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": 6.54297388429011,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.044,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.478513988373168,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.0408,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.3610513361699885,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.0225,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.2955090048609685,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.0174,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 5.945225576567738,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.0137,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.858841560164762,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.0298,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.454544083250418,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.9914,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.059627022201706,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.9659,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.723970270751247,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.9408,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 2.6123060825169677,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.9447,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 2.4653357736545356,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.9207,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 3.860118286501855,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.9419,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.150135930213129,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.9344,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 3.964594045963413,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.9058,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 3.6168150629832456,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.9039,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 2.713333885588272,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.8675,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.500370229159504,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.869,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.021881450644816,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.8337,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 1.4866676248565023,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.8223,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.1013943692201984,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.8236,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.048614351126827,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.8062,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.21457547614379,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.8014,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.0885572312419416,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.7854,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 0.9626403313434233,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.7921,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 0.9502695581790847,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.7947,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 0.9250738082111004,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.7807,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 0.8245275747538591,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.7455,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 0.8125580689358093,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.7402,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 0.6909565434474426,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.7531,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.8389981885976906,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.7263,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 0.7042647762631075,
+      "learning_rate": 1e-05,
+      "loss": 0.7475,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 0.6413191770281536,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.7278,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.659280418237527,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.7316,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.6422877628806536,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.7259,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.6346214673275423,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.7216,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.5895210536348506,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.7231,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.5444794918901437,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.7153,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.5721288517679979,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.7201,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.547837185733015,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.7011,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.6296970753706612,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.7036,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.5774295480539557,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.6887,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.6025451229064763,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.6923,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.5863797409453076,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.6741,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.7441481941816019,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.7046,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.4875744284631483,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.7011,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.5444678501059769,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.6905,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.5045829702371525,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.6875,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.48863847221953133,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.6971,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.4662693692448641,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.704,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.4967851425188479,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.6897,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.5529722383310639,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.6944,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.5961058724307372,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.6676,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.49408967892500893,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.682,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.4631279394562096,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.6784,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.507857998151672,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.6699,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 0.6097904940682387,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.6567,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.5084910854910024,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.6875,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.5821541461879335,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.6685,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.5680950646794105,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.6647,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.8603278170673331,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.6638,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.5997036815163177,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.6489,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.7741524840542362,
+      "learning_rate": 2e-05,
+      "loss": 0.6704,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.8797123761065213,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.6853,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.7538115404547725,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.6652,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.7753370442093064,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.6664,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.6198892344645441,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.6552,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.5738249597344739,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.6657,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.7270729983708695,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.6634,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 1.0586327279226564,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.6522,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.8685168062707064,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.658,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.9735507887402329,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.676,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.9926370358247442,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.6502,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.9887632710316665,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.6655,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 1.0563530071929152,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.6669,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 1.0117154519424882,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.6541,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 1.2754763903646331,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.656,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.6308778579340987,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.665,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.5884483786543485,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.6383,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.8221914811490717,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.6579,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 1.1695209065315333,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.646,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 1.1015544099097276,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.6503,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.8605832690656502,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.6635,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.802406839805688,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.6404,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 1.015883125620469,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.6484,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 1.1570932281749249,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.667,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.8030310520865008,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.6546,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.7332181663415983,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.6581,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 1.0270876629408459,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.6365,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 1.233311348120077,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.6561,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.9268048902845496,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.6521,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.3469342702801297,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.6287,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 1.1069129852354542,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.6368,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 1.0186434558083182,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.6411,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.9443118111539656,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.6378,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 1.218791150023356,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.6457,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.7236852578402513,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.6464,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 1.3173536020974541,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.6494,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.7178070962710712,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.64,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.9032776699858382,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.6432,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.1389256968080086,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.6499,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 1.0810526120853239,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.6455,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 1.1306505408099508,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.6546,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 1.191617051146834,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.6472,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 1.1539105915168222,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.6457,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 0.7296250068429828,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.6319,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 1.229711339872855,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.623,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 1.0043050217065117,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.6422,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 1.04812875648899,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.6506,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 1.1267373176357938,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.6379,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 1.439175980804842,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.6564,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 1.5145305286668922,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.6449,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.7841264982420506,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.6358,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.600117759452762,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.6223,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.8855291346079193,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.63,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 1.0165096578541064,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.6464,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 1.104807565056701,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.6325,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 1.5976098156431662,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.6155,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.9279458212047694,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.6466,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 1.238647581098739,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.6425,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 1.237269833113962,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.6338,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 1.309483093779508,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.6431,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 0.8455657806116739,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.6306,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 1.69694303530633,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.6613,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 1.2488984938751513,
+      "learning_rate": 4e-05,
+      "loss": 0.6408,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 1.3839382245610032,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.6371,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 1.432391652054566,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.6366,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 1.1884786750128176,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.633,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.955752782948825,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.6628,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 1.615005270995625,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.6496,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 1.5697447390191595,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.6418,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 1.3224700084344776,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.6383,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 1.3730879049936944,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.6396,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 1.0068635066854532,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.6221,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 1.6337158330344492,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.6384,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 1.107548024237694,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.6473,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 1.564721425912208,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.6345,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 1.1615861716448843,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.6406,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 1.4902301543405898,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.6369,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 1.2622149597337473,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.6109,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 1.2016939366759645,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.6262,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 1.096030086291758,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.611,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 1.095656563574066,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.632,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.9568523064216561,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.6269,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.9613675950133961,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.6282,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.9224396498868845,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.624,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.844780905851037,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.6355,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.8936758793340379,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.6013,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.944873001123428,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.6305,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.9243912829824783,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.6264,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.9700174672268779,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.6316,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.8056855975904442,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.6357,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.7201812652851558,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.6071,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.680431796712001,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.6298,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.6953227911876712,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.6132,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.7723147573169493,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.6153,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.6557969688759357,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.6274,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.7637795461755923,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.6165,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.6887946962461379,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.6166,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.7217087442032908,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.5964,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.6861714306187402,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.6184,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.7303687175710094,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.6372,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.8006779772365897,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.6246,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.721661478357974,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.6217,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.7244833048370023,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.6221,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 1.1093300502366743,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.6349,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.9817181075789236,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.6376,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.6010458291147878,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.6175,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.6329451021726243,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.6252,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.5472569178483336,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.6119,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.6290094230514961,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.6212,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.691424230934777,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.617,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.9595461792411586,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.6142,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 1.0678713942561384,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.6197,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.8815951587968242,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.6243,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.6963547650985056,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.6274,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.6208694853660576,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.6142,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.9107419090240193,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.622,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 1.0314675661284094,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.6249,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.8024226505128518,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.6265,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.6798512467143558,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.6226,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.6240101159235505,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.6188,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.8097331534572286,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.6081,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.6501929387030264,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.6342,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.5849878220261902,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.6224,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.7689440084751216,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.6059,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.6950997160079558,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.6202,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.6333715880182729,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.6235,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.8485496392484501,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.6034,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.8620956107017369,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.6339,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.7668115857878265,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.6185,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.596082556810814,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.6184,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.5765593325055318,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.6126,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.6041252517498985,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.6131,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.4642394205514685,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.6316,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.5692594370008345,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.5981,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.6664121805867167,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.6337,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.6143642829927998,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.6054,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.5708717811771643,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.6146,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.5271240707910438,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.6255,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.6300190490580725,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.6077,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.587460303417004,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.6042,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.5571421366765608,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.605,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.6151451776295939,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.6082,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.7166276568692412,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.5802,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.7188307467703589,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.602,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.4865210356018533,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.6206,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.5987033985696025,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.6176,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.8587420188365464,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.6032,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.7458031970593563,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.6138,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.4219898538784674,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.6019,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.5438935476883472,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.6015,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.6692531872672267,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.6046,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.560122147206034,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.6102,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.496987551261406,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.6002,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.4279956933491659,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.6189,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.4779929096766363,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.601,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.4593093217496535,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.6074,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.4546835271650199,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.6148,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.4784911413566355,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.6166,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.5227250771241043,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.6081,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.5989407950927542,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.6044,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.5344537441027372,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.6158,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.4851655610844948,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.5965,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.49889178953913355,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.6174,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.46028926058205594,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.6194,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.4850116099624815,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.6079,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.6506640798684126,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.61,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.7048554023076098,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.6068,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.5230185108485159,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.6094,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.4728500625392583,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.62,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.46495912732269423,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.6128,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.4596034128381929,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.6068,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.4769445994613367,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.6076,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.5411141735576702,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.5979,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.5094721966306482,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.6111,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.6099788182100713,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.6091,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.8748294035198233,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.6064,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.7847544015763178,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.5901,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.47414073130536294,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.5941,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.545043611564174,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.6155,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.5278042738414513,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.603,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.6714838139166309,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.5991,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.6472453500503696,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.6175,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.6236518750224544,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.6167,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.5004386827215208,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.5889,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.537424469264863,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.605,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5478603762798139,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.6108,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.7046123221518731,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.5623,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.61518598774554,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.5497,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.5890410711037817,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.5611,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.6614089020743565,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.5605,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.5994859688185743,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.5655,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.6419298321365445,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.5552,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.6477859672615357,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.5564,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.6064431529729457,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.5549,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.6101759687030778,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.5742,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.6177840898639708,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.5768,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.6019996846442438,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.5598,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.6194883693618679,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.5641,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.5089678182485745,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.5847,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.49470029980007957,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.5702,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.6391558270988463,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.5537,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.7205900351112345,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.5532,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.6800631267707413,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.5624,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.668448261320995,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.5608,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.6279616914991825,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.5636,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.5985195952253778,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.5631,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.4448628584682874,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.5456,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.5849466413924853,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.5581,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.6526888582353247,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.5641,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.5780407671942869,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.5629,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.4529392210978806,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.5628,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.46041529770033124,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.569,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.52333393709489,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.559,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.5672398818635047,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.5401,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.4579931596391269,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.5422,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.5286768622210627,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.5516,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.6299432845331695,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.5711,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.6689740281813948,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.5492,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.4306151041135981,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.5562,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.6335167672863588,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.5541,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.5011827325279985,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.555,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.56023835448926,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.5526,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.6194532448140461,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.5729,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.7323663819965754,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.5648,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.629617149867259,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.5496,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.5014542678506709,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.5486,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.6116944586735217,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.5574,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.6342223593741498,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.5616,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.5448400755457774,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.544,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.5643561645868612,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.5472,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.5094422548370747,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.5699,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.5864289818162597,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.554,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.654779734909922,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.5588,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.7518928276094192,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.5595,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.7063452022981225,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.5603,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.4741534719105488,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.5656,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.5056661076268703,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.5496,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.8142191683529203,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.564,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.7873201744316574,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.5516,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.517557061967328,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.5327,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.4695884573712099,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.562,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.5568282764875672,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.5599,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.5052440113433488,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.5554,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.5506281318308207,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.5616,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.5731357225855426,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.5643,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.39494866920596383,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.5767,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.43331627753880003,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.554,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.475125777572782,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.5466,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.47911928794883557,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.5688,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.44180990883592786,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.5463,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.5356876499017008,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.5635,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.5651877344711482,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.5425,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.570653158838979,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.5648,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.4717415085800711,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.5659,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.4304821284351938,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.5606,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.4993459340531862,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.5573,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.4060543153949394,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.5301,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.44074611886325016,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.5616,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.47585006574675026,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.5581,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.45003306444648256,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.5674,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.46774381265356557,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.5448,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.3798957879872776,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.5514,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.44243833650994063,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.5512,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.49352491180473823,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.5507,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.4943590454311403,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.5555,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.3955343567049488,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.5559,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.3972927536930408,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.5619,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.3766865979997967,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.5578,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.5183568774323366,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.5522,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.43301704541143465,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.5536,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.3604612952706816,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.5623,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.3922070020496394,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.5429,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.40829772927362223,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.5582,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.3525976955603173,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.5555,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.34519429754659076,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.5547,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.4266140428598264,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.5752,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.39346483803517635,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.5604,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.4219953931321321,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.545,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.43527413905679974,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.5578,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.45593117222766694,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.5467,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.4867479981373752,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.5604,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.47078699352691467,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.5611,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.39594813610849805,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.5598,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.39660252718958044,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.5619,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.39520687991267855,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.5626,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.40189741200601775,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.5689,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.4689796183702496,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.5559,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.5562573772252011,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.5583,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.5187437082568627,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.5504,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.47873375068849705,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.567,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.4437290627111576,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.5619,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.40966843364909145,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.5527,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.4226451765217786,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.5586,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.43033293167336834,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.5494,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.37827563495463273,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.5583,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.4651597214754255,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.5475,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.46220110207530846,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5585,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.44300705467066853,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.5525,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.3445499411056897,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.5594,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.3385489583170658,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.5655,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.4046552493306492,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.5433,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.3789275871242677,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.5587,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.36761942864178115,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.5508,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.38207939199562724,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.5486,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.5123353396263114,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.543,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.3912235129822814,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.5553,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.4691092726867642,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.5539,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.5553331620468946,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.5512,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.47144082809631027,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.5554,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.42981196921718906,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.5463,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.4291257771801819,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.5461,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.3944926767304932,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.5375,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.507401421485124,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.5551,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.4408842903820842,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.5523,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.4763144988887003,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.5515,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.42177186507841935,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.5488,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.37649942102975814,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.5581,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.3639475036992518,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.5516,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.3555240324699287,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.5456,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.5678896129921959,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.5537,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.5792921398702016,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.5525,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 0.3813126953863021,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.5483,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.5364329409216799,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.5591,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.5154902450202639,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.5412,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.4399355637985282,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.5412,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.3826069925925746,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.5626,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.5221771442372606,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.5601,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.5238363736442909,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.5346,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.4862467769306201,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.5591,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.43050624779057683,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.5547,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.5351502740941674,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.5607,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.5129229059670624,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.5398,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.46822654913111134,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.5432,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.3832074281180265,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.5568,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.5386678963468294,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.5573,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.4259779774339028,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.5636,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.4756813343945755,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.5665,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.5424691346016505,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.5363,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.47828086636596173,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.5468,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.475791895505287,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.561,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.4946528512863168,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.5437,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.5470738457418449,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.5536,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.4129508054890389,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.5549,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.5155606646862104,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.5468,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.46270479703313716,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.5601,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.5291845034723518,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.5593,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.560513998193393,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.5625,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.41339561961256105,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.5422,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.6025947096465057,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.5516,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.5063205514772435,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.5447,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.4150339376881955,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.5537,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.4878380261281231,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.5459,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.365327418959227,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.5514,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.41551490419956205,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.5427,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.3880375614268584,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.5415,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.42928702846439915,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.5559,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.40532547722660783,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.5468,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.4336703732427423,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.5466,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.43101326817185526,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.5436,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.4188742024425209,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.5591,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.380321389389417,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.5404,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.44806719689352537,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.5379,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.4137765023667395,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.5458,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.3946624666490224,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.5586,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.504535355687136,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5479,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.34735787083136627,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.5434,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.4452953653220499,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.5643,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.3453910608132859,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.544,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.40898276552137736,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.5503,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.35961813908145773,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.5681,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.3852946714692432,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.5509,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.4190285725417996,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.5432,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.4073573397422204,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.5419,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.4297468010529677,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.5604,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.39622437258662474,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.5605,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.4625150303117316,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.5495,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.4093690263796008,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.5505,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.4628029201035852,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.5519,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.5342396924776115,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.5621,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.43871763895526616,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.5433,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.42823961220853896,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.5518,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.5038415128517514,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.5487,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.3557284763385091,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.5638,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.42956731700581674,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.5459,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.359963170704742,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.5489,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.3734490029507308,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.5505,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.3483188359541883,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.556,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.3769316402529617,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.5499,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.35058051854897465,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.5455,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.4073107725394978,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.5551,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.42364326517312306,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.5562,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.45796821684224953,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.5439,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.4202799278930839,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.5501,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.5060512887416367,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.5432,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.462057534063152,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.5397,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.3667787525094176,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.555,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.4465430361129101,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.5714,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.4836127101881475,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.5586,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.3982344060039588,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.5283,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.49300808934373674,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.5549,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.3947163341167545,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.5503,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.5237897285201443,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.5355,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.3856630056349095,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.5523,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.5370661936794425,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.5504,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.36844098761785576,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.5525,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.4639538473011158,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.5385,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.3563624236485184,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.5435,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.45375902483643593,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.5463,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.3610329880872194,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.5394,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.4450733743087252,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.5373,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.40103128012148354,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.5553,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.32049779593255207,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.5434,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.38720707520484543,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.5428,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.3800308119410429,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.5382,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.3557690282617048,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.56,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.4381275504104802,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.5557,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.41117808495059416,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.5602,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.373964484055866,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.5467,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.3815558087166116,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.5591,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.3832810791084611,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.5463,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.4204179742136049,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.5477,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.4231127809067307,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.5436,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.394646041606415,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.5622,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.497719837277611,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.5503,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.3539787911518485,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.5524,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.3884772303971677,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.539,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.40980751070663096,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.5414,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.44629014008978984,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.5422,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.3946809915589734,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.5432,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.4588906066163366,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.5503,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.3882186666670153,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.5497,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.5542543364378265,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.5418,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.4287454642497271,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.5399,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.554754821243918,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.4861,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.545743420620779,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.4798,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 0.5658088126946466,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.4886,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.5741168057832691,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.4928,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.4857902115978857,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.4834,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.4231140337265413,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.4906,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.45886484437284697,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.4918,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.5039770350000454,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.4871,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.3910102412189536,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.4835,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.4593211077510333,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.4881,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.40164628364638705,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.4742,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.519363209607065,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.4766,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.39080725739550554,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.487,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.47833247962863523,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.4819,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.4164824182227974,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.4898,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.39298997979144473,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.4934,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.5135196391843505,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.4803,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.42336030568543914,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.4869,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.3776826823188418,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.481,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.4907600030166504,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.4784,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.4516937305380755,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.4878,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.46626971250083316,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.4825,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.39262898930963,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.4924,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.5287624757826767,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.4621,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.4106545123531904,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.48,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.38623216678880484,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.4777,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.33841564737684504,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.4827,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.33770922196783465,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.4748,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.3441872383705684,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.4824,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.3158690922878844,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.4763,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.3900232026591817,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.476,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.3440166461285994,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.4864,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.32496335177792574,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.4916,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.3889643920926402,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.4867,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.33245489928760064,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.4721,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.32978550658973554,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.4829,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.37970777536838607,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.4872,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.3544849795017511,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.4784,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.366625903485653,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.4872,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.4533440041937536,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.4736,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.3535113450872792,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.4854,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.40009276802692045,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.4864,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.36845378847716465,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.4865,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.36874298145395634,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.4928,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.3534268667774682,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.4692,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.3059150763513404,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.4963,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.34322225254725075,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.4817,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.3016806126963221,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.4948,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.34830408574510163,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.4901,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.3002222672302038,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.4859,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.34613349312306346,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.4852,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.3368440590835921,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.4833,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.3192159735274027,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.4751,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.3873351479378658,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.4872,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.33020774936413905,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.486,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.3428424227790982,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.4719,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.4103454835479282,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.4822,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.39788360439279846,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.4815,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.38523513994464614,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.4917,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.37511859284614313,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.4809,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.4011928425105952,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.4908,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.40698173519375175,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.482,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.34672879782302585,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.4965,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.40730650996637197,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.4749,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.3848639940460819,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.4722,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.3884680392126513,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.4879,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.611654109595553,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.4853,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.47264300890462974,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.4787,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.36305407317029115,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.4847,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.36869432641158106,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.4793,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.400887758807889,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.4904,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.2991437831998725,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.4711,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.3798119268268107,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.4729,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.2940185776296549,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.4819,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.377709444543999,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.4911,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.3433360322270319,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.4763,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.3900329475245157,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.4819,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.36117465810987975,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.5024,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.3516883612654434,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.4701,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.39486776802356044,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.4869,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.33341260907955983,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.4812,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.3603590063532126,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.4973,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.29327829460822497,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.4903,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.34917182212492864,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.4825,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.29864888539213574,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.4893,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.34838357757524996,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.4946,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.3409963897729681,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.4815,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.3940998396056094,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.4714,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.318835514250764,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.4884,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 1.0138623547800356,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.4886,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.30984497886497336,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.4767,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.3417528249990639,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.4805,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.3829421843575417,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.4825,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.2951461776946434,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.4823,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.34619013720658753,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.4909,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.32744305614188907,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.4895,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.38469233092353694,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.4926,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.34786250193501683,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.4849,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.37721994352093374,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.4948,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.369397668470289,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.4856,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.4029249926222225,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.4865,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.3764379097541486,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.4989,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.381631675098176,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.4835,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.3932170107517469,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.4994,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.3987395072322803,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.4947,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.3242161559561427,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.4878,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.422056215823352,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.4813,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.35323337021296614,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.4902,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.3840132026227925,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.4919,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.3308291880571455,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.4905,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.315733738546684,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.4978,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.32234784259837745,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.4937,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.27912425443411537,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.4822,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.34215210670057367,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.4902,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.32897478557759163,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.4838,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.3266221517266906,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.4795,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 0.33333717523736706,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.4848,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.3355028645217648,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.4879,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 1.1107415879069613,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.4988,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.39238417941294607,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.4864,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.33212154228201796,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.4931,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.32623319276649676,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.4824,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.3908243563363243,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.4942,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.31756800050887807,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.4858,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 1.2774339844410285,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.4841,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 3.3776459395825333,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.5207,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.6812189048715483,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.4795,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.6357384907166447,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.4716,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.3924474223211006,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.4745,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.38505161385940545,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.489,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.3350230430784962,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.4981,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.3843642188527194,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.4786,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.47771342764507563,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.4873,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.3477122373279191,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.4903,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.3587222793249495,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.4908,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.3340225164758689,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.5028,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.3257239566486329,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.4825,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.372049783159658,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.4748,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.3121377037129043,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.4815,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.3204264462484106,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.487,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.2932943359016784,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.4873,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.36211681347935365,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.4982,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.26192056060027885,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.474,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.34870371649166476,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.4759,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.26713305545866856,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.4872,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.334996017150231,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.4782,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.4702488445744811,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.4818,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.28931925112531254,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.4795,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.34805036163822295,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.4773,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.3086479397183277,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.4828,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.35248037707934216,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.4871,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.29878648585418743,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.4982,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.36387309578549465,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.4912,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.30608748062094754,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.4898,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.3518167048152296,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.4951,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.29338165177131215,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.4924,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.35239631230896823,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.4849,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.30648276329123086,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.5013,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.3595125743726005,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.4855,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.2994203141973345,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.4998,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.34069669353431414,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.4894,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.3174554848113794,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.4808,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.2643867552947516,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.4789,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.31581393656603285,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.4881,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.27993834620292485,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.4867,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.32147975889908836,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.478,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.3047708642102016,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.497,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.3003735197906575,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.4997,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.30138288811809755,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.4847,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.3049949600354941,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.4789,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.30096332017090466,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.4687,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.3395137137112261,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.4837,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.309845878452941,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.4919,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.3132467348184248,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.4848,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.3060043709034966,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.4705,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.2875667715947219,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.4795,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.3057686739502379,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.4959,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.2870517426067134,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.4784,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.3246296342172927,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.4876,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.26714507285508376,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.4743,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.34233050811247273,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.497,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.2992889649467614,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.4908,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.2979214154803331,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.4771,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.30808948001112485,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.4846,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.2972719548676234,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.486,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.35691944901938366,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.4926,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.28683164836568553,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.4833,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.3565652328123348,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.4767,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.3052164542976162,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.4927,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.34455206515987014,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.4899,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.3476876992217214,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.4927,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.334970371151864,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.4901,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.44903668859672413,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.4845,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 3.3924158237229802,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.5757,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.36005174945587,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.4913,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.29975308422364516,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.4749,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.31053279560666486,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.479,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 2.454561682205652,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.52,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.467290123980283,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.4845,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.3340755739681563,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.4939,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.375922178472052,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.4976,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.34129623771736933,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.4803,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.35207165817322655,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.4844,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.34689710450662814,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.4848,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.3226030999063548,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.4923,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.348977317690657,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.4964,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.3080749578839063,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.4785,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.3496148145872439,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.4883,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.33957141992211615,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.4797,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.3210601609740697,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.4873,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.3256572876855761,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.4912,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.3058673969925996,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.4773,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 1.2087659424888628,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.4873,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.32995434107138577,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.4833,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.28668896472730004,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.4811,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.3365882761480398,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.4728,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.32748811156412994,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.4786,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 1.802429027759739,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.5081,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.34081294715223787,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.487,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.30724246852461706,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.4965,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.32735006731767335,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.4937,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.2934831424798933,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.4837,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.29126296223822357,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.4729,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.2982794502700161,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.4859,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.25907925314454905,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.4779,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.3798512663995208,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.4841,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.25414763029242676,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.4892,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.2753743249068604,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.4846,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.262028392364149,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.4844,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.31657117035547006,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.4777,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.25478845018908497,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.4805,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.3050239654768483,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.4944,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.2670227632833275,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.4752,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.27813526986440484,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.4859,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.26048610172603504,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.4934,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.310851163130656,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.4876,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.2633364060530586,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.4752,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 3.0524829097435813,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.5228,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.36346235977118124,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.484,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.2763577679439961,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.492,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.8596416549638942,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.4866,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.27702615916396717,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.4814,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.3548901680930176,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.4823,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.2291806556994912,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.4825,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.3192994037433034,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.4826,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.25186343261457744,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.4882,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.30752215620295303,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.4805,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.3972829437426428,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.4217,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.36191573634600444,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.4244,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.48677108377134326,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.4236,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.43465101932063666,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.4191,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 1.1477566850859422,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.4289,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.4953533779416447,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.4338,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.40570704273859826,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.4289,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.3998171492740688,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.4229,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.44420375649038396,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.4257,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.37592317300326,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.43,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.37646180356983966,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.4166,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.36616331067679453,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.4204,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.4051607579329709,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.4182,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.35480644375001336,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.4194,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.42015325028880524,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.4222,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.3413922710701675,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.4195,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.35003321907674884,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.4144,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.3174951216914049,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.4191,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.3110320032132791,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.4204,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.3081801035972107,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.4113,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.3246555131468392,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.409,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.30895248888103277,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.4268,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.3376217940896481,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.4361,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.29940673275070545,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.418,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.3145795714481807,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.416,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.2705020470127929,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.4118,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 1.7118914576736424,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.4308,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.3510593552138557,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.4262,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.28297887059963356,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.4304,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.3719051118343025,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.4135,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.2649237031336904,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.4281,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.30395842389410616,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.4096,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.2914858460295883,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.4237,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.2812596484136079,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.4234,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.27777914650774477,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.4176,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.3044579668915092,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.423,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.2812216285528743,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.4194,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.2768848852505602,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.4193,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.2740638054582664,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.4102,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.2904555280967881,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.4172,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.24480468033812028,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.4173,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.30712001433556463,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.4149,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.2781769808063252,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.4234,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.2730640070698941,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.4184,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.2633492265854985,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.4236,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.29938632210800675,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.4207,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.2540030416063094,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.4131,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.29928362304929146,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.4195,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.25837385564777376,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.4112,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.270790868199557,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.4111,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.260505410726488,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.4151,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.26112256632135594,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.418,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.2783232190233852,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.4267,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.26847145285435703,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.4203,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.28336708576887426,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.416,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.2825214775825125,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.4293,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.2561877066190111,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.4203,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.2558224385139739,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.4139,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.37980723620697565,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.4247,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 1.8034208527132338,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.4993,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.3459069585105733,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.4263,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.3063324527335935,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.4257,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.28142002097147695,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.425,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.28190734891678915,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.4167,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.23717738042758443,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.4247,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.2675283887588312,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.4218,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.2553702857799187,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.4182,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.25510954698505117,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.421,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.26086985013166836,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.4372,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.24310594577059985,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.418,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.24962479505345242,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.4184,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.2517295374835364,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.4146,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.24798272365154123,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.4212,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.25098072109613045,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.4214,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.2463931922636782,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.4229,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.24100755170346524,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.4233,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.2519723326724292,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.4251,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.6782185112583891,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.4281,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.25396658960467366,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.4219,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.7068778102805425,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.4144,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.27190411576950235,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.426,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.24015296088784194,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.4211,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.26926742309154617,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.4126,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.28402332277060227,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.4139,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.25433686120132576,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.4167,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.2833085879773838,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.4252,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.24673727005644763,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.4242,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.2666138844703155,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.423,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.2440017599765623,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.4269,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.27474866144977017,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.4188,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.2683949683158654,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.4277,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.24834830912510625,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.412,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.3182331291401319,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.4266,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.24209196090623306,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.4146,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.30571672382548665,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.4274,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.2708341093880287,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.4137,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.2543393071686141,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.4138,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.26636073294392976,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.4217,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.2694797412785414,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.4121,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.27210532547168564,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.4223,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.2781851237947751,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.4262,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.3036799150498283,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.4246,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.2939854942883591,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.4306,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.31585811085302834,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.4206,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.2626348379689296,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.4224,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.3097106181886677,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.4147,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.27977044813947316,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.4149,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.2752478095903011,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.4244,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.2815752869962845,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.4244,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.287761058671538,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.4339,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.2729419237107765,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.4137,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.2749366290582891,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.4145,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.2686320586662743,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.4261,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.2781883814609276,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.4164,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.23492758021172713,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.4197,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.24317059145802397,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.425,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.2714737655668381,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.4164,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.23943369511637455,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.4196,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.25896063066220815,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.4319,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.26936824636461765,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.4311,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.24624095798561405,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.4236,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.23289795174911035,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.4212,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.2508606359889803,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.4138,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.24662238701182712,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.4278,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.22297637156255842,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.4167,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.25664244696216765,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.4314,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.24285849268024565,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.4214,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.23751576533958463,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.4156,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.22534748468073806,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.4102,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.225622706585816,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.4223,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.23086561089779523,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.4259,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.22682771318272993,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.4281,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.2099513720001625,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.409,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.24410735919072551,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.4269,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.23431448880902653,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.4166,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.23940160786641096,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.4209,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.2372061172138042,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.4226,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.2373343071916907,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.4181,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.22873847129909405,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.4268,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.22293428103840723,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.4214,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.23046248718648862,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.4267,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.22397944697549424,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.4112,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.22039640541904307,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.4198,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.2399362817344911,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.4207,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.22078747157006318,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.4011,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.20761556902357128,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.4195,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.2331345185059278,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.4204,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.23127754405240125,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.4079,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.2192656966907477,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.432,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.2188062336389143,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.4238,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.2401879296083238,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.418,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.23017778348441417,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.4202,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.23918279874742823,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.4307,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.6094858455503853,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.4356,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.23457166495703322,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.4153,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.24645379776708423,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.4124,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.24447622835373226,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.4215,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.24592724276703778,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.4216,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.23027532503582618,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.4184,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.23459123432695578,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.4239,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.25204970523165093,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.423,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.228274505293236,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.419,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 1.125795842804083,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.4339,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.23025502114454618,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.425,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.2183300235191779,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.4213,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.2321765606211286,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.4211,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.23330661720368323,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.4191,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.2188159472770866,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.4134,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.23509895102701073,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.4129,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.2389153507591906,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.4099,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.2198044272113932,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.4138,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.25074956765720585,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.4175,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.23373858956645688,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.4205,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.2252082269635187,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.427,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.2342215009745082,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.4034,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.24012706663621813,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.4293,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.2287460401835426,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.427,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.23048129442323184,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.4255,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.23484617410580585,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.421,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.22819342467000034,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.4135,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.23832561634456687,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.415,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.22556853878790986,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.4274,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.23472785461930318,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.4287,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.22754439032462304,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.4123,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.2242447967940963,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.4207,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.2283494681227522,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.4113,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.22181549057104152,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.4207,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.23675417006146518,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.4086,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.22916431901605216,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.4236,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.21715959497225823,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.4216,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.2158304575690221,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.4139,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.2324551440227255,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.4226,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.2160351111668501,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.4094,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.23544044847712564,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.4087,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.254916761251563,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.4109,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.22560738415091977,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.4198,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.22360377232620815,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.413,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.24160716785787395,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.415,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.24612981952364468,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.4238,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.23641943519450062,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.4176,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.22053480810455078,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.4252,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.24828652042871355,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.4192,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.23034700388877252,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.4228,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 0.2193692592953392,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.4217,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.223687330769549,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.4246,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.23666517608629542,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.4172,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 1.1634735543016006,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.4211,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.22226723704467466,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.4187,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.8983637080578181,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.4125,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.21146584047419495,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.4149,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.2385145825608252,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.4254,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.2515886907459947,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.4278,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.2137994483253148,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.4114,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.2245574768056866,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.4188,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.269485745095042,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.4194,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.21942155942314054,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.409,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.2329333869597521,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.425,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.23975893194006523,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.4179,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.25000945077935066,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.4178,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.2266995085601006,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.4151,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.22456576091895677,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.4188,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.23063420879410337,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.4129,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.2308017968804769,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.4278,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.2157086319695383,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.4168,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.2257873856662602,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.4154,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.23909683935497328,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.429,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.22236963989901762,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.4187,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.21100073908297595,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.4209,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.22205831575495466,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.4013,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.2434470359238555,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.4124,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.23387474249089585,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.4125,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.21693783281601334,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.4268,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.21755850659780376,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.4114,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.2168411885697064,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.4176,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.20865600809990864,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.4281,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.2181489013275563,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.4201,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.20929489951924307,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.4118,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.20892205531715627,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.4203,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.21148578155892653,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.4199,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.21168830327475813,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.4141,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.2130546132755014,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.4209,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.22664965018021146,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.4233,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.2230174201038645,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.4196,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.2110545644463685,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.42,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.20844988762290056,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.4223,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.21653504189414052,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.4138,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.2050331249189353,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.4213,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.3785102007509296,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.3797,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.28880110175787554,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.3814,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.25137747973778285,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.3824,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.3272010151155223,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.3716,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.34187468383794006,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.3744,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.3301466969558241,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.3769,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 1.0787562695506419,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.3749,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.32384811440765593,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.3832,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.2972558910826124,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.3726,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.3073649808999794,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.3724,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.2817237917948355,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.373,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.2686695767827279,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.3846,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.2559553354090125,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.3781,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.26614954309728256,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.3753,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.25703986229846,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.3612,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.2513592091907806,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.3767,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.23774061939141897,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.3653,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.2465041422181261,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.3745,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.24437749408489795,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.3672,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.25384049215621307,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.3791,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.2319212751338709,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.3787,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.23181530389821106,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.3839,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.24325026936499305,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.3825,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.2430223628548571,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.3717,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.6206783524494142,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.3716,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.23530237428398657,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.3727,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.24391476440739743,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.3729,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.23212373419550775,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.3717,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.22842911053054643,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.3756,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.22554605592689864,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.3767,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.2436479152941388,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.376,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.2303458719791958,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.3787,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.213297144812893,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.377,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.2159294116823383,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.37,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.22389171456640616,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.3789,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.21685161152058002,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.3652,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.22571894730574316,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.3655,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.20976993775800226,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.3805,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.2115606434317671,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.3825,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.22129350364864944,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.3726,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.22883205167856951,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.3712,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.22329942652299703,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.3631,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.22329768836379235,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.3725,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.2243024795877846,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.3761,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.21352992000329105,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.3722,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.22608723794781282,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.38,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.24794277964090014,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.3775,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.231712130725902,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.3628,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.23239330663363403,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.366,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.20385665958284846,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.3843,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.21099200341491534,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.3749,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.21557349395991318,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.3637,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.22604444767522403,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.3727,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.2052486748119046,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.3721,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.21624680958973924,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.3795,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.21011830354767164,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.37,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.21869950750933137,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.3585,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.21195236450662264,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.3814,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.21463067906696673,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.3781,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.20837444249748863,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.3739,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.2154545776192865,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.3755,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.21572993520297498,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.3724,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.20700781015519226,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.3588,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.21558563513141185,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.3802,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.21367660157353627,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.3691,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.21516662255288777,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.3777,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.20497114095413405,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.3862,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.22209201147191465,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.3726,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.22998505272878617,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.3736,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.21542223873884414,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.3737,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.21259081699301116,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.3795,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.21317249002632727,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.3773,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.26974621231275053,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.3877,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.20972093222292196,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.376,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.22240674169101135,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.375,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.20492457531882985,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.3723,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.2181230084075852,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.3795,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.21010671544643658,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.3916,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.21294458816636014,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.3692,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.2126866890303431,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.3747,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.2008805515490747,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.3732,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.21602508276330523,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.3808,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.20989930085400485,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.3693,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.2039940257086165,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.375,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.20569498341105316,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.3703,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.2101268734601231,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.3791,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.22167936540002778,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.3581,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.20666859342568195,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.3825,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.20167274537610702,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.3612,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.20712448633289987,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.3719,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.2028208011772307,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.3711,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.21266316470515748,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.3724,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.21018201637467568,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.3701,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.19896049399646237,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.3753,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.21517952741166427,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.3612,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.2128419507363525,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.3712,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.20478288246733142,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.3744,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.20136890444624284,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.372,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.2067418620651178,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.3762,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.20920965355637716,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.365,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.2103345262824817,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.3676,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.2097223598367682,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.3625,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.1977074158159701,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.369,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.20735829734171052,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.3697,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.2074048048773947,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.3767,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.1954916381657991,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.3693,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.2129039063856292,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.3757,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.20861029520138458,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.3816,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.2107324378705956,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.3651,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.27148465587707765,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.3775,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.19958496847461063,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.3709,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.20814218513563149,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.3876,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.20337792867658497,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.3819,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.2077772986304015,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.3864,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.20303308945541804,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.3763,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.20377068247898092,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.3847,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.20019343541122447,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.368,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.20605961508226348,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.3756,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.21616686946459682,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.3675,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.20252715645932054,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.3629,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.20843310418077818,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.3764,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.19932426854850485,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.3659,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.20087162426073116,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.3658,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.2049859562803763,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.3825,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.1941053506443968,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.3704,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.20563539560047023,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.3763,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.19450373569448903,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.3823,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.22131934073966705,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.3807,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.2096408033319603,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.3743,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.20602901524962822,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.3817,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.20705971103951337,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.3734,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.2005177042240757,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.3652,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.204267608119766,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.368,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.19994928333215015,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.3938,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.20456278258330907,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.3747,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.21808718450032144,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.3764,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.20022806757065856,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.3603,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.2121717459577894,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.3733,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.198156305148303,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.3801,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.20779189224881744,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.379,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.1996440224969902,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.3762,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.2056909024623437,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.3715,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.20177983320237713,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.3716,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.2076042953806993,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.3789,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.2086486145856037,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.3747,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.20407503177095726,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.3622,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.21720418238691172,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.3739,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.20872686416175906,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.3739,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.20107382272368096,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.3713,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.20441952174627895,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.3749,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.20361950305504878,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.3769,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.19713489237396,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.359,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.19320101804631096,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.3853,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.21190434725282997,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.3706,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.20220714026940248,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.3665,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.19893071411233304,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.3779,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.20440615608543583,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.3776,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.2077611821281078,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.3745,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.19722812116816701,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.3781,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.2091062638373662,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.3711,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.19903022509848795,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.3759,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.1951205040738541,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.3896,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.19440682227545444,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.3689,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.20196563312528829,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.3778,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.20426447806247489,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.3711,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.19987786915803074,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.3751,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.20345209727051689,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.3804,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.19670362135244143,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.3713,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.19896878372032106,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.3773,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.1947541954803566,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.3701,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.20165824314922637,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.3776,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.21058399812998005,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.3772,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.2037261179919708,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.3626,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.19561130379905523,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.3666,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.19998374653023177,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.3655,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.1986107060383188,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.3736,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.21432602050787464,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.3733,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.1970093470437174,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.3754,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.20611244733287906,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.3783,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.19796599629098108,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.3815,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.19938941756562262,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.383,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.1964643010871256,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.3688,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.20331839981696234,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.362,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.20087653592712704,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.3673,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.20397242009176186,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.3694,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.20119582039575667,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.3641,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.2000762879684694,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.3727,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.2032029505869459,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.3684,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.1950476383288398,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.379,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.2124275822938688,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.3784,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.20498565739562807,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.3745,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.19228893445016668,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.3741,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.2098355452873232,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.3773,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.20088183422924208,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.3725,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.19935184480175963,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.3644,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.19996135425543504,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.3764,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.2027699529317874,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.3736,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.19501910874203943,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.3813,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.20600143418501005,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.3771,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.19230862701591223,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.3778,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.1965779241878704,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.3747,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.2047830227903774,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.3761,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.2070490416675477,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.3653,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.19786834169268036,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.3798,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.2020734891734032,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.3782,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.2019365085733824,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.3704,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.20499192044662323,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.3764,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.20423878987829286,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.3801,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.1979142230706447,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.3788,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.20053076991348207,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.371,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.19710177657226485,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.3669,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.1947828924582454,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.3693,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.19348141151837003,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.3699,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.8110580897734982,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.3943,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.1972918540268682,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.3766,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.19638934932674718,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.3802,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.2229356315809666,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.3775,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.19475134953662385,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.3843,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.19648311304172691,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.366,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.20762946183468403,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.3656,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.19851644235305269,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.3653,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.20527627646943422,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.3704,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.20032712856373663,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.3768,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.20374988037701855,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.3787,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.20392184422675685,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.379,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.19921375650039475,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.3773,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.20289423013627797,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.3819,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.190284518004219,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.3848,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.20332792008774403,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.3672,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.195117295841438,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.3698,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.19740647651911275,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.3765,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.19856737701489902,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.365,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.1994339149152077,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.3691,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.19431989915955045,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.3738,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.19829078250872897,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.3832,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.20529361076491734,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.3735,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.19686192473308103,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.3854,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.19887116812006742,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.3768,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.1990744973566862,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.3696,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.20331147045669754,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.378,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.20114814508853768,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.3751,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.2070742056881589,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.3735,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.19748610447047418,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.3809,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.20317493735602413,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.3705,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.1975710769494502,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.3732,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.21102692834030767,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.3804,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.2029917779662122,
+      "learning_rate": 0.0,
+      "loss": 0.3792,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 3851473559486464.0,
+      "train_loss": 0.5004229312483598,
+      "train_runtime": 30565.2663,
+      "train_samples_per_second": 5.169,
+      "train_steps_per_second": 0.04
+    }
+  ],
+  "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": 3851473559486464.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}