3974,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.4925692836585598,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.3835,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.6391057419835987,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.392,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.6724709857531772,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.3968,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.5837962254862783,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.3932,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.505307048925815,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.3875,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.6605255491349271,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.3958,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.47503744567969586,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.3979,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.467106495137601,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.3789,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.5563934216042508,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.3795,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.5138688608377102,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.3802,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.5673016912842296,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.3852,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.5734942716303989,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.3921,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.5727175832674363,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.3921,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.5672470444330511,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.4047,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.532918589423326,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.3903,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.5459657183581108,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.3974,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.4684082206428702,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.3892,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.6250204981923342,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.409,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.6750100767955813,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.3784,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.6697962118676349,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.3933,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.6756189513463061,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.3839,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.4342672145124222,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.3947,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.686276187629147,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.386,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.5239997356114313,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.3804,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.4955363937673751,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.3912,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.46537862951506964,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.3898,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.5650663124601191,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.3909,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.5696264501394686,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.3875,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.5179909081371625,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.3948,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.6584316165443344,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.3912,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.6786068289588245,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.3932,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.4138513947654111,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.3854,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.6141617533382998,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.3819,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.38780941733083224,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.3861,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.4786112582895802,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.3866,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.6560867913650186,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.3987,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.4324452843491432,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.3896,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.5086950446003374,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.3851,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.4590029291706237,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.3854,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.5160244828120208,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.3848,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.459773268174046,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.3942,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.4484836073725009,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.3915,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.5139989734963349,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.3911,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.42270278465596817,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.3943,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.5965982815517131,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.3939,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.4391827243928895,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.3894,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.3877472630848821,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.3946,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.4275592268137969,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.3957,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.3478105887856853,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.3806,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.41226742977222286,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.3939,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.4015116167735344,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.3909,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.45272602157343106,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.3876,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.5992229735768623,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.3943,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.3878478128652179,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.3931,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.5091601095389361,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.3911,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.5401102261495089,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.4063,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.4780512114319128,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.3821,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.4891156441680742,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.3762,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.4304234824886197,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.3907,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.5732639154630448,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.389,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.5369247664033313,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.3874,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.4498666616675876,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.3844,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.4063510445490687,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.3852,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.47960453425825195,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.3867,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.5705171417381589,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.3927,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.5376764341013457,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.3859,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.5055682901343901,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.3979,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.47701226241730443,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.3712,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.4823876994135204,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.3865,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.4408428799030248,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.4059,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.39713524816571066,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.3877,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.4980068276939656,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.3773,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.4276128554317761,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.3914,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.4293289315970315,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.3821,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.4238344224738794,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.3897,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.3966243495456184,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.3973,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.54211300306432,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.3987,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.6089498321029162,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.3922,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.43025892374982677,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.3893,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.45657955929736777,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.3868,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.4841087111686456,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.3848,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.5599922332961976,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.39,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.44428634273390827,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.3884,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.34548884747627606,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.3877,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.47969008255922385,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.3897,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.386797934817711,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.3871,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.49890122465386655,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.3972,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.39579120388041295,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.3876,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.5275334214114501,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.3797,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.45059956617657937,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.3847,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.5501303355459896,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.3871,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.46735124131429073,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.3934,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.41168586251446027,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.373,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.49645840894409615,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.3915,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.3589485854438496,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.3876,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.513565326110708,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.3835,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.41750420557119294,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.3739,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.507962221906822,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.4006,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.5171215003637244,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.3841,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.4857360153263103,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.3979,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.4987420320768395,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.3902,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.3986584082008368,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.3953,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.5879552223412854,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.3725,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.45413592239131984,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.3851,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.49723113357304416,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.3881,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.48250565575594534,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.3895,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.4444090232610576,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.3814,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.4383346342003474,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.3925,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.44556969210277453,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.3967,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.5720660562400122,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.3869,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.5104184611082161,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.389,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.4663026178646765,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.3897,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.45006126620214876,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.3876,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.5383593781049091,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.3775,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.515564008810841,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.3908,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.4220773650142081,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.3833,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.5046274418317748,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.3891,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.5100600557056891,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.3867,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.4452477838616047,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.3745,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.4701884483987027,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.3838,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.5837521239490957,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.3816,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.45763704479689776,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.393,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.5135354806510457,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.3827,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.3833395075875008,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.3873,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.5094507821815292,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.3906,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.48677042601763654,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.3748,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.4059665290423617,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.3851,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.4018138744709745,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.3704,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.5171690925457062,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.3829,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.48106976634042453,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.3887,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.39754597076215487,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.3845,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.5513847069492093,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.3869,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.40931999530177776,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.3794,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.45358565746843754,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.3889,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.3778430785407747,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.3812,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.4187452084106712,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.3986,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.4892851164330255,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.3942,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.41227537685904164,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.3868,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.46860604777432613,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.3758,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.34650894580661523,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.3753,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.513060798877634,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.3906,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.3714278617447204,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.3782,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.4584800234157039,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.3933,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.40418610168011976,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.3828,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.36822944584929573,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.3885,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.5472840927253942,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.3813,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.35854342395120575,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.3774,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.392923435033652,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.387,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.3435553833558924,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.3786,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.34217957361613993,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.3894,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.3771108561239909,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.3822,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.3938143741826976,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.3904,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.4490043530993501,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.385,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.40185588106852116,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.3767,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.5303556623894291,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.3474,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.5152908934482027,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.3373,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.5400680097403973,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.3355,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.6510450081765825,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.3413,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.6259701515775314,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.3371,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.5248831501474021,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.3321,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.6164809872621799,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.3318,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.5459455155844398,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.3326,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.5648188451362511,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.3452,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.564358552764898,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.3325,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.47621214895910796,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.3375,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.7314845148569669,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.3336,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.6544534747616907,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.3397,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.568539814879701,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.3315,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.4925225641132062,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.3392,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.590794107873537,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.3465,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.5425331123514353,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.3403,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.3539327872456876,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.3341,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.4948499051615696,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.3323,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.428093823089664,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.3363,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.4702779156776278,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.3432,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.40985107701196016,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.3344,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.47744555619741785,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.3367,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.3774427909500078,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.3353,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.40476310515291647,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.3457,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.49227557692230195,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.3445,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.3633491927329453,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.3287,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.5212013980216025,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.3337,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.3737475645256372,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.3376,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.44963721289773023,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.3328,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.445999596879948,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.3334,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.34898105381862304,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.3391,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.41948030794638846,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.3347,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.4366164965149465,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.3468,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.3891317907105391,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.3366,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.417567910403333,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.3431,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.5194656199252959,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.3388,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.3671979637532448,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.3297,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.43005380071617577,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.3398,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.46489031915791434,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.3517,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.3866362209106381,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.3439,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.41753430167399413,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.3324,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.4058982049547496,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.3356,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.329766159081924,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.3381,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.40956129254068535,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.3357,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.39119591233780726,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.338,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.3523642001953874,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.3349,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.378744219081539,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.3355,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.3694555651535086,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.3375,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.3997450817977694,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.3333,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.327379426335728,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.3363,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.3866124561674292,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.3261,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.3110999676875282,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.3293,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.3912447566243538,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.3342,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.32442038211167723,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.3318,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.3562660182954932,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.3402,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.334944335177732,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.3376,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.3365916156346822,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.3425,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.3453017610021559,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.3286,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.2999085370665485,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.3343,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.34580872533695745,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.3426,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.32442134588222177,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.3292,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.34269918208803934,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.3395,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.34986788202651425,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.3482,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.36979775806372384,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.335,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.29796301876578185,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.3361,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.3805201816726468,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.3275,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.32591311286994534,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.3394,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.34722761447019085,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.3445,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.4038184920155738,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.3254,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.316594263750228,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.3397,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.4389898336854991,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.3355,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.38780918039720597,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.3372,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.3489202476556766,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.3346,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.35264470886595944,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.3298,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.385826665670159,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.3365,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.3909928620271686,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.3407,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.2983342645512899,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.3476,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.3719230907869111,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.3419,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.3417790024992912,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.337,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.34469736250865995,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.3303,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.2890282296850651,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.3387,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.2878792842979019,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.3354,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.31262992513716803,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.348,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.33296788302858044,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.3416,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.27891626472861386,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.3411,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.31835511261804017,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.3373,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.34920141291643664,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.3402,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.33380368514936226,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.3414,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.34614322079210824,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.3375,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.311298283499031,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.3415,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.3339072582563793,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.3325,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.30437859796890576,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.3366,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.36908933761995716,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.3434,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.33946184037354304,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.3291,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.42006356767936587,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.3381,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.41618512428945553,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.3409,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.38770950030351226,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.3385,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.41890666941405424,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.3378,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.425389721797591,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.3422,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.4502434838953622,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.3249,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.3966507777400277,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.3381,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.45268397687022777,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.3412,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.32852359344650783,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.3339,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.396075533954901,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.3377,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.3324061092510544,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.3344,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.37774805644368925,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.3356,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.36848992295382776,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.3255,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.3754373296031159,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.3347,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.34417661027142826,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.3318,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.4084406130515676,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.3336,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.31069575784885584,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.3311,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.32532538669314964,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.3454,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.3627798987530789,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.3424,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.3097166829159115,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.3409,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.3233737021580825,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.3363,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.32329134330241216,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.3389,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.34276831755405335,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.3343,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.31757519716645266,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.3359,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.3429891949156796,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.3294,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.3190267386438318,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.3368,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.3822988750235917,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.3451,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.33744652019788535,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.3437,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.31932377620300784,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.3416,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.2891638377699828,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.3354,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.30379977230487704,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.332,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.3486390467973211,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.3446,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.28727563523389504,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.3328,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.3030584839848433,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.3324,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.31310648711672523,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.341,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.2847004682858129,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.3301,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.29030128239023245,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.3381,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.3257329473670702,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.338,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.30313771654288924,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.3415,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.2984276419740891,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.3387,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.337181929197356,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.3276,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.3465676001471749,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.3418,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.29236085122531585,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.3439,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.30474691561112294,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.3403,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.26408547330551735,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.3325,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.29493818812198863,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.331,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.283584247376802,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.3393,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.3395465409306191,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.3466,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.3209818055748033,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.3165,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.3222605959537157,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.3387,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.33977841536427866,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.3451,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.3431841405504492,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.3499,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.33601480472369927,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.3335,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.3524454129375857,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.3436,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.31262173754946576,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.3394,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.38149558075874485,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.3365,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.31447473595762265,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.3542,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.37697861220625,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.3513,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.34298562870195204,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.337,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.29547283175619743,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.3346,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.3294613645502698,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.3354,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.25504809385681526,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.332,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.37116105402276955,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.3342,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.2733425480033472,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.3297,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.3928423008928823,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.3479,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.30205691994823985,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.3457,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.37933062355208663,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.337,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.3015598875470479,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.3325,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.34418125908501196,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.3405,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.2867522044020269,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.3347,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.3835127712680058,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.3505,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.3130998768375347,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.3373,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.36851413740706107,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.3379,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.28341272178508775,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.3252,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.37922199949636426,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.3494,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.29216900113404454,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.3407,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.36790775026266703,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.3342,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.3181377946027446,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.3487,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.3841274180055995,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.3401,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.32242044700793077,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.3399,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.3509328315526967,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.3365,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.3301393754440888,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.3343,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.4438399857281464,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.3377,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.2900807072327278,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.3437,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.44178603229122526,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.3385,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.3168235701984305,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.349,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.40851088550856324,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.3427,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.27581117528873433,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.3198,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.4066667771725973,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.3522,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.31811744317762064,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.3397,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.31450547465920925,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.329,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.3006415809742694,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.3456,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.28438397791870246,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.341,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.28055002914477456,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.3426,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.271628605206101,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.3357,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.2928901111249546,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.3306,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.3311006709167612,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.3523,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.28131076944792754,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.3421,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.34120858708381363,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.342,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.2860720922037707,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.3295,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.3265093393036698,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.3392,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.25298653097477863,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.3314,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.3277932909507565,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.3389,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.2973659983693103,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.335,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.3300690744773407,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.3305,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.2702477383294841,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.3447,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.33818788120870374,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.3364,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.30351584468247206,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.339,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.31166141963556737,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.3448,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.31632344691206926,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.3304,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.3150887567331993,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.3366,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.3228909138724453,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.332,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.2792264036090365,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.3261,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.3114480491383736,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.3308,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.2972495349736527,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.3386,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.28498100752822664,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.3478,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.2859600627338636,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.3267,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.26401396366239077,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.3378,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.27792076827991774,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.3388,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.31213234104885895,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.3377,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.29375272956341036,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.3419,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.3056578486713288,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.351,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.2712075594678839,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.3367,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.2794235428982334,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.3305,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.2724487635155296,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.3411,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.25581395359935394,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.3413,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.2639486302108489,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.3409,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.26846749933517616,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.3339,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.28939734804483136,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.3354,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.26407128753459425,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.3295,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.29341565976130307,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.3456,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.2602538738498735,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.3345,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.2920025242828723,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.337,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.30682416691084075,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.3371,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.32337997815938174,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.3293,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.29889930275382975,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.3297,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.3382542428623257,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.3369,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.3183741843331982,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.3388,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.3306130509702418,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.3309,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.31501596372453833,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.3525,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.34967373970813626,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.3484,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.3141892383697478,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.3435,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.3094273935851772,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.3454,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.29550825004652814,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.338,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.2912483494378447,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.336,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.3179102819600449,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.3383,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.28418213425344385,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.3254,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.31449582695301814,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.3408,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.2867581158048848,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.3309,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.2932861430107474,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.3378,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.26361680744634186,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.3319,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.2744164862499861,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.3337,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.36611559861950954,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2858,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.28551888552366905,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.2774,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.4128038120495307,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.2897,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.4267339662020001,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.2875,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.3583933338545516,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.3021,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.3985185186795101,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.2875,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.3923034084242149,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.2916,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.3419566658030182,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.2972,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.38427692353523474,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.2871,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.34532028871006765,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.2912,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.34340169450030744,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.2781,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.33382968803936275,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.2877,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.31268110354956286,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.2931,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.3281757921402249,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.285,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.30519498297658476,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.2887,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.2960938163467874,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2775,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.32025731318611383,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.2887,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.2905673213376449,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.2956,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.27802125782382225,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.2956,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.31150181754391826,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.3014,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.2712893534950197,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2805,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.29270608740117987,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.293,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.27711961026833337,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.2934,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.2938657725569671,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.2824,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.3025864972450615,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.2916,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.28173022433780515,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.2836,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.28528758661645803,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.283,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.24046449301036865,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2855,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.32605254344476003,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.2883,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.22353694632083174,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.288,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.31998849032857457,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.2958,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.29452208666792745,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.2926,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.31556831536783214,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.2865,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.30037398759445705,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.2902,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.28782620095849915,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.2906,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.2850766453438943,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.288,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.2975453395814113,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.2909,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.27685520641026806,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.2982,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.2682252241195558,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.2832,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.28431934787976404,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2945,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.23543515118620537,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.2853,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.250316695099336,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.289,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.21326719731316784,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.2834,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.2518843231353911,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.2793,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.2290346989062734,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.2861,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.2620375627056031,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.2914,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.2453126409162077,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.29,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.23658372484993373,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.2865,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.2385189842399488,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.2949,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.2253525563042739,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.2819,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.2546184061868644,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.2859,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.22251535637938208,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.2903,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.25088618267049106,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2922,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.22854099118874363,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.2889,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.23673414915080654,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.2934,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.2362620982889696,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.2873,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.22861892620354393,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.2871,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.22629966697371134,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.289,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.23332627585310467,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.2879,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.23826449505686348,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2857,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.22641371714089417,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.2871,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.21195954441654916,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.2824,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.2542658665610915,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.2817,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.22286428849306833,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.289,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.2160008500904822,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.2787,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.25605938497887754,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.2886,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.2198528872107236,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2854,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.22841571916537343,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.2908,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.2356013474253358,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.2936,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.2247872775332124,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.296,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.21499718775700083,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2886,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.22749417245002915,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.284,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.22215037051762593,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2934,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.21193787905859762,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.2782,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.2361526734048553,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.2865,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.22070048503287917,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.2904,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.22299405805148428,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2904,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.22445360043996798,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.2972,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.23393064570346908,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.2908,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.22579090468014673,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2907,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.2159208023172492,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2827,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.25598115256796844,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.2942,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.21107154266783174,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.2928,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.2375110402443226,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.2885,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.26063902818704365,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.3006,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.2336913292584425,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.2927,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.251247290925097,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2875,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.23074532508285506,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.2863,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.2305241005851657,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.2875,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.2174810057589093,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.2836,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.21478756790982073,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.289,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.20121113131763738,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.2828,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.22377470909227842,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2886,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.22207540103193626,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.2891,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.21642056588628392,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.2871,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.21569365366875928,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.2937,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.21801131135971286,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.2916,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.21572070231764973,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.2819,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.22349813323790904,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.2916,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.25523628472039,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.2888,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.23637804475744828,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.2953,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.21920840364465627,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.2943,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.3069064043854183,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.3045,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.2169183417214614,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.2951,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.25050110353038724,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.2963,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.24568590868725798,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.2917,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.21309083828384998,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.3005,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.23102772160447635,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.2903,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.2379449845446301,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.2923,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.2094897163837021,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.2901,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.23757569019930275,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.2987,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.21156711964211092,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.29,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.21953341992392464,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2855,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.203568721905517,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.2763,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.20965506048716256,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.2899,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.22104342054552298,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.2902,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.22427569124673444,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.2875,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.20755020902841234,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.3033,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.2148824139281095,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.2922,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.22376323187085428,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.2942,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.20654028350386738,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2923,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.2015002352630099,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.2895,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.233377303547359,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.2902,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.19495019734196287,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2815,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.21517114700154194,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.2916,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.20693539589796486,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2831,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.20743209601572649,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.2938,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.19995442775117045,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.2883,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.19576351252472082,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.3002,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.21236967004058552,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.2947,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.22308819799921176,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.296,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.20384889015606744,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.3002,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.2150298424696523,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.2857,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.2128796677072654,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2896,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.19834155707986573,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.2833,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.1988476724977661,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.2765,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.2190951128805634,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2885,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.2195905504068831,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.2838,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.2020040265665253,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.2865,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.22372786515728485,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.2776,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.2238491575088446,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.2951,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.20207950852981366,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.2897,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.21372864426040872,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.2921,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.20706999502023396,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.2938,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.2053644193914046,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.2935,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.19101782131271236,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.2753,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.1929538705588472,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.2913,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.2018635127186374,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.2829,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.1989929107900467,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.2826,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.21368883438199476,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.2972,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.18489995916862406,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.2883,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.1945701359435098,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.2907,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.19801203605300258,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2855,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.18862877900497133,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.2831,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.1855847697460044,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.2879,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.20935493157179622,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.2973,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.20021804928364792,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.289,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.1865235263754341,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.2821,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.20291839656089727,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.2863,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.2233472607419647,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.2958,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.20100117022093075,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.2923,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.20428456583663274,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.2807,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.2037657837548517,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.2952,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.1954880310531426,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.2864,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.19836892049152177,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2964,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.19959890969724553,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.284,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.19751531591596747,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2946,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.18797308425000764,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.2874,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.20232514249530983,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.2783,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.21082586394360892,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.2821,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.20404086739900693,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2832,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.19983202128285615,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2806,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.22842860948810392,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.278,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.20377480940997325,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.2893,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.19766199007990715,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.2814,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.2141083227955186,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.2838,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.2033061953111166,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.2868,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.2054468407780253,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.2854,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.20404073061013303,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2858,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.2037392777505931,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.2965,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.20972312851861222,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.2908,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.20130012063515468,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.292,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.1894042808731602,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2966,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.19963244464007876,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.2857,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.20532322067048556,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.2893,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.20563306948902546,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.2896,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.20503885308433983,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.2909,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.20080691705194062,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.2932,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.19335341270486275,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.2871,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.1946153963811525,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.2841,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.2112667564809226,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.2849,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.19978988006121748,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.2988,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.1933162227787063,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.2895,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.2101974346592077,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.2927,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.2007598625344376,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.2958,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.1945574465294295,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.2904,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.20531263791625004,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.2834,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.20143060250493142,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.2907,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.2028827397105636,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.2898,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.20579717250899035,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.2951,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.19171615945510195,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.2872,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.18960028347782,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.2899,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.19754723899070584,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.2815,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.2002324694896748,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.2928,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.18887350347475387,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.2863,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.2110041916215512,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2893,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.20933701990578954,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.2885,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.1941415779942822,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.283,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.19327296173966593,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.286,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.20162051824866947,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.2879,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.19681288590425253,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.2967,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.19005738153402227,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.2891,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.19434334172746484,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.2884,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.19794255590260404,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2892,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.18860502736672596,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.2891,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.1997941804266184,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.2938,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.19585367915679674,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.2864,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.2065944602329389,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.289,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.205480085431489,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.2806,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.19509227085329955,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.2823,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.20464449781360167,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.2952,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.1916723598876134,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.2899,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.1917969696249266,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.2772,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.19823648511597597,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.2839,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.2221841085250693,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.2904,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.19358840559797336,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.2814,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.17936856044803773,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.2908,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.19776180616107236,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.2915,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.19663176294669807,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.2817,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.18970396857223798,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.2805,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.17679597739482955,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.2894,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.18116659043634462,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2837,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.17852820799366803,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.2829,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.19123524895552094,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.2894,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.17506900093775998,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.2887,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.18898156933910368,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.2861,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.18642781369976413,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2829,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.18524448737891677,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2886,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.19752402435262048,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.2849,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.17740970653608032,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.2808,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.17574668057360443,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2852,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.18962849788687305,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.2835,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.18395530283281028,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.2906,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.17774240839355845,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2805,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.19727616449903643,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.2921,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.192490067457276,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.2877,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.18030448796047785,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.2813,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.30395861963546555,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.261,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.2632906294998707,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.252,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.1973937615637014,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.2537,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.23722831737244515,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.2575,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.29687935658051107,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.2488,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.2628096243801315,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.255,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.23765493165598262,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.2581,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.23506089564031213,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.2517,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.23733855157372657,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.2416,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.21985007354475822,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.2508,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.21500452627538316,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.2424,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.22144300048659302,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.251,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.21539608722761772,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.2504,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.21155443924967424,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.2569,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.23300747743280842,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.2562,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.22524180578473893,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.2549,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.20947262265328345,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.2464,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.1986847402880331,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.2488,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.23351019177985446,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.2524,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.21097095264703827,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.2521,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.19134238743660142,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.2493,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.20682489628770412,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.2622,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.20665303323326065,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.2552,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.20062856486190064,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.2526,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.1957078899920919,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.2564,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.17969245015796093,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.2511,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.19049504037572443,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.2501,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.2011822831571168,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.2594,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.19612258063820362,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.2573,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.18501670582649407,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.2523,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.17650269566731178,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.2599,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.19989874870525423,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.2594,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.19120331628168352,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.2521,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.18610579578137101,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.2471,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.1902570865248062,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.2571,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.18013970867144083,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.2493,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.17895939579860975,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.2508,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.18675646524642087,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.2564,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.18443174668822465,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.2538,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.17758384581489703,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.262,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.17886518218668596,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.2512,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.19227895279622798,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.2553,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.18413568632778965,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.2525,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.17485817546535035,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.2567,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.1715277297820428,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.2519,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.17968053805454132,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.2619,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.18115370364974845,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.2523,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.1823311155724449,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.2522,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.17058313798866237,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.2489,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.17607221603217527,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.2546,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.3505562400905427,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.2558,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.17701536461808398,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.2486,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.16754863203412132,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.2612,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.1734768878079203,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.2483,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.17630564066136478,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.2495,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.1677678707283956,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.2489,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.17530475567905465,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.2555,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.17477534672190692,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.2524,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.17372670918716807,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.259,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.1751714784220081,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.2532,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.16536509009609063,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.2488,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.1705168085771883,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.2537,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.17336701021311335,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.2518,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.17390412096401703,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.2518,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.1672694704337311,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.2466,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.16780050723139942,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.251,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.18376618534492664,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.2574,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.17068138825235377,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.2551,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.17037935216115294,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.2515,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.16586090486088054,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.2502,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.17003753959686252,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.2541,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.17122709193517102,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.2514,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.16962334646233448,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.2598,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.1773356822403112,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.2543,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.1726539030511917,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.2496,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.17583491534494927,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.2495,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.16593005665235172,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.2445,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.17135353749555204,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.2482,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.17432344114574977,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.2553,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.17311159412987276,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.2548,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.1714144667542426,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.2516,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.17013880478107168,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.2473,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.1636854621883899,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.2469,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.17550593840572123,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.2504,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.16817104046648543,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.2455,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.1744055033198684,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.2498,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.16945592765449996,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.2485,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.16670226604297195,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.252,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.17054323986037329,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.254,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.17072467077261533,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.2511,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.16428597808313494,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.2515,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.16933095639742954,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.2485,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.17263055977065145,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.2562,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.1711348477495078,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.2531,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.16786294937424526,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.2497,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.1731033716232654,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.2581,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.18064280391750978,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.258,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.17254301004485317,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.2569,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.1679695516353815,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.249,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.1638503317454199,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.2431,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.17272361961141894,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.2525,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.17272302121666103,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.2523,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.16525677878581085,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.2486,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.16530842640422697,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.2523,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.17012488416623908,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.2529,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.16729141065170847,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.2513,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.16621962902038054,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.2557,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.17458978403159248,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.2546,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.173134148427128,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.2541,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.18684731404292365,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.2421,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.16610251933377038,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.2584,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.1751611486699881,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.2583,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.16265337834193364,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.253,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.16690686149144687,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.2482,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.17303966902793227,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.2528,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.16835307422532958,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.2596,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.16832572301238694,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.247,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.16596578948003624,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.2514,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.16615998130507054,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.2549,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.16732950203648686,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.2481,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.16366302707221714,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.2494,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.16553326661251444,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.2494,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.23175040111570472,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.2537,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.16900453504356966,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.2493,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.1633076516993753,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.2464,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.17745730220997424,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.2511,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.17237240630655043,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.259,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.16945522454324072,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.2551,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.1650332305919656,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.2561,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.16757447306801107,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.2507,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.1668614909814492,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.2551,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.16614860012451466,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.246,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.16883006629965489,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.2481,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.1657239826298354,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.2577,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.16218846333025078,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.2587,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.1700015422013648,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.2585,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.16577604122893094,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.2548,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.16222729144664755,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.2482,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.16436124911567987,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.2513,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.16866304378493732,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.2521,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.16726245162642536,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.2466,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.16080986773884495,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.2519,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.16861214931050378,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.2647,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.16205831892890385,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.2498,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.1622406950379493,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.25,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.169301154462786,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.2532,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.16338362820224187,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.2532,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.16365607588037548,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.2498,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.17246935528981344,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.2542,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.17152621929958417,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.258,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.16988260654339232,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.2575,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.1599842930581676,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.2429,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.164716156094345,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.2598,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.16567320521525128,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.2551,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.16407189184995222,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.2523,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.1683618069212438,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.2596,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.16478785785178365,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.2556,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.17024102762661128,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.2571,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.16604897821727974,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.2578,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.16390829592345169,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.2531,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.16214106365965375,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.254,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.16733668678309704,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.2513,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.1657200827517337,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.2557,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.1665277371845228,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.2545,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.1715068312249911,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.2519,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.16183307757828783,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.2486,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.16571765756316176,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.2565,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.1616924370612381,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.2539,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.16346300831519917,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.2531,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.16050307516682483,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.2428,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.16381482192469446,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.2621,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.1659759757649514,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.2548,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.17086912501313833,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.2641,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.16060529515589,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.2486,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.1669916605159866,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.2515,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.16366159337409325,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.2434,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.16607355646149166,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.2558,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.16487293599318512,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.2572,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.16210286055958203,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.2545,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.16203963622805428,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.2507,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.16186470610545783,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.2525,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.1628087844721581,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.2497,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.1654369680950424,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.2484,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.16333466756121845,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.2556,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.1681292237228783,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.2591,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.16165720508085596,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.2526,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.16409625737685132,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.2489,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.166822830120079,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.2516,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.16785606035415598,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.2536,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.16651452904187303,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.2524,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.16100775204744083,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.2479,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.1612398260780192,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.2563,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.16429736301985023,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.2497,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.16171379121051424,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.2506,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.16168482150449226,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.2528,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.16239323973569572,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.2538,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.16532371154118808,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.2543,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.1654949863073274,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.2506,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.16655812402281372,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.2594,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.17008987808767714,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.2457,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.16591881055921284,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.259,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.16281014380768333,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.2524,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.16380551181471362,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.2549,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.1609775197112163,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.2533,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.16549172507235782,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.2585,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.16622061504793545,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.2581,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.1583028874546832,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.2566,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.16041191136650362,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.2519,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.16467198893516163,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.2574,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.16111472766304075,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.2543,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.16440175657458958,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.2518,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.16616404818305885,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.2605,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.15738639762708603,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.2506,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.16648803351663727,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.253,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.15890588152617258,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.2523,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.16142554298255152,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.2548,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.16410594107432946,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.2528,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.15757656628477978,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.2465,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.15654053863685488,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.2472,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.15929103829139393,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.2518,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.1671161743950628,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.2494,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.16424213720243758,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.2466,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.1626274502217075,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.2542,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.1540348323853417,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.2435,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.1573919781613214,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.2444,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.16684048137177454,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.2561,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.16371631062011685,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.2571,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.15913982694453477,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.2562,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.16429522670089503,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.2532,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.16355351192949433,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.2541,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.15939217706382836,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.2478,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.16680382900496402,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.249,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.16047846806184704,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.2507,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.16525665490048955,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.2504,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.16434075586344696,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.2549,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.15849121965553586,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.2541,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.1632532854901493,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.2574,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.16872763357672133,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.2471,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.15804790113758763,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.2481,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.1626067409622289,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.2565,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.16618929660455709,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.2546,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.1620863283514671,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.2528,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.15875657673335292,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.2522,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.16261590474089135,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.2597,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.16105463800665362,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.2425,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.15911884113943414,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.252,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.15851645711529483,
+      "learning_rate": 0.0,
+      "loss": 0.2457,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 5.479593102698086e+18,
+      "train_loss": 0.3508631761257465,
+      "train_runtime": 58784.5389,
+      "train_samples_per_second": 2.688,
+      "train_steps_per_second": 0.021
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1235,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.479593102698086e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}