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": 6.468109765992959,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.0703,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.344720326748741,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.0487,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.4149934291354205,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.0382,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.332087642396387,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.0385,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 5.878640377946109,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.0144,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.903974057506374,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.0441,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.632694804891455,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 1.0153,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.229549481477283,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.9964,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.7237614586778554,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.9483,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.533261880461903,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.9676,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.4760433113677385,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.9554,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.7248693531384696,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.9522,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.038021303365605,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.9467,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.871887578499984,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.9369,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.652484117602737,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.9214,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 2.6752792329268975,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.912,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.4619066731564314,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.8774,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 1.998360673271129,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.8619,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.5506829955993457,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.8614,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.027192980383625,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.8581,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.017754336506802,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.8439,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.167687969481697,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.8164,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.1079430245719972,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.8133,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 0.9581724850300047,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.8008,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.8688023715613664,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.7901,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.9517674429296324,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.7865,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 0.8960959248383995,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.7845,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.7441945975018933,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.7786,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.7487272391227141,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.7743,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.8530509049709812,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.7749,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.7270678981836934,
+      "learning_rate": 1e-05,
+      "loss": 0.7637,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.6503788815268033,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.7518,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.701273036290457,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.7603,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.7707111367416855,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.7545,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.649315009316372,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.7408,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.5548797418063167,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.7334,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.5571558681912275,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.7426,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.6611045280554834,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.7303,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.5342983395615125,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.7352,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.5935167599872428,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.7328,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.6002823050783924,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.7328,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.48333769457973785,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.724,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.5230794849062826,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.7211,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.5842624897969625,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.7367,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.5030775038560161,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.7148,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.538486423733782,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.7187,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.6221034259978968,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.7059,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.48292085836041604,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.6982,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.6123940852107901,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.7181,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.5581443556353028,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.7185,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.52013146605216,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.7083,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.4057676664692682,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.7096,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.5675527858481032,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.6975,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.5633642707823989,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.699,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.6031077757049423,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.7011,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.4369714344454582,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.7109,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.5309113344629813,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.6902,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.5813401607205175,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.6827,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.5596719878765363,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.708,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.5187859976511674,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.6963,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.7873812608251413,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.692,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.7447981372912805,
+      "learning_rate": 2e-05,
+      "loss": 0.7035,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.6853581341223728,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.6952,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.5204836175012415,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.6883,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.5471441056172252,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.6899,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.52869574394853,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.6856,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.5345476958948412,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.6784,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.6872545257714947,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.6855,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.8822482350954779,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.6971,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.7514842749572818,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.6851,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.9473455288956615,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.6697,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 1.2246952619498739,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.6836,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.8472672307853,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.6856,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.9835728187033558,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.6843,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 1.1093508508190393,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.6986,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.6067858991144882,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.6716,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.9513830464129313,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.6847,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 1.3644193806161344,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.6686,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.7481282732954673,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.6736,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 1.7673528516289936,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.668,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.8709140493531121,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.677,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 1.6292189400944017,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.6871,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 1.119547955993395,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.6731,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 1.563911938517457,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.6619,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 1.2700080739212778,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.6825,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 1.4018984388608773,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.676,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 1.26214292616709,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.6696,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.9754717433878335,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.6894,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 1.1710423206632896,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.6728,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 1.0134879669504346,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.6703,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.3688792184144003,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.6832,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.8348432364201743,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.67,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 1.0797563801836392,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.6683,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 1.2179809183869847,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.6648,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.8755730894781338,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.6725,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 1.087677096913398,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.6832,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.9729177863631804,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.6756,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 1.1310571260714033,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.6703,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 1.2557076403082792,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.6803,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 0.7676376500183847,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.6618,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.9081692681003226,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.6733,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 1.2553547691803142,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.6733,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 1.3954692621640143,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.68,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.6425292654309613,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.6583,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 1.182707490344512,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.6715,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.025866303967039,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.6763,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 1.105638807811866,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.6589,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.7911471696345186,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.6648,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 0.8796853710186825,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.654,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 1.5303692107303948,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.6677,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 0.8183781972539376,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.6749,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 1.3770529662330038,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.6696,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 0.7628231662981719,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.6681,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 1.091382117705418,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.6647,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 1.1764082705794554,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.6481,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 1.0318389788738973,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.6608,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 1.492985398067675,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.6597,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.9122615049558864,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.6779,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 2.04915795871591,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.68,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 1.1064576373600152,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.6786,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 2.3005138548134862,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.6742,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 1.919080484754206,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.6881,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 1.8354260231308974,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.6795,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 1.4134193496425136,
+      "learning_rate": 4e-05,
+      "loss": 0.6705,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 1.9852297799126326,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.6944,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 1.4619821129737787,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.6735,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 1.6852260926896556,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.6807,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.379012043528442,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.6587,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 1.6161109969778849,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.6628,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 1.233316748086653,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.6655,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 1.7768383174817193,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.6555,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 1.4318769147003994,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.6757,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 1.6136765600009235,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.6583,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 1.5320591189321267,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.6715,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 1.288315820161813,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.671,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 1.204067168782626,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.6799,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 1.311894391836944,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.6663,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.9668006035160325,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.6607,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 1.2147838153629265,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.6693,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.8554182139637558,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.6519,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 1.1938725432581965,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.6668,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.7968686819790375,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.6392,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 1.0476182672429508,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.6607,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.7143098771675199,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.6506,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 1.066211154807631,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.666,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.8941876829755752,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.6457,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 1.168476968496031,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.6515,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.7788045964271537,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.6406,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.8511357786261358,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.6585,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.7809055910019252,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.6587,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 1.0860442322974189,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.6451,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.7130526168810114,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.6556,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.8107377623249106,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.6531,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.6337303681327097,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.6421,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.8647842497552319,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.6543,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.6205623911999734,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.6514,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.6021261876688216,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.647,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.5987756650280364,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.6659,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.710529451485544,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.6486,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.8378178637496151,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.6516,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.6836361329499052,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.6483,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.7908993995923895,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.6436,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.9443981392208999,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.6552,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 1.2402872371186617,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.6486,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 1.018336698316595,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.6438,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.5891659294845437,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.6319,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.7185697574866865,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.6488,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.9667363191134639,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.6546,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.8237155929441756,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.6759,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.6146164677913598,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.6384,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.6165312540410313,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.6535,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.7467517085454958,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.6545,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.7297013210720285,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.6502,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.8278060978117822,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.6397,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.6259851675791982,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.6419,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.4253097306122165,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.6407,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.5804631759549862,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.627,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.5578699637863188,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.6469,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.49033602527298004,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.6474,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.5676752842121191,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.6566,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.6037796006786365,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.6387,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.5319127221226274,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.6359,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.8584159857915639,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.6596,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.673559509454145,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.6449,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.5723833249522872,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.6385,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.7488878350450471,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.6544,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.6176234290887035,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.638,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.5368064087032255,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.6491,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.5026819792027188,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.6387,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.4889344296178706,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.6366,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.5059040767607228,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.6325,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.45783865026145193,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.6391,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.6040510063646028,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.6466,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.9831302524115538,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.6308,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.7137771780011558,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.6251,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.5193841388914379,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.6312,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.8463884063659829,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.6557,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.6308168170535164,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.6363,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.5272053005152946,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.6315,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.5758186339009932,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.6346,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.656898881015555,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.6407,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.6170336330181433,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.6133,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.6250915485605649,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.6277,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.7386240703166879,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.6589,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.5269121092657277,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.6267,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.48086865136290047,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.6346,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.5779886201469309,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.6338,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.5965261553108728,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.6575,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.7404643759217113,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.6254,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.8092773798196258,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.6411,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.7214101018169119,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.6242,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.5256462662286692,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.6172,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.5297171354962475,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.6422,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.4923392593384204,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.6496,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.5201537440479553,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.6268,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.5188921241982254,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.6266,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.5211883642645548,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.6253,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.5669627740571819,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.6467,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.4661305033694006,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.6274,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.5580032183281534,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.6259,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.451113761430687,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.6341,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.45558957402678607,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.6465,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.3873007041022168,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.6158,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.4701526111160403,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.6407,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.4137905216318286,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.6337,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.5144827105212197,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.6368,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.5642193138763121,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.6311,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.5656655182415025,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.6333,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.7708016422667034,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.6364,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.7576982906321226,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.6451,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.5004306139985089,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.6372,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.44932490845447187,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.6506,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.7469980710325189,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.629,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.8403118597781046,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.626,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.6297536699312064,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.6429,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.5046599764092574,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.6256,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.7225000007181877,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.6252,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.8519374970416779,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.6297,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.5533113549298865,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.6191,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.577437346548963,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.6321,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.6130091619690886,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.6343,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.5789006539134625,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.6332,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.6212654189106194,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.6257,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.553089472146024,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.6349,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.8155292273015877,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.6382,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.49583244994021586,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.6235,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7108504057387817,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.6376,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.5314155128383439,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.5823,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.46686404453982133,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.5965,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.6407090961729659,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.5823,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.6580048901808531,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.5738,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.5501056061218143,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.5899,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.8204760227595117,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.5677,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.7454644211678966,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.5985,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.7088099027326181,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.5819,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.7377629491428992,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.578,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.5632374158262449,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.5903,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.5748229935207868,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.5628,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.5912376899766753,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.5787,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.5947919715571378,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.5837,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.6693301484262061,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.5743,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.4643734055345145,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.5803,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.5299815362078812,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.5764,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.6278232395286171,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.5813,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.585268812643381,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.5792,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.5179595147339898,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.5825,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.525497888881259,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.5666,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.4987793051124585,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.5781,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.562524009915611,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.5743,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.48483457747549585,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.5851,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.6144794466889696,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.5604,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.7157508422064507,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.5897,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.5905040334312258,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.5723,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.5416269563304246,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.5785,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.5499037516983251,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.5715,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.5990618670841819,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.579,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.487357704696692,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.5778,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.530500883868585,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.5919,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.5044232122122166,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.5685,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.5696255269051674,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.5865,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.450729130718413,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.5773,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.4621569730774247,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.5662,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.4634562296192485,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.5749,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.5748815263014613,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.5858,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.4219777757762244,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.5849,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.48392830742415704,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.5765,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.5800258007565128,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.5774,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.5254503587003061,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.59,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.5144409627645512,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.5606,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.5531223382707695,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.5775,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.5472429420933893,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.5878,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.6276543336616816,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.5814,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.5790593404304276,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.5868,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.5549912360669516,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.5743,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.5096131160973747,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.5772,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.6430179465725279,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.5854,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.8725226974025424,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.5767,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.5415599436404113,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.5638,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.6175002210116763,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.5786,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.7591655966977433,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.5843,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.7539238300189914,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.5791,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.5694806405981085,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.5935,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.4859583785773087,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.578,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.6907987457615984,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.5814,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.6654888225364667,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.5833,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.5042746350159831,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.5839,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.6320633478318144,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.5873,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.5950396726777643,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.5802,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.4001834489131382,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.5776,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.4781389736914105,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.5698,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.556707925650086,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.588,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.542749124752331,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.5848,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.7084764751496512,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.581,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.4702413483764299,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.5702,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.5534525698160969,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.5737,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.8017629900905101,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.5858,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.46012265742751735,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.5774,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.4864097088019957,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.5718,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.5625668941296641,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.5701,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.4292659656311284,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.5714,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.5856828755381612,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.575,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.46672389483299503,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.566,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.5101430979787078,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.5707,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.4835559033299343,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.5748,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.46041438298989656,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.5775,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.6056721356151675,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.5782,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.4632999661357482,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.5745,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.5260118275758405,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.5767,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.4353924156657436,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.5818,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.44189154912716366,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.5843,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.4500190014370777,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.5782,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.3622718048903468,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.5734,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.41323451071507133,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.582,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.4664054029031822,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.5751,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.4174897876474871,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.5715,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.3881877932077531,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.5705,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.393529325530471,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.5861,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.4098849661183463,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.5803,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.4197502382883329,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.5821,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.4061144685557598,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.5843,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.5931212355260799,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.5693,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.5747606620103699,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.5786,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.4166301886438857,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.5883,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.5572640983732021,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.5838,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.585979422704714,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.5698,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.5360483434807496,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.5697,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.44481696622868827,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.5693,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.5378376559106056,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.5851,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.4598352493009354,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.5626,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.4833279582172982,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.5824,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.48438354568359476,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.5783,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.5587481598965377,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.5713,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.5561725097817287,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.5858,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.44129883361210315,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.5634,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.5039624631060948,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.5787,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.4913480089344011,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.5819,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.3673731544026551,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.5545,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.6296051301123883,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5902,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.4478887466300343,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.5679,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.4769539008968257,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.5816,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.5576219442979026,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.5849,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.5114520358298641,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.5631,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.5100950804619335,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.5834,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.6590284490078259,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.5602,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.3749440302221245,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.5842,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.533737476679607,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.5813,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.43544149198977083,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.5801,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.43261853673644746,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.5786,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.48924513954148174,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.5769,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.8501882754426849,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.582,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.4751092135317685,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.585,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.4762438564281442,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.5761,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.4745234971819465,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.5572,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.39486975803932733,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.5737,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 1.5960380077966927,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.5873,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.37386221548219234,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.5735,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.4773539487597005,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.5813,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 6.43937606796315,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.5782,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.9146316692788132,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.5812,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 1.006911577635789,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.5874,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.5182993699590813,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.5736,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.6781222092723089,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.5907,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.7799381687285848,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.5659,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.7217829953753383,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.5758,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.5407962585257036,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.586,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.8091873678694935,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.5768,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.5238318907933134,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.5759,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.5092450765334783,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.5799,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.7174318947391319,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.5793,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 1.0417842093759684,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.5885,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.5729389572254955,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.5636,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.6674262616800668,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.5734,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.5353687723211406,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.5765,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.6553170365479115,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.5776,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.6479196067536364,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.5713,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 1.2037581014635714,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.5824,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 2.8681713426469164,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.588,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.8704944047087104,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.5888,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 1.2307441361464506,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.5882,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 17.594904039778584,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.608,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 9.107900741789482,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.6081,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 2.5293558890081487,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.5873,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 1.0553387682079496,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.5834,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 1.5770749010604637,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.586,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 1.6104532890474104,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.5923,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.905462767843162,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.582,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.9547003489695844,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.5763,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 19.10256793391456,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.5685,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 1.6287384805184768,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.5818,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.9352588291551833,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.5917,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 1.2103506880924872,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.5837,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.9646628104542234,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.5769,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 1.1585247380658914,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.5809,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.911038602423582,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.5722,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 1.0076842641311652,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.5965,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.8893851268510462,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.5815,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.726751081166138,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.5764,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.7366810207029251,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.5748,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.6953433980867739,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.5646,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.7468518882846402,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.5842,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.9403754169084851,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.5782,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.499728159242601,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.5681,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.8134798427985863,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.5848,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.5132423008948825,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.5762,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.7489970245917877,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.5703,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.5095791093448965,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5864,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.6932378624758002,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.5772,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.5289402650998729,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.5815,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.627488357771487,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.5704,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.4869548661687247,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.5709,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.49916050509436266,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.5677,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.5776195754861082,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.5735,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.4701451495680357,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.5554,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.6110126545531873,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.5744,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.4313600187692805,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.5712,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.48159946079937893,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.5669,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.40471774564454316,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.5688,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.4201991326060129,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.5765,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.4170978212427515,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.5708,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.3342610992917528,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.5772,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.3533684444929981,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.5867,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.34363455741135585,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.5769,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.3638135367344141,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.5817,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.3671865737738111,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.5675,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.39123553533629274,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.5689,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.35721329985023925,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.5801,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.438987497232136,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.575,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.35197926678497554,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.5765,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.4340924338879996,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.5586,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.37092230415265515,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.5671,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.4115058774504886,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.5673,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.3877359920821406,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.5736,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.4435788173403523,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.5732,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.3675274021312594,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.5746,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.4628016211049096,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.5871,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.35079656780617885,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.5691,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.4362429052016364,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.5904,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.5150743985617391,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.5796,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.2994293781211465,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.5669,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.4692083175816105,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.5846,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.42216698910855527,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.5765,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.4312261719299481,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.582,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.44605028432322347,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.5594,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.3652627978729729,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.5713,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.408895103778046,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.577,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.3666469548031559,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.5678,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.31181492926892906,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.5703,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.3484110993176375,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.5703,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.3143505142942207,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.5645,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.31473861447722123,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.56,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.30146285767562214,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.5754,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.30733801281423345,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.577,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.28366411184800294,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.5672,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.3607307032335403,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.5669,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.2853820501168446,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.5659,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.35233847163780363,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.5731,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.5524009638480728,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.5849,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.33986561879981503,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.5668,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.2956772024964904,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.5697,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.30275114498244177,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.5727,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.3086831979851712,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.5786,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.2926574395067476,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.5685,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.33941152828194954,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.572,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.32377766315221357,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.5674,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.3377866155058963,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.5761,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.3568390033358345,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.5817,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.3403339463708182,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.5637,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.32027610501018083,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.5657,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.3248633375718466,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.572,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.32385661262164583,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.565,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.33453520158786615,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.5895,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.38631218383263966,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.5902,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.39647881002301283,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.5644,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.3584931247489666,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.563,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.4455526138446247,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.506,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.3720613627154453,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.5135,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.5390211618392352,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.5019,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.4146491728379713,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.4933,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.48766618940297884,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.4837,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 1.414573724462544,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.5031,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.48473149403936017,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.5,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.4289082000324169,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.4887,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.44783477485268847,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.5084,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.4978940115457792,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.5117,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.40785678012809423,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.5159,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.4109069261301344,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.4917,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.43550361543416544,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.4997,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.9861531033673167,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.5124,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.4631439181320009,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.4885,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.45286796781678434,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.5063,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.4237612850129412,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.4994,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.46977167219948557,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.5104,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.4227691815128243,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.5063,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.42104456765085585,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.4904,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.42481620326659647,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.5027,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.4240128573706683,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.5003,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.3826644424085992,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.5027,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.44417997291657724,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.512,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.35421701045939347,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.5208,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.43734397647143075,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.5048,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.3356820474843771,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.5006,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.407789887174987,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.4859,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.37618789795373275,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.4977,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.4398725539981299,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.505,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.3717674964653083,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.5043,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.3422671445094512,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.5097,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.35232392560004855,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.5079,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.3363648712729031,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.4813,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.34621687095638415,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.5046,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.3096014686222518,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.5061,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.30486654747635,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.4991,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.3380140598065726,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.5074,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.28890748821955303,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.4954,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.35660148960025917,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.5048,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.3053332131787813,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.5012,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.3681028802807221,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.5016,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.35384248134595836,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.5042,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.2703024888848084,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.4953,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.30917914010972336,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.5153,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.27800319719330047,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.4954,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.3156744579755463,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.5034,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.3394937203842575,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.4932,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.30688471365003595,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.4866,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.29765439575106595,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.4908,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.3424094359743664,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.5049,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.511295428501913,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.5314,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.33717871940134325,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.4868,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.4302634455477321,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.494,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.3000959651863418,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.4987,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.3704102521655032,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.5028,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.3311100672235221,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.4894,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.3908569084840423,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.5089,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.28249318175053584,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.492,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.30516383308470274,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.5031,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.33035328279721055,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.4957,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.3298564672079724,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.5143,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.3008262004555948,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.5062,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.3595863834491008,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.4898,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.3086444109921619,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.5154,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.31435780532686824,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.5037,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.3275761305401726,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.5049,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.2885837634417834,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.4967,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.302041496624443,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.4962,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.2939147839479889,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.4957,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.30972712046518863,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.5063,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.3529516931857674,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.4912,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.2699538008311739,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.5057,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.31728011262221195,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.503,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.3008586341463175,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.5006,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.3052429525911378,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.5086,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.31948464573785257,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.4941,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.32119678946224567,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.5037,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.30799723318628625,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.5126,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.35175970094207826,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.5076,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.31364857138777236,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.4967,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.3365079484915002,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.5052,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.3173429195428106,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.5006,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.328787914284625,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.5039,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.3258575507782665,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.5054,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.31901228657052305,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.5104,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.3609886588789144,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.4968,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.2688924725858756,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.4998,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.2793717274104096,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.5033,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.29245182463078495,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.4924,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.3208464227255332,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.5212,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.31888775565410643,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.5046,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.7939415297640601,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.5208,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.30807310787148384,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.4993,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.403485846738937,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.5068,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.3801345059919768,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.5105,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.2977275713547556,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.5024,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.42501688189412684,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.5115,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.3191597031076031,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.5052,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.3909768022669353,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.4966,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.28545719775928,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.5138,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.34848326408004576,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.5047,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.33455636792190135,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.5036,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.3309395111137566,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.4867,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.33301917306167067,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.4987,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.33672594943203477,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.4984,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.30186333205709853,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.5108,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.2888446364327021,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.4924,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.6631534970405282,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.4922,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.3244590496073278,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.5009,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.33044079634445894,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.5111,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.29555753988990124,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.5002,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.31702518846580113,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.504,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.28279176556427993,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.5151,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.32392152410440744,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.4919,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.29891805536009386,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.4964,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.2889316226691634,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.4999,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.2919800927403157,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.4922,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.349300020310836,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.4907,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.29270465103748555,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.5126,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.3832192148653457,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.5103,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.30045672722615924,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.4896,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.4313003982145353,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.5081,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.3848421318253727,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.5077,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.4270819828529916,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.4977,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.3916500276399153,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.5045,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.4221507220058398,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.5059,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.3453301486594238,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.5135,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.37144816954187787,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.4936,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.3285805626436351,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.5052,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.388895033353152,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.4948,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.3173489008726406,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.508,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.3898790678777697,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.5163,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.2971580511464276,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.4931,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.3618771420695581,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.4961,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.2733907108385662,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.4981,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.3227086288118721,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.4947,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.3174994515343733,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.5199,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.26129482448705915,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.4962,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.29883978831847835,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.5066,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.30957550694870584,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.4985,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.32161403252786946,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.5079,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.30062484732431716,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.5042,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.30944580126461924,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.5022,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.26606900159921776,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.5007,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.2911244666985617,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.5072,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.3021533182346288,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.5002,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.2963736113074982,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.5056,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.3136726170412457,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.505,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.27177796353121253,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.4908,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.32147061277954486,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.5085,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.336352630021967,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.4999,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.32341379165517153,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.5014,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.33059270154474063,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.5105,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.35322863599634485,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.4998,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.2933658687633399,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.5162,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.3221719272100289,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.4993,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.2890514793171781,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.4907,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.3065137923824865,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.5086,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.31363588318754565,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.4968,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.7067463953548598,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.4973,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.31921740493362294,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.5082,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.28672374062749806,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.5083,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.3618812156593899,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.515,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.29841032641377174,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.5027,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.3367236834960781,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.504,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.2962420607962803,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.4945,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.26766091321385754,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.5052,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.33657430723141957,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.4912,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.2784416947750599,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.5048,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.32833400420623865,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.4853,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.30523040111504307,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.5088,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.30178340669198034,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.5078,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.3049081606072509,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.4958,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.32704998368948,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.5004,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.2757684432905046,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.482,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.29146132993410356,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.4978,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.7637165526956056,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.5324,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.36796902712371127,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.5092,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.27519442126467725,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.5045,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.3101188756211647,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.502,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.27328784349709134,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.5072,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.28642298411692546,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.5061,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.2705207790303987,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.5115,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.29767940785873304,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.5021,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.3022713918150079,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.5091,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.2698826948452932,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.5082,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.3068490582200055,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.4904,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 1.783589463149265,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.5124,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.33611132133075894,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.5025,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.3097175440703449,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.5024,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.3199632950882559,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.4937,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 1.346781813439481,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.524,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.3651035150055135,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.5149,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.29413736461414614,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.4977,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.31020735315869374,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.4998,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.31972322362427547,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.5018,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.2890736249167476,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.521,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.3323719395545838,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.5169,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.302642461148754,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.5025,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.30705765902258336,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.4977,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.2769057897229502,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.5074,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.305580932737304,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.507,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.29847665189076383,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.5048,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.28914822250306493,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.5007,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.2633362990068182,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.4991,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.3445235713981302,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.4982,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.2702893279058683,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.5165,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.2891954243693674,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.5058,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.28132403891851465,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.4995,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.25049844586841347,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.4973,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.29923476608166394,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.5092,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.2657698354434878,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.4984,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.28460401733252827,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.4999,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.2716217466924454,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.5037,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.24362464014149904,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.4975,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.26966556825756177,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.5111,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.25416464222736085,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.4881,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.24849279620164136,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.5057,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.2748101854303737,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.5042,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.24754234921665522,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.4961,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.29532297486584574,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.4999,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.2729508614688553,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.5122,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.2857909317883689,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.5048,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.2961890647983563,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.4939,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.28735786610867087,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.499,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.27694782034879545,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.5081,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.25479656635448195,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.5058,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.2492268420109934,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.5137,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.2855308009526863,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.4847,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.27469752228062916,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.5031,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.2931974789071143,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.5071,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.2917484387634891,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.4975,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.27194186899574063,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.4891,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.3018099718373258,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.5069,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.2658705033804946,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.5083,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.25851898420041136,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.5151,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.27280363349882675,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.5029,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.24647769061907207,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.5002,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.24275125352132976,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.4972,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.25552436103866016,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.5,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.2693255039106339,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.4973,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.25524897135871766,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.4995,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.25638840960661385,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.5032,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.24770822918601107,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.4961,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.2383793095925144,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.4989,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.23199386582422438,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.4976,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.4549492764323232,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.4338,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.3525896459959793,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.4306,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.6160749358224051,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.4413,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.4941909886947903,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.444,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.44752307827684595,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.4261,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.38340204198105277,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.4387,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.38063979400256015,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.4282,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.40587176821080617,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.4413,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.37366332209220066,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.4299,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.37871858582890827,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.4466,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.3897611688618514,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.4462,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.351096388839258,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.4364,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.35735248981549206,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.4195,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.3845287838131827,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.4367,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.3222819381563978,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.4237,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.3546159626542012,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.4293,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.3087483254176499,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.4279,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.35177032984167617,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.4421,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.3449264121608564,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.448,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.29532031073152193,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.4336,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.4739365347175253,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.4482,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.29487669259807187,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.4361,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.29074539954005646,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.4305,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.28521792303001053,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.428,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.27844375884327127,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.4233,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.29108648120094727,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.4183,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.2950681753053369,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.4292,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.27694681718835523,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.4442,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.2779323617445222,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.4348,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 1.034540295058086,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.4489,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.26434400961112486,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.4385,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.26645880588968096,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.4347,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.26866597085244065,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.4446,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.26933500446674713,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.443,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.3035048384396216,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.4342,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.28796243341084027,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.4333,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.26994898766942216,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.4227,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.2537967232757423,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.4388,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.273300840164435,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.4362,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.25723603157737307,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.4429,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.2371137466081644,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.4539,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.25773164392227466,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.4312,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.24619292197839296,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.4358,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.24270414445597738,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.436,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.2373906574686306,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.443,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.23518559870194436,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.4393,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.25147657512942106,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.4321,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.24764316576748807,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.4425,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.24143167245841599,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.424,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.28788222991476625,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.4429,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.27437927352820235,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.4325,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.2523998280568165,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.4333,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.27973257068591756,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.4366,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.25883869394821846,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.4288,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.2660425467320997,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.4311,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.2695188472096521,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.4361,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.23652288864024354,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.4405,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.37199375217226766,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.45,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.2587307273156187,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.4324,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.24296551422221718,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.4352,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.2556554776601863,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.4445,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.253598117464109,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.4484,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.27684113786015296,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.4295,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.28581534779993756,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.4474,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.27701890365730286,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.4173,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.2509419282676904,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.4241,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.23177237642686954,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.4542,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.23751787710506744,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.4331,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.26442213699961925,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.4431,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.23813973197648441,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.4339,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.26942421340230815,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.4432,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.252717616481509,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.4466,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.27614879678866794,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.444,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.2524690392599185,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.4374,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.2490074778580256,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.4383,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.2618891584656124,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.4205,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.29338043764735555,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.4393,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.2607656210149901,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.4162,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.25265457529833923,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.434,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.28243124877888,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.4355,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.7574852551458602,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.4553,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.30652632379891537,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.4404,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.2405727229082847,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.4268,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.2538552701628543,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.4323,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.3086196008617793,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.441,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.2584350293582898,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.436,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.28916536894802136,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.4419,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.26773215118217697,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.4235,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.24846389182907413,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.4527,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.2578955863480322,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.4267,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.25054192166040434,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.4362,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 2.5534022050870977,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.4406,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.3034142014729515,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.4418,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.24161057598869343,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.4409,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.26060500238700135,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.4482,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.2944187094036855,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.435,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.2422747326499594,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.4424,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.2961983306763831,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.4539,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.2555653003426344,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.4459,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.23977564597189316,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.4399,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.25146683600855746,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.4088,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.2449840103265125,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.4227,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.23570777059849385,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.4299,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.23792812826855325,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.4438,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.2507642815840023,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.4492,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.246944309135122,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.4291,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.2482475138823122,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.4419,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.25338556717628014,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.4356,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.22907806742644526,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.4368,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.23725546246835316,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.4434,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.22139369474797893,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.4368,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.4013425135090327,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.425,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.24661719862235446,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.4362,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.23469816165195315,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.4243,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.24749752203149814,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.43,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.24349814808052642,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.4347,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.2314958698840421,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.4276,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.27231567730528916,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.4559,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.2368172164464056,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.4404,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.2393136204547464,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.4381,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.24191801716174538,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.4515,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.22933035596818546,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.4264,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.22264598015980047,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.4371,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.23349235472898686,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.438,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.22953727726557885,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.4322,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.24886003187104666,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.4301,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.22887176401589157,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.4302,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.22913056996260886,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.4513,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.2355979509008494,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.4269,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.2246245573318474,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.437,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.23668842093072373,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.4385,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.23800026857124615,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.4483,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.2392647928985238,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.4417,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.2402426293546199,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.4338,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 1.5394046312915908,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.4479,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.2525028625940517,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.4502,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.24466117739824841,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.4445,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.24079776101869307,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.4321,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.22565525162630057,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.4375,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.24894310695530456,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.4375,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.24334039861062023,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.445,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.2214958994490029,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.4327,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.23097331596474646,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.4284,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.23589749832716314,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.4329,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.21734323727269694,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.4301,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.2226191584652408,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.4405,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.22427369086212298,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.4379,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.22666507960377053,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.445,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.227795492189423,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.4381,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.22696219135677753,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.4463,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.2331570777157063,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.4398,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.22173372520401444,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.4383,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.23368246041349477,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.4428,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.21730867978252746,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.4358,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.22984060886784605,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.4386,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.311034543828598,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.4386,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.22786301247424717,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.4384,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.22831088878706535,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.4265,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.24302177669636077,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.4297,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.2237378997214366,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.4384,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.22960998438812952,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.4179,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.22711445655053095,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.4412,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.23211966893664676,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.4342,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.22124250384305288,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.432,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.22885680644110065,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.4562,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.23230388012442238,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.4401,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.2238732930419573,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.4233,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.23117331187200874,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.4467,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.2558591563249465,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.4303,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.2375561314115263,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.4277,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.23495720348349775,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.4202,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.22345403541419995,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.4423,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.22000230146677374,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.4209,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.23369351383257148,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.4247,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.21907051175868175,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.4308,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.22830229608577793,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.4458,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.2262080975617713,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.4215,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.22217754616706595,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.4488,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.2230667595035534,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.4379,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.22178170457630153,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.4493,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.22301982568562886,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.4277,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.21028523145373446,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.4408,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.21043774078811212,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.4511,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.21562871706453612,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.4334,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.21406853278582252,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.4377,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.21349366173178194,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.4435,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.21131326620244678,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.4289,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.2371815796763936,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.4493,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.22669623253242235,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.4248,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.20990447390883346,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.4284,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.23605881342933543,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.4286,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.23826478048967256,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.4491,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.20967011926048335,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.4403,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.22984565817751001,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.4342,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.23452674132472848,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.4319,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.22102522911852296,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.4319,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.22585626161977926,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.4389,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.23673533584675213,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.4217,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.24852576219339306,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.4388,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.2187605919057179,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.4373,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.2242676338565513,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.432,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.24386212655732328,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.4385,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.23041187708605448,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.4417,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.21245103834347429,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.4446,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.2187543590941638,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.4189,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.257545014220748,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.4275,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.22214173727914363,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.4247,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.21843315152060894,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.4309,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.2186365824314662,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.4345,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.23580389579155175,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.4359,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.22055584216074833,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.4228,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.21687651602764477,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.4499,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.23143419109290905,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.4449,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.2149151377640905,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.4525,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.22187080286454844,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.448,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.22891287709752128,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.4513,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.22183328535604988,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.4229,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.21154258980530824,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.4131,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.21903755052185356,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.431,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.23197391012866495,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.4342,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.21450776669923316,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.4356,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.2118076504299908,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.4421,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.22850712461605185,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.4387,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.2215119020264137,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.4395,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.2156519491828405,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.439,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.22401967006804527,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.4374,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.22353521844606097,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.432,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.22689284969318887,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.4197,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.21250880775824993,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.4393,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.2133531085020732,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.4663,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.27170640084831,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.4311,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.2196847006566226,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.4391,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.2090356930858301,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.4343,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.2278550626385792,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.4375,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.21905174234975572,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.4376,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.21464095222197813,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.4233,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.21662440420751689,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.4297,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.223582534977822,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.4417,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.21940945688314473,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.427,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.21666705536115127,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.4429,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.2058414256186936,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.4347,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.2296303613705075,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.4448,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.22146379170289573,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.4275,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.21481395036255987,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.4159,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.23314994556995455,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.4397,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.2281711505580352,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.4375,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.21620634590956703,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.4266,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.3645697328522445,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.4169,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.303384805692071,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.3941,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.2326364496979524,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.3861,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.3528160418711453,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.3914,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.3553666123878872,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.3945,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.2552156447825777,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.3894,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.306050675509667,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.399,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.3346882602251698,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.3929,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.2994290480255684,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.3802,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.23173235455856236,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.3942,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.277183504129002,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.3983,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.30448508843952504,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.3714,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.29752244611588735,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.3949,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.22802257175171856,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.3782,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.25499963080649196,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.3744,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.26647927837329877,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.4013,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.24829328467233105,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.3866,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.23082631148528623,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.3952,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.24255760872037546,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.3922,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.2631460921741403,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.3927,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.23612069861634127,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.3906,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.22387892243099936,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.3858,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.23572225052745668,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.3807,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.22889809900490826,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.3859,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.22323360525411262,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.3903,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.2130569783705088,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.3771,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.22115997142165847,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.3868,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.23471406571681616,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.4046,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.218312868379773,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.3849,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.20726852753984387,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.3955,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.2147468612622889,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.4057,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.2306026766184789,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.3927,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.22891101956860604,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.3947,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.21757993810175577,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.3928,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.21149335154128832,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.3955,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.22727255102350136,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.3863,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.22686527985536264,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.3899,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.20456244397535933,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.3929,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.21373039064728874,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.3708,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.21558378224056116,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.3845,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.21769141926309832,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.3913,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.21718368849393624,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.3873,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.21291725806552492,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.3706,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.21940837789712783,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.3923,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.22294252662964792,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.3851,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.21871467807329753,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.4079,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.20924053683774407,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.4042,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.20564691453878495,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.3984,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.21361580713582543,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.3983,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.2109825787158314,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.3978,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.21031209356961347,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.3833,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.20825143792134881,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.3787,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.21402146351802143,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.3896,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.2070096445108216,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.3888,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.21807374140886743,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.4037,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.2174113329910912,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.3694,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.20968828134324183,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.3924,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.20632375804039094,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.4059,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.20897925579956791,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.3801,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.20994611659677126,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.3864,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.19950252474212943,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.3901,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.20568036210498006,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.386,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.20219570296307465,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.395,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.2020132514558141,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.3892,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.20485428735970226,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.3879,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.20139128035641227,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.3893,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.2056533798648585,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.3878,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.20410434015650364,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.4063,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.21525495148330803,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.3904,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.2070303905682487,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.3609,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.21023310236519538,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.4032,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.20378758636105823,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.3879,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.20079962734404228,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.3996,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.19690515748586496,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.4003,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.21289774722600854,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.3839,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.20586354791357922,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.3792,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.2117900144666368,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.4097,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.20359182167917542,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.4013,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.2009121461608691,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.3888,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.1978560158525822,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.3966,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.2086375611384555,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.3949,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.19758279947319016,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.3757,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.20065000672563763,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.3997,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.21090007785394366,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.3851,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.20057076553578942,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.4026,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.20160920082290773,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.3935,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.1932582729494484,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.3864,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.20214843732991158,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.3865,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.2014010916174826,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.4087,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.2078311015481782,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.378,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.2009904449091929,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.3968,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.19672207849414639,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.4138,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.2046251284568484,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.3729,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.20246766927592008,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.3849,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.20246520855768876,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.3993,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.20332147508467865,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.3926,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.20145404757744198,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.3954,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.20295678271668938,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.403,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.20663337774060367,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.3956,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.19397283850305297,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.4058,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.20338166542953562,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.395,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.2130674365904382,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.3864,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.20899142101323065,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.3802,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.20591773957273846,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.4019,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.19627030351113556,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.3994,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.20609436004544912,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.3977,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.19533497835790864,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.3865,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.2249249531710021,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.4064,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.19676598100319614,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.3973,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.19301505933518232,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.4059,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.19506910658478008,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.3884,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.20655305790642753,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.3961,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.20784336423108524,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.3955,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.2028161516569148,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.3894,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.19877752155488124,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.3973,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.2094759513914305,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.3726,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.20182229968241588,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.3923,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.1982036477170145,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.4025,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.198549897238847,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.4047,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.205129062413334,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.3829,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.20054373844555343,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.3971,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.1902566306712916,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.4073,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.20496331283715583,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.3821,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.1934046545945333,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.4044,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.19333695486224456,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.389,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.19797061695598966,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.3854,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.20144184066694248,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.3916,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.19947613825858204,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.3894,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.20073949565208682,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.3963,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.20275106780256594,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.3932,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.20147108257743523,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.3919,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.20027130849996447,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.4038,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.19069350450753872,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.4024,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.2057400454452374,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.3997,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.19709525400653244,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.3846,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.19224721979198386,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.409,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.20363233792218258,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.3807,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.19788384874922693,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.4096,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.19666432161271627,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.408,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.19944550350148602,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.3897,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.20549998985434773,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.3848,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.20074590839691195,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.3798,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.20132352869313816,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.3926,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.19948919533369275,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.3729,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.19286675864291108,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.4,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.19983603755359783,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.374,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.19691118554023743,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.3905,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.1920354910308016,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.4041,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.20019980616782584,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.3975,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.19906367697004387,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.3964,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.19363331190828006,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.395,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.19681182266290648,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.3908,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.19448909043286608,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.4032,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.1948017232318065,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.3757,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.19431731710952535,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.3922,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.19429159083872316,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.3949,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.20278447078887915,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.4062,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.19462876435327534,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.3982,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.20997022085563977,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.3879,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.1860294632501331,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.3951,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.19182966742908775,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.3988,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.19930600536524484,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.4051,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.19893226112728948,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.3806,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.20742809519964822,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.3884,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.203046290078911,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.3884,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.19437539606449492,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.3871,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.19306736804740027,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.4052,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.1959728357781069,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.3899,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.20572429198953948,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.4021,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.20371680582559434,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.3882,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.19381366078090298,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.3975,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.2052959158224277,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.403,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.19577964772097503,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.3924,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.19689916827572418,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.3856,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.20470686825764417,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.4016,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.19169679289075553,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.3944,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.1948555408628416,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.3812,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.19968343823819482,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.3791,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.20140080852087977,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.4021,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.19401594380734438,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.3904,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.20682806841579995,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.3749,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.19656419031547875,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.3834,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.2045836259273087,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.3927,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.19900591594586198,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.3999,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.1919316957731677,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.3836,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.18881086078526318,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.4047,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.2021072987551945,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.3808,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.20145086440564283,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.3717,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.1937889147630969,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.4009,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.290856023417042,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.3837,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.19840861427301332,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.3947,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.19381831355622653,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.3895,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.21806353600067646,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.3865,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.18982310940307565,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.3835,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.20235423555846369,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.3907,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.19626481452308595,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.375,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.19529490029080693,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.3939,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.19807542861045052,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.3852,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.19184718982596627,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.4148,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.19501918999545262,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.3788,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.18989841305566585,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.4009,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.18871807458163078,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.3991,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.19533235727786202,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.395,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.19432387041275234,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.3985,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.19443046267576075,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.3928,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.1968747540176252,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.3848,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.18964093250696032,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.3884,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.19256896408813073,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.3955,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.19573569275108607,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.383,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.19076626993509263,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.3824,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.19158035599115794,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.4012,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.1962722023088182,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.3866,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.19284362157198642,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.3836,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.19003837279780894,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.3919,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.19810410827397149,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.3838,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.19713321303901277,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.3835,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.19675194138888485,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.388,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.19641472175782282,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.4023,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.20541125315919948,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.3774,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.9761465453831671,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.3983,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.19464589815099798,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.3881,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.19602819380679581,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.3723,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.19762319479384763,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.3845,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.19358737170787194,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.398,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.1968927639129562,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.3856,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.1934875956088849,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.3974,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.19495317997697748,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.3999,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.20372575079257027,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.3882,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.1993341382356279,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.3844,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.19139987815235301,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.4133,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.19469334569414654,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.3856,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.23895458309831047,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.4046,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.19938877022432186,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.4056,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.1952087125219556,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.3851,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.19383558234786236,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.3919,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.19724413433992968,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.3736,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.19432074028135096,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.3851,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.20959560416696332,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.3951,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.19681985030996507,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.3858,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.19341568812093787,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.3933,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.19693086432979742,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.3841,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.19428951112181073,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.3945,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.19188574055819477,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.3899,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.19993710418507693,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.3851,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.1996529620798354,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.382,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.19617265566606604,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.3921,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.1859316107864917,
+      "learning_rate": 0.0,
+      "loss": 0.3999,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 5.150555634529206e+18,
+      "train_loss": 0.5203020626475454,
+      "train_runtime": 29473.6788,
+      "train_samples_per_second": 5.361,
+      "train_steps_per_second": 0.042
+    }
+  ],
+  "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": 5.150555634529206e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}