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": 6.720313632683381,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.8372,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.865866547910736,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.8435,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.924945470437719,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.8501,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.816484438567485,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.8602,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 6.324635659798169,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.8217,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 6.208074722284902,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.832,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.789018982243938,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.7835,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.55782759186916,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.801,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.5270760096821094,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.7259,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.330860293568401,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.7473,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.144013983908243,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.7286,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.9470134479750167,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.7292,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.088678501071537,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.7104,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 4.445688263853117,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.7329,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 4.293965157496694,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.7151,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 3.324496910045219,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.6677,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.9548813432509218,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.6425,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.325541902349651,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.6394,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.7509010777851737,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.6405,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.5559913741550915,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.6217,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.6277280002475063,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.6007,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.6660935301005269,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.5885,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.4377977218966231,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.5834,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 0.9924049703748143,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.576,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 1.3968835767862378,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.5573,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.8754834368457042,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.5702,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 0.694643078890811,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.5529,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.7062936704654795,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.5451,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.76930291728848,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.5579,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.7638096895574201,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.5575,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.6329333395830534,
+      "learning_rate": 1e-05,
+      "loss": 0.527,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.8398488181958437,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.5353,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.649326788796873,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.5439,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.609075147447315,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.5348,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.7760609298195079,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.5329,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.5085039526045684,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.5122,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.7717055471223562,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.5195,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.6920269049648219,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.5172,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.5011177231884446,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.5158,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.7287249479099216,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.5105,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.7196541506560883,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.5126,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.5424848263877725,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.4996,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.6400572767966559,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.5027,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.5241245785095697,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.4921,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.5804366194180663,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.4996,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.4729507016299286,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.4934,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.5099628584881852,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.4963,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.4866470114817898,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.4871,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.441320883026727,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.4739,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.5546047083130496,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.4909,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.5624470171357147,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.4927,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.5512888906457746,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.4774,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.5312942534957518,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.4855,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 1.169333750786725,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.5003,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.5605211391547605,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.4804,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.5100855892120887,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.4785,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.5365644729343734,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.4844,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.5148443698652201,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.4792,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.5225274349736068,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.4999,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.5183297043917037,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.4706,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.5703673172407892,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.4626,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.5943486463325229,
+      "learning_rate": 2e-05,
+      "loss": 0.4718,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.7316301344488954,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.4829,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.7000572062843409,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.4721,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.6434979360251097,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.4793,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.6039082958826385,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.4627,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.5515556656943064,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.4718,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.5476712937070156,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.4623,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.49474486034029774,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.4529,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.8285504430377411,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.477,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.7676506881266095,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.4706,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.6815968958303046,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.472,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.5988596737289354,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.4619,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.5737786363163616,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.4676,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.6013479423524069,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.4537,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.6920960403199063,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.4619,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.8521571266106306,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.4566,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.6293385278791191,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.4586,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.8292955999307067,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.4513,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.9096671145260958,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.4621,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.7163467592024004,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.4621,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.5793296198672107,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.4461,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.772233054210313,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.4651,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.8536147690265288,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.4607,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.7362208449603332,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.4528,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.6227094127897949,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.4605,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.7279285351280159,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.4307,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.7224935669383586,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.4561,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.6136121265429374,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.462,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.624538946495043,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.435,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 0.6190882871668518,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.4719,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.7536810532489604,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.452,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.740949031196268,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.4623,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.7560158630332877,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.457,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.8724799602979018,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.4548,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.6445522040875817,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.4593,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.5815584442710945,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.4422,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.8270239949650832,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.4493,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.9299635779323513,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.4516,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 0.728780357456473,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.4414,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.6903478695741618,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.4445,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.632752291991796,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.4519,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.626358341002859,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.45,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.7453570322793344,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.4548,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.8641706107583838,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.4665,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.2032758698762966,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.4585,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 1.0421808246568929,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.4664,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 1.3177412473204864,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.4611,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.6301484393726319,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.4487,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 1.4073288321010515,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.4626,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.6626453952288496,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.4391,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.855815098383975,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.4397,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 0.603998955468158,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.4421,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.9016612024188702,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.4499,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 0.8802130824205491,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.4606,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 0.851926145537898,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.4555,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 1.0276907568399918,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.4453,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.7175712962774428,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.4468,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.7978521694294324,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.4457,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.6050655643181089,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.4437,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.6489202968958862,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.4442,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.6732524133388296,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.4308,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.6838006386598261,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.4428,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.6517362772633797,
+      "learning_rate": 4e-05,
+      "loss": 0.4463,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.7451912993695795,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.4499,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 0.5934986561093645,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.4486,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.4905099927155084,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.4233,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 0.6103181646722485,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.4365,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 0.7297155510904937,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.4439,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 0.5837618878451047,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.4331,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 0.6178869628645598,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.4472,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.5814109615369392,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.4449,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 0.6446588028863152,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.4581,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.5665025250598587,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.4302,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.7185313723572266,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.4395,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.7988882095798309,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.4347,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.5313474440117925,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.4461,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.6433741992471147,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.4386,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 1.0950839765666855,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.4709,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.5968618785208732,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.4512,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.5086629787885952,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.4458,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.6047268292285155,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.4525,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.45431142421967374,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.4454,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.41182572329300804,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.4395,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.4890294749921196,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.4304,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 1.660899143790332,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.4421,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.5727605173875056,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.4413,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.5973473778714804,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.4328,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.6726706679942572,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.449,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.7318126253203148,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.4373,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.6949946220840727,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.4478,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.7992645827545217,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.4439,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.6934015107349057,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.4368,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.5439081727505936,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.435,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.8059871987450756,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.433,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 1.6092569964355177,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.4611,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.9240283432627293,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.449,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 1.3069763274089556,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.4341,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.633616918375799,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.4292,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 1.1716402217369528,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.435,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.6676633696202126,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.4157,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.7556250288006953,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.4364,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.8432261161914822,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.4454,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.514184993691894,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.433,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.6475245910035559,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.4311,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.5859033087923647,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.4223,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.5308333468033517,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.4139,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.6274928110125843,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.4337,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.6450996463998665,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.4458,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.6644196754991113,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.4318,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.6028855534258594,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.4097,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.5940335924458224,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.4266,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.6726844293966269,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.4327,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.48683483739239036,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.4148,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.6799744007714286,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.4194,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.5706083455222054,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.4245,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.5285077438766009,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.4169,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.4360202173131749,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.4225,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.45648517757042323,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.4298,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.5010202844670069,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.4279,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.411784120947612,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.4344,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.4669167986294635,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.4401,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.3818670615508354,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.4209,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 3.0006544933346335,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.4877,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.6557531688700068,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.4216,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.6337779585256715,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.4332,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.45326269596568936,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.4298,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.5235799209996737,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.4293,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.4235281594093527,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.4183,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.5792487357047983,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.4118,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.3777054556174785,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.4101,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.6058729326544898,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.4224,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.5098705948856941,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.4229,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.5399028090291087,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.435,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.5346252487499162,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.4395,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.5240183705547296,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.4377,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.5919438212144604,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.4237,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.4312500074187652,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.4226,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 1.924322108268979,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.4453,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.5469133865667605,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.4158,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.5486262129760139,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.4252,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.58101319343812,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.4162,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.5147497044571169,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.4221,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.4978986436820908,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.4327,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.5322747603235845,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.4346,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.42283560135898457,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.4207,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.544749670696631,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.432,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.5913785308621513,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.4216,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.6036232338627572,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.4246,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.4242728299446276,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.4157,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.5971967053598449,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.4262,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.41468847805417197,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.4242,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.5079037833186744,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.4249,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.43627216107140243,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.4275,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.4612333921092053,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.4267,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.40006925084113487,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.4265,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.4841022120235549,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.4147,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.5266663051531908,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.4211,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.3603020506105271,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.421,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.4888870734749114,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.424,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.5093866403248749,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.4388,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.4574826432714574,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.4235,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.43091484257726137,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.4221,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.4475632475962844,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.437,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.4259474155397758,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.4191,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.47366101846995246,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.4208,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.40351985349229463,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.432,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.4539910497128911,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.4222,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.502931819148987,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.4141,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.6760326468295894,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.4284,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.39887404530072285,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.4231,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.47974538139128586,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.414,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.47853985477664057,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.4245,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.40233740996678785,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.4249,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.522311923490341,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.4085,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.5271891422779577,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.4257,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.3989082785804531,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.4002,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.5355977686100452,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.4288,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.5098510138404377,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.4121,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.42615103781931496,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.4295,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.4804423228852851,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.4111,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.47296184158269255,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.4138,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.3589724326106711,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.4133,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.4538007564834648,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.3941,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 1.7448738040607357,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.4261,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.5225759247156554,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.4336,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5142967865206043,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.4212,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.44909921352654936,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.3869,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.6266375716049284,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.3926,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.6431598135693586,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.3666,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.5349799981780707,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.3901,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.6704777007006191,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.3742,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.5338310655747841,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.3676,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 2.7664478522204163,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.3969,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.6855775738315114,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.3941,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.7247949070635261,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.3752,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 2.462251822225763,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.3947,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.984524371792189,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.39,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 1.4421667294842464,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.38,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.903498781453985,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.3984,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 1.6063575662650715,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.39,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 1.199123642548432,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.3779,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 1.056542709304203,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.3627,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.9512519471770691,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.3741,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.9553638060608256,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.371,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.6338297061816031,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.3766,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.8921294741224384,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.379,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.6380784083693715,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.376,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.6931383741774363,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.3713,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.620119538817013,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.3758,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.5049243728647953,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.3643,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.6004446233776141,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.3821,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.4963151453843717,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.3796,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.48151923944633096,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.364,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.46109802119323323,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.3638,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.44029134868502723,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.3693,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.505426480382134,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.3754,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.46512940832526,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.3854,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.4943683271872026,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.3704,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 4.209405286497348,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.5032,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.7361035611386454,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.3967,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.5330282485860742,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.3703,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.5738832517225787,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.3733,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.5835002749857904,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.373,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.5510877938422433,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.3853,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.5007139518930799,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.3701,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.5165372233432705,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.3749,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.46647309201174947,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.3819,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.450423257642982,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.3722,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.5729731872330088,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.3781,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.4194773683176032,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.3809,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.48829144537453373,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.3744,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.4102082569261092,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.3674,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.48528479226794086,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.3938,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.40119066991190666,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.379,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.505115185243081,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.3854,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.44588755881827913,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.3734,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.42823318763486456,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.3723,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.5040334882449713,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.3901,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.3887966348958811,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.3654,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.3899481866609106,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.3867,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.450623879937821,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.3709,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.4018173403265111,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.3739,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.3877738968646148,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.3824,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.46745989460248677,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.3833,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.35077105425787797,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.3759,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.6849730411879219,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.3862,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 2.008336981606847,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.4234,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.4874113872004471,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.3778,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.48271527291422583,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.3709,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.38641880488392033,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.3664,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.4450153039491475,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.3747,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.4858233052527487,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.3728,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.3680180231999003,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.362,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.48650296827609385,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.3941,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.36652448748775035,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.3857,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.3903530744447606,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.3811,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.3648006271024427,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.3553,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.35978599909185804,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.37,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.39501375751864254,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.3698,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.39374906986154906,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.3852,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.399965649484609,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.355,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.4098852942089664,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.3804,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.44817222061417167,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.3715,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.3524990556775313,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.3784,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.40874878895699795,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.3772,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.34028363972540954,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.3811,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.44171724295431086,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.362,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.3427489755282385,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.3681,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.43465543349984104,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.3656,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.36772634436217433,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.3759,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.4573778533276756,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.3719,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.39839497584069533,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.3722,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.36753988054190606,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.3671,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.3703819331227409,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.3837,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.3184538796027115,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.3798,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.35184848991714557,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.3646,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.4417848740628385,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.395,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.3372154701961584,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.3539,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.41153620949016234,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.3809,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.35025472056775925,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.3681,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.36676876641049566,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.3801,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.3889536110044916,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.381,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.40072874658358354,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.3592,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.3502580971797869,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.3687,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.4113774605762822,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.3716,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.3120709008405903,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.3828,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.3840627910467601,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.3685,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.37931619848966613,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.3762,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.35681516743936975,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.376,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 1.595903395895561,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.4131,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.5125091789234335,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.3699,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.4838440493001961,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.3648,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 3.546257668923366,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.4848,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.5038307954009387,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.3679,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.4368756036134776,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.3894,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.4141244922600816,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.3724,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.36409559292399307,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.3892,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.41560270266686433,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.3835,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.361916488245091,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.3763,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.37530152673767647,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.3587,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.35181576070753035,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.375,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.41316728914812423,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.382,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.3226916059325309,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.375,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.3823664565999465,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.3743,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.31717443044561805,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.3671,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.3622672494277445,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.3784,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.3516941586890796,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.3621,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.34501444656635993,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.3662,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.3629499257473195,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.3619,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.3421570161735196,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.3648,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.34025786237257094,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.3663,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.4211279540109525,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.3888,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.35353724521426144,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.3671,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.40417433944392905,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.3675,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.3600227522883374,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.373,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.3941239637592317,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.3642,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.37614185032997965,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.3756,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.36881126486952404,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.3791,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.41838488012191255,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.3702,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.29640001863848703,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.3758,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.4110429213027619,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.3694,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.36339832322530297,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.377,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.359042903063112,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.3718,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.36824581866834066,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.3887,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.3476857099483462,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.376,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.39337785955331445,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.3734,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.3308270213620097,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.3907,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.4325827673533607,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.3702,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.352434856720576,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.3823,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.3815264058220359,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.3658,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.4393636654713821,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.3756,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.3035037809441835,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.36,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.4666836392659794,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.3637,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.328787801017876,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.382,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.3980687747020068,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.3842,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.31794454193523736,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.3864,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.4154977371218738,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.3774,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.31308864031983225,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.3732,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.3529037102373268,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.3566,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.30071411229140343,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.3612,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.39750753601143235,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.3854,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.28464001997270627,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.3566,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.4340974242621353,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.3812,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.32196751016385483,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.3554,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.37898295508344676,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.3683,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.4089996437539883,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.3771,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.33628277856113237,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.3669,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.4426201806155919,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.3696,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.342377972554429,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.3707,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.3253411503725598,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.3827,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.3507996305521197,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.366,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.32892716972375813,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.3597,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.7242258162057098,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.3834,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.3257664895582676,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.3728,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.39325211412020494,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.3829,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.3016421576558512,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.3659,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.3817636689847074,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.3848,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.3310894955065262,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.3722,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.3405878242033459,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.3828,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.3693975581243673,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.3763,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.3291294020524623,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.3662,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.3491348690882125,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.3745,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.35834118571649254,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.3851,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.3363063855348039,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.3703,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.29369288709426816,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.3672,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.3267750521881444,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.3672,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.3545783642446784,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.368,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.3659892791682909,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.3813,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.3598158702857331,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.3681,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.3194890657602714,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.3737,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.3761029352578444,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.3658,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.28703440041021805,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.349,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.3635089076652628,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.3718,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 1.7118069236976985,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.4605,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 1.9786997212517359,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.4072,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.33408825254624136,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.3687,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.45670677460379716,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.3739,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 10.380537714009913,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.4087,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 2.2399716684369104,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.4221,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.5180043304769243,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.3768,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.3851937091854489,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.3666,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.445300998494924,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.3758,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.39617251580824414,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.3584,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.39638393072023237,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.373,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.37809753776915045,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.3657,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.4716365350113756,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.3791,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.34207111825554354,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.3599,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.40609229813188463,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.3684,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.34104909206239126,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.3622,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.3727260951765164,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.3841,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.3500808299409244,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.3713,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.3434775311643408,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.3738,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.3489752948328182,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.3747,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.38145671940173004,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.3783,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.36749263381373665,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.3657,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.37718398439665646,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.3685,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.41871434834208804,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.3708,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.33614462649835525,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.3774,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.388373595289996,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.367,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.35085000249816256,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.3759,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.40583405403996353,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.3659,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.31112961494129976,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.3587,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.3627267756805008,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.3586,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 3.1395927463564695,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.4978,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.4966092031640311,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.3703,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.4272085812876261,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.3602,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.42111683099259073,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.3791,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.501297855047176,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.3794,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.4178709240316266,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.3786,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.4776194688978658,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.3616,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.39548126294127395,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.3583,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.4428702750832463,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.3617,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.3925485488510172,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.3709,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.45475683460513755,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.3819,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.3666382298230754,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.3603,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.3812158047090916,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.3816,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.37110774626649223,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.3662,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.40283545338812815,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.3621,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.35081016153779254,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.3667,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.3892694533855677,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.3723,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.34998061169953365,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.3737,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.420409546406995,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.3795,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.30255867515980794,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.3711,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.37224542553547174,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.3593,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.32708987992319744,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.3866,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.40804819444744994,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.3679,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.35180375853170476,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.3482,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.5012504705265298,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.377,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.3236968737234415,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.3559,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.4512354537906134,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.3626,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.3497665221630918,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.3676,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.3228148878605614,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.3544,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.33508323945938145,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.3589,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.41576347910689093,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.3103,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.4786995914451808,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.3182,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.44246349385427114,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.307,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.42752416632717777,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.3195,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.4147248301290547,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.3042,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.41441544280572723,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.2997,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.4327544675159407,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.3157,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.4165903491954383,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.3103,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.3785458923768498,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.289,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.4301470323226086,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.3168,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.33972760377383937,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.302,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.38510885053554755,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.3014,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.40098769223756353,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.3203,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.3573521250965285,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.2941,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.36278006458594214,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.3052,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 1.4226300984214175,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.3597,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.4097912007462446,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.3045,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.36281193373330195,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.3126,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.3639327374484529,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.3045,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.34340013060839997,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.2995,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.39209341812823717,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.3043,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.3353382576145703,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.3118,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.3451340477884477,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.3091,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.36950774233643424,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.3096,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.37510397335409024,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.32,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.37657853048294043,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.2985,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.34545162683337344,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.3201,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.3588811340356346,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.3022,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.31819544888163304,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.2967,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.3537655943128026,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.3106,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.3289074270607059,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.3049,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.3628073630188078,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.3015,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.3438663444603401,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.3159,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.3612420034295469,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.3224,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.38517766491532585,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.3261,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.34979118685031535,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.3109,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.38167903759894567,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.3104,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.3445199214336256,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.3287,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.37727329994392544,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.3039,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.36081613509856214,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.3103,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.32736805992294127,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.304,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.32992665607658067,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.2979,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.3120020267750035,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.3004,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.36325185673455423,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.3235,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.3491503436284882,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.3141,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.3304707572485454,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.3125,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.3353935078434942,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.3057,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.29837653410445947,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.3052,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.3455966006702602,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.3125,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.2942142642970729,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.3138,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.3026911519384316,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.3211,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.307894153513738,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.3091,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 1.8685818281747864,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.3734,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.36822452178605675,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.3075,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.36331020496786215,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.3112,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.3192785061135888,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.3108,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.38867244227867875,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.31,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.3196555295528127,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.3205,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.7558245260803729,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.2973,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.3446056087642026,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.2974,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.401028225515519,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.3377,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 2.2250302040618353,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.3235,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.4591154530366185,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.318,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.3488395368352645,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.314,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.37270257334224577,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.3074,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.3717269822992869,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.3248,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.35056429061544786,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.327,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.32446085012217074,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.3252,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.3388514668650879,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.3205,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.3289878953025212,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.3263,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.2889741349629684,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.3166,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.34823337844980806,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.3078,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.3155052825771145,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.3172,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.3554241554118333,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.2936,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.3496185892036018,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.3289,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.3254710544305437,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.3292,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.30110470084425234,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.3102,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.3401910790400765,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.3031,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.3075897480912695,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.3153,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.31937871976840837,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.3223,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.37030699478193535,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.3256,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.3326855628709171,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.3207,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.3663603223082838,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.3101,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.3309621326586483,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.3285,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.37409103168322577,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.3027,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.3068827691345087,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.3237,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 1.4287764180186648,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.3707,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.3971522573154079,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.3103,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.32485520543754937,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.3223,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.6672131920294637,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.3626,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.3061709753310188,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.3084,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.3241254390216559,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.3067,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 10.510761720318758,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.4094,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.4082350947184431,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.3129,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.3394824526012302,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.3238,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.34177804109586574,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.3065,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.33987959818729246,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.2978,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.3567353758266389,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.3174,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.35338502505993136,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.3203,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.32181832138265365,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.3195,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.3740973092037561,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.323,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.3128546747247001,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.3156,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.3378403469559353,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.312,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.29974909244690023,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.3263,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.29521634920651557,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.325,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.3191834604394154,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.3278,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.29350612502475126,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.3045,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.29839217890972114,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.2989,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.3061299362072471,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.3144,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.30747596198345645,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.3028,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.3361455794786873,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.3223,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 1.886371106514409,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.3137,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.4059085533893071,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.3167,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.3378324393090295,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.3148,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.3370014490691685,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.3015,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.39760084185074096,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.3257,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.2997896318482148,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.3177,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.3426729909188098,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.3101,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.3213268989651252,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.3161,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.2722356426109417,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.3148,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.3167284443496681,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.3018,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.2785415346734228,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.3095,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.3555810475635118,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.319,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 3.0517251698211134,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.3619,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.39367377458089,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.3141,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 1.6513708967993865,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.3839,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.4501417811000595,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.3172,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.30449995349702574,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.3075,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.6920683860595005,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.3176,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.2906492140744565,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.3217,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.3070675015630379,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.3102,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.29686169093631665,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.3187,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.2969229553901235,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.3103,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.3214972295761771,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.3257,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.3019420503442239,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.3178,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.3185428718412979,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.3172,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.2779269913793396,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.2957,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3491115167795953,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.3088,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.2824898487827363,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.3231,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.3013166123587632,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.3021,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.2897551092717085,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.3114,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.2986671509547089,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.3223,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.2767466053302283,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.317,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.2872373484696063,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.3026,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.26785945963257113,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.3082,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.2652731005153018,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.3084,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.3047123766705143,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.3201,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.28887638272192123,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.3036,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.29108693247302814,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.3354,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.2928705340488569,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.3076,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.28705893653814046,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.3052,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.2938505522600221,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.2914,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.28370455261613314,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.306,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.31970559417485295,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.3333,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.27655483124638525,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.2992,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.2959515336568435,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.3253,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.2594407982629827,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.3247,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 3.0913562852635987,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.4077,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.3783227995546468,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.3146,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.2861713555400562,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.2985,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.3609202727540535,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.3195,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.29193506058392266,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.3143,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.30934749634850167,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.3152,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.4341666154085339,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.3169,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.2788029109669657,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.3117,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.3267448834796272,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.3165,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.2737686107954253,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.3033,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.28787315065257324,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.2987,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.28397270496868976,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.3046,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.2598073113211885,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.3125,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.2833951389214912,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.3036,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.3198044192958963,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.3222,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.2610694361577784,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.3092,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.32327265154923046,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.3051,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.2840862367412291,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.3056,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.30144390019566447,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.3012,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.33572394059258315,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.3005,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.269687779564153,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.3106,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.3348968384181083,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.3078,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.2710712427904591,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.3002,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.2950930410943403,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.3157,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.3158910013766789,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.3065,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.27032013343351846,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.3156,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.31353631364454243,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.3191,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.3080361120188048,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.3073,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.2990637324498321,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.3192,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.3297104826243715,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.3185,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.27718258371721577,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.3231,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.3443919264911349,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.3144,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.29799587500272984,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.3058,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.294056772911086,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.3259,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.29244331705163945,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.296,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.2587711803722175,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.3037,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.2909175497608899,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.3223,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.28274509096250966,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.3099,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.2663632794534064,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.3024,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.27255486937658774,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.3151,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.2598800995987235,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.3099,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.28391790072221473,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.3052,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.2706479580262863,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.3048,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.27111214903741065,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.3156,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.2831524250986165,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.297,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 1.1164260635238576,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.4298,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.3008485118226466,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.307,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.26916573500685564,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.3163,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.25941675138997633,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.3173,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.2972900822700718,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.3148,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.26730358019892436,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.3117,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.2623646385372355,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.2977,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.2828229712312875,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.3137,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.27636480771530497,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.3023,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.2695943594725663,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.3046,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.27915151067100047,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.3255,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.29779882258345697,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.3091,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.2993256907950365,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.3042,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.29296460845945005,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.3004,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.26224971892718013,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.2998,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.29362447625654076,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.3061,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.25869967289800294,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.3164,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.28545969356749984,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.3256,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.26970858404698045,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.3261,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.24341097268320327,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.3026,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.247537017243565,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.321,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.25399165181098815,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.3139,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.2421537269377279,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.3078,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.2664437992906547,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.3102,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.25525992967650113,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.2941,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.256766989695889,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.3161,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.2583736490307536,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.3137,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.24915576484350038,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.3181,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.2519340911187712,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.3013,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.26413831936223536,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.313,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.26901489203966195,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.3204,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.2727595518542456,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.3216,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.2520778495766931,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.3018,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.34319368966116864,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.3086,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.27773128050434226,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.3005,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.30045684385903143,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.3191,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.2923020347254866,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.3031,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.2608492405193948,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.2984,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.29717838978053823,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.3192,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.27482784356851353,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.3041,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.28496527708482583,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.3078,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.2676633859640053,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.312,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.2761418534723955,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.3091,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.2982080596846438,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.3217,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.2870704193085936,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.313,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.3785660005756884,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2595,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.3492606064735215,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.2673,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.5273215150960083,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.2446,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.3931081409118327,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.2696,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.4512121133995068,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.2457,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.31546452800831853,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.2648,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.3808160717548833,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.2647,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.295542241184904,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.2436,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.6346949411660153,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.265,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 4.110419568884763,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.3179,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.4919296525720336,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.2691,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.27704856520843935,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.255,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.3286344361423486,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.2656,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.33543279699833545,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.2606,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.27920948836621173,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.2581,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.30710975496504894,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2416,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.2601968488598077,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.2597,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.3215354534914666,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.2658,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.2988297976755901,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.2532,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.28752892416953635,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.2518,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.32484641865724273,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2665,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.2626776774981013,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.2402,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.29108169612609386,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.2593,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.3120212892281153,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.2581,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.2668341159518244,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.2443,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.5493160579548971,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.2528,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.23696316964034259,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.2661,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.25321679147578846,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2674,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.24207951312943182,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.2848,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.2553931936015642,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.2583,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.30045081421212355,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.2665,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.2503142108406727,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.2636,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.295613386492542,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.2585,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.3170226629870658,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.2498,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.2653125254349608,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.2375,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.282695500347274,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.2564,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.28543518688246927,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.2438,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.2988143854969707,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.2525,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.28894502333977007,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.2457,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.26535442746387783,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2493,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.31016112630546155,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.2546,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.239347411237621,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.2425,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.29811171197528763,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.2527,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.26860763905483476,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.256,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.2520841134303888,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.2692,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.27641216718839545,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.2541,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.2636740836670682,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.2836,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.25966660436860184,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.2721,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.25923939316441735,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.2733,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.2654632629946347,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.2545,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.2516646137810142,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.2503,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.25948743249210193,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.2549,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.24262482346903314,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.2658,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.2521888592734121,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.2441,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.24865382588677212,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.2351,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.26712039504900004,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.2688,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.2698711276866268,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.255,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.24607318286555202,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2587,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.2728400337285896,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.2552,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.2406493540256583,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2713,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.25503295844458074,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.2468,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.25661506151856034,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.2646,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.24795384438219617,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.2611,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.23976564244070603,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.2515,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.24452697437219953,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.2776,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.23209782570376267,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.2558,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.23287673459009262,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2515,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.23146074485840656,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.2678,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.2614838334046144,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.2708,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.24009132729275243,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.2543,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.2512060105038407,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2514,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.25484126772491406,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.2601,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.2517890530913261,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2513,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.24991814165087783,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.2665,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.24687297432935643,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.2687,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.2729942747979708,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.2679,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.22533900056053516,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2542,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.23554150957441078,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.2477,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.258740130964835,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.2596,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 2.4798412621096917,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2635,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.23832901629268227,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2665,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.22837120646370568,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.2689,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.2356973492439687,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.2564,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.2308746621385288,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.2627,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.2552307136547455,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.2593,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.2443967941228206,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.2506,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.2632706227760034,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2701,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.21742772576402172,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.264,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.2660237589307662,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.2674,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.26391925911763126,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.2408,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.2559556732133556,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.2631,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.25732230023465696,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.2546,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.26377498379513953,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2631,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.22769544839257358,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.2507,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.2436494708312758,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.2551,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.24915569581523717,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.2713,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 3.129529707363287,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.3546,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.27698517213094,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.2683,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.2310847919223229,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.2442,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.2485873202220095,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.2533,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.2603073739810961,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.2732,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.22818413100201668,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.2807,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.24034268689360797,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.2455,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.23417535703137546,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.2484,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.23210733402501152,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.2605,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.24062837280951552,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.283,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.23250814170448703,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.258,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.23671932574895996,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.2713,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.23464453535794083,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.236,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.23584622836972802,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.2599,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.22668597775075272,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.2552,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.23836714266566172,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.2547,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.2547402456152415,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2438,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.5961403889919263,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.2538,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.2199597952603292,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.2533,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.2467467070931257,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.2632,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.2527515157982986,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.2538,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.2348872728860179,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.2694,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.24199163761084758,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.2529,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.23895743308940554,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.2435,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.23355410572468993,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.2583,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.23512211758313514,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.2554,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.23908293699068534,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.2562,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.24369491808565416,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2435,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.22903336637915064,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.2482,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.23029392534395518,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2561,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.24556696137260944,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.2589,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.24011928825989093,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.2436,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.23510039163775298,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.2414,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.249631985420191,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.2467,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.24667059878732872,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.2561,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.2437053616596856,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.2215,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.24187908187183188,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.24,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.24700156194315212,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2423,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.24092821412671772,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.242,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.23004037179138398,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.2376,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.22963129070669946,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2813,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.23235845434802543,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.2519,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.2405794222679829,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.2574,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.23644132599794299,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.2661,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.2391236919104885,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.2525,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.2664538924906814,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.2415,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.22985244890511058,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.255,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.23692441546182533,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.2497,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.2444789956678517,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.2464,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.22812163245633493,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.282,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.21984611736729437,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.2621,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.23492102329435666,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.2506,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.24181100152626095,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.2596,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.2185740700004013,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.2574,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.8061513863746467,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.2731,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.22527928392996224,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.2564,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.2429692840111762,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2536,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.23144321280194935,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.2517,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.23379526275714532,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.264,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.23224211112630067,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.2636,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.2280086392196385,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.2626,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.22649528259625418,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.2637,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.25399164840084554,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.2836,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.24112185989323903,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.2506,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.24010053007129492,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.2675,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.22800570725759495,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.2719,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.21936178223588743,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.2502,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.24495823870785047,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.2443,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.21719650370965815,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.2479,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.23429855635137067,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.254,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.23642020476276426,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2675,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.22230098176846108,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.2707,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.22820040290469285,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.2579,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.22095062732806642,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.2795,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.2194151996618836,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2516,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.2237018776475978,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2561,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.22156783719611184,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.2498,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.2369995471752677,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.2276,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.2582790807073243,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.2481,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 2.647818289134684,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.2757,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.23117410376260505,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.2508,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.24395636488948638,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.2518,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.22898200973865906,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2489,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.23068571466002805,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.2642,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.2525406992724317,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.2657,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.2507260765158667,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.2705,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.2443784216715562,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2531,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.2351032195998317,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.264,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.24390888034218638,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.268,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.21963293909635254,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.2396,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.23799437517150088,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.2613,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.2572327979376714,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.272,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.21837984951418649,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.2684,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.23693442795834632,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.2302,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.22980784757209835,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.241,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.226220158043529,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.2534,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.21659292447926637,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.2776,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.21287777522240292,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.2587,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.2169077428625463,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.2511,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.21394140978737966,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.2489,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.23007779796923003,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.2427,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.22295341968588017,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.2585,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.20187499484357668,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.2683,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.22142534534359565,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.2692,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.21978424101436098,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.2554,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 1.6765915582844608,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.3139,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.22099244508157892,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.2571,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.23338357159046402,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.2493,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.21903842424565434,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.2485,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.23855451509833664,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2587,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.2224067506217238,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.2439,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.2180306201638381,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.2586,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.2207075315480691,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.2572,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.20361545360007163,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.269,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.2152262388129453,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.2524,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.22401281275675838,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.2542,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.23025688895705299,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.256,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.21216674094409652,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2422,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.20916279221271614,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.2511,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.21718866698123565,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.2471,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.21716962143992058,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.2599,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.21670613182821563,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.2606,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.21833561131232238,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.2643,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.23027523022468077,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.2422,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.20622454067768806,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.2688,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.21160466746235027,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.2561,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.2197698314356089,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.2307,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.22627107156450255,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.2594,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 1.8943889528719315,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.2987,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.2230839812781848,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.2521,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.20504624958394135,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.2667,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.2222324021333317,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.2408,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.21483273703945374,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.2565,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.23848820252919456,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.2492,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.20559483730302255,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.2663,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.21612594134943383,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2435,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.2337735283096324,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.2615,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.22810807441765504,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.2456,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.23205889467136687,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.257,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.20635201377609716,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.2433,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.21939634832962074,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2574,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.2178387000127759,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2521,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.21965167781036074,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.2471,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.21813650483293243,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.2535,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.2107353923106093,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2598,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.22111329671330895,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.2452,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.21689702960832885,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.248,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.21433209460811956,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.256,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.2185918489476274,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.2614,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.21403294859087646,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.2612,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.2170133372828556,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.2729,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.3664408298539393,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.2329,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.28054390414018177,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.2128,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.2607773591389432,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.2251,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.4147618992268116,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.2125,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.3954849460294434,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.2158,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.22881211849565375,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.2245,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.29755961473806847,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.218,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.3873364431279107,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.2463,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.31536326334591136,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.24,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.2390859105448519,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.2121,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.2493819154395797,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.2396,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.3245203492897487,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.2058,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.32470734429658465,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.2283,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.24976406509120108,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.1987,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.20780471874824477,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.2214,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.24063609468504354,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.2285,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.2574029682439464,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.2201,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.2773896213337786,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.2333,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.21260423218226418,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.2171,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.2522246453898413,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.2109,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.2680349520844927,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.2041,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.21862219940815864,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.2163,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.20464248089847287,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.2232,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.1956265091971109,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.2406,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.23903594635314626,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.2137,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.21065504273540284,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.2343,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.19466373497866374,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.226,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.2028311762257227,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.2407,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.21185845302844697,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.2163,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.20907298168762778,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.2235,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 3.9504979155750917,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.2475,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.20985841360927496,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.2089,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.20757534752453985,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.2089,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.21816319348546603,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.2127,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.20150207093138875,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.2182,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.1964630669133695,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.2288,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.21068484536153761,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.2346,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.20024773053998285,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.2274,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.20417850681019176,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.2388,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.19720050303143224,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.2245,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.1940308718529191,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.233,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.19863953740458146,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.2186,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.20179549826977627,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.2173,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.19910596235089056,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.2248,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.19952503843777603,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.2318,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.19793068504143047,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.2224,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.19236254440252307,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.232,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.19682297342507935,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.224,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.19999967974653474,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.2228,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.20013011266920766,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.2207,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.19971561619645617,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1976,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.2076420584055411,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.2286,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.18939867807039745,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.217,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.7935692555032997,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.2419,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.20291036302962803,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.2085,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 1.9388600940697014,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.2521,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.19119439626131288,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.223,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.2006808669678455,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.2204,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.2009323607741129,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.211,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.2038333785558482,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.216,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.1960476591919314,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.2201,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.2033861413178178,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.227,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.2022385489007153,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.2231,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.19317858192395784,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.2198,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.20209062169707054,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.2331,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.19537716454042453,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.2232,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.1925223277414691,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.2169,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.19363679457927555,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.2511,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.19389418363809124,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.2288,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.20604666092014934,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.2201,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.18871566283903815,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.2245,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.1928492486181577,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.2264,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.21156637875597756,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.215,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.1860129807405497,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.2254,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.19269869051704577,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.2037,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.47859192718225907,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.2346,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.19277039572593208,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.2249,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.2020356997313471,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.2223,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.2060465101565481,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.2205,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.19493179037624708,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.2224,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.1922948277726153,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.2231,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.19250777273941075,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.2208,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.19672154959860294,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.2207,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.19617637790378106,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.2176,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.19827333052983848,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.2258,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.19845893494579764,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.2046,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.19758386620541665,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.2204,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.43475341974791787,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.2177,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.1887711111618614,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.2134,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.19339156895384774,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.2364,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.19317944947275043,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.2035,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.19957480504741693,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.2375,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.20054905639855175,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.2396,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.19834398407979276,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.2106,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.18889126246587304,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.2288,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.9721627378177307,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.2631,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.1995238793515951,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.2258,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.20423509420768501,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.2081,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.1965928986102977,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.2254,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.1882611083168841,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.2166,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.20315750673034066,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.2184,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.19273940567662234,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.2049,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 1.9643360732643327,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.2714,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.19722651325814103,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.237,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.189663176315684,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.2016,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.19413456601005566,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.2055,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.19445858572533284,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.2231,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.18879231375478814,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.2344,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.1904527073773965,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.2276,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.1926949927639869,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.2211,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.19054210916787626,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.2139,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.19352864097370293,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.2374,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.18726237342410337,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.229,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.1987797235127207,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.2035,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.19194074185908155,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.2142,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.1924011890772762,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.2212,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.19963540019223028,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.2067,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.19989056977854508,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.2184,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.1885608232506812,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.2109,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.18753281714723,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.2348,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.2101521062979427,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.222,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.1858087250886233,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.2216,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.19661029941596672,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.226,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.18786404726764466,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.2321,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.19422129813234912,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.2039,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.18735285834805154,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.2441,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.2600095099390707,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.226,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.20204307928469462,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.2197,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.19218407606419596,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.1932,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.19953987869549714,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.2185,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.19881148819393446,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.2199,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.19497279174973103,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.1946,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.19365511492830348,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.2389,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.1860642255205971,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.2372,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.18409422816941237,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.2184,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.1994219196582933,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.2413,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.19082633793122997,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.2256,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.18468606713549082,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.2265,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.20088011131436875,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.2097,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.19324222083488177,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.2215,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.19429472715567372,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.2312,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.20151095647775583,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.2023,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.19419600053361488,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.2138,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.37020539627043897,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.2111,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.18901543358303285,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.2302,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.19678480166127674,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.2097,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.18575163671494418,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.2215,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.1843128581778202,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.2065,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.19151622934908985,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.208,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.18384921273878035,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.2293,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.18452972301493759,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.2144,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.19919139359882912,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.2065,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.19069828184882426,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.2162,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.1904775673139846,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.2256,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.19785955998610402,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.2326,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.18755854025746058,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.213,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.18994571276042535,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.2222,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.18713817661849386,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.2282,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.1944563786641667,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.2046,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.1910323050680513,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.203,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.20013859347476434,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.2095,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.19122318268301491,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.2186,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.19595645780771095,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.228,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.19443183587753157,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.2053,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.19640972626035705,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.218,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.17758190402359061,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.2194,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.19709347153154316,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.2212,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.18892457887246314,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.2208,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.1815658176840994,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.2425,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.18807844776112395,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.2297,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.18685921370008118,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.2282,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.19723282692247607,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.2209,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.2003214038411181,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.207,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.20264228207304028,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1936,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.19702310698928652,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.2313,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.19185620773055284,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.2117,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.20580648833577214,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.2245,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.20230832813335806,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.1931,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.1871558386569946,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.1984,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.1968483218883922,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.2262,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.1818626611626006,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.2268,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.1894136188913656,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.2163,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.19454914993883388,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.2256,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.19329245900910008,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.2186,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.2016604646478333,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.2028,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.19628613864809027,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.2183,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.19205553769872497,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.2064,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.20063334416761805,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.211,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.19595629940077455,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.2221,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.1887795162395183,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.227,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.19693141781128926,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.2129,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.18967996960985198,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.2122,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.18504206469002032,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.2278,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.18467491344448833,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.2148,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.18066669791206674,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.2056,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.18474608010412996,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.2163,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.18182787652426352,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.2244,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.1874619498829968,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.2194,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.18923952871283314,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.2191,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.19762902613173952,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.2015,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.19010541426753974,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.2124,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.19499451738602808,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.2131,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.19274969989045496,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.2317,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.18990632655027198,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.2222,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.1945505357506228,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.2268,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.1859325779580991,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.2267,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.19342877180616275,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.2172,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.18981260143460352,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.1989,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.20306630550063237,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.2032,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.18454716444764696,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.2252,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.17994001669875848,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.2241,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.18581832707967494,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.2084,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.19133094143683343,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.2257,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.18858010801523836,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.2176,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.19434024784896886,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.2164,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.19026637817884054,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.2096,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.18962769368743254,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.2266,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.18973866846058757,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.2186,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.18816565164795138,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.2161,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.18462094865990794,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.2315,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.1836283427363123,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.2155,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.1879438356578156,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.2293,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.19249379342721704,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.2098,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.18651325675691302,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.1954,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.19631068959054934,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.2109,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.19162083155353066,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.2127,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.1906513738533297,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.2132,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.19071611840505634,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.2149,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.19050600144502372,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.1999,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.19170892023162456,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.2203,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.1848097131031078,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.2252,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.18529033297477557,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.2118,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.1920785455232159,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.2165,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.1856948098772787,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.2411,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.19779355074622887,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.2158,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.18761060503798502,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.2046,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.1954633381892792,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.2035,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.19573178371699937,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.2099,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.19430241872311624,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.2155,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.18871001261044035,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.2264,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.19767468262487714,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.2078,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.1874255250598259,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.2177,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.1976978693715012,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.215,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.18621887672996412,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.2146,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.18933533657301435,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.2149,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.34451416098657917,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.2308,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.18534798246049602,
+      "learning_rate": 0.0,
+      "loss": 0.2252,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 3.5717155096593695e+18,
+      "train_loss": 0.329093080264354,
+      "train_runtime": 28327.3716,
+      "train_samples_per_second": 5.578,
+      "train_steps_per_second": 0.044
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1235,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.5717155096593695e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}