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": 6.048504446042652,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 1.0177,
+      "step": 1
+    },
+    {
+      "epoch": 0.008097165991902834,
+      "grad_norm": 6.165743983683067,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 1.0537,
+      "step": 2
+    },
+    {
+      "epoch": 0.012145748987854251,
+      "grad_norm": 6.0793757522168805,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 1.0413,
+      "step": 3
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.023370073923658,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.0426,
+      "step": 4
+    },
+    {
+      "epoch": 0.020242914979757085,
+      "grad_norm": 5.8112930031546215,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 1.0196,
+      "step": 5
+    },
+    {
+      "epoch": 0.024291497975708502,
+      "grad_norm": 5.673697714275445,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 1.0211,
+      "step": 6
+    },
+    {
+      "epoch": 0.02834008097165992,
+      "grad_norm": 4.443833227781453,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.9746,
+      "step": 7
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.345897098044867,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.9903,
+      "step": 8
+    },
+    {
+      "epoch": 0.03643724696356275,
+      "grad_norm": 2.259581419354385,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.9454,
+      "step": 9
+    },
+    {
+      "epoch": 0.04048582995951417,
+      "grad_norm": 1.9633382281598797,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.8906,
+      "step": 10
+    },
+    {
+      "epoch": 0.044534412955465584,
+      "grad_norm": 1.8890532377054712,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.9013,
+      "step": 11
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 4.000068079704403,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.879,
+      "step": 12
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 4.4155704463660665,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.8972,
+      "step": 13
+    },
+    {
+      "epoch": 0.05668016194331984,
+      "grad_norm": 4.5666381466713855,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.9128,
+      "step": 14
+    },
+    {
+      "epoch": 0.06072874493927125,
+      "grad_norm": 4.091227572286978,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.8951,
+      "step": 15
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 3.3257084125125327,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.8935,
+      "step": 16
+    },
+    {
+      "epoch": 0.06882591093117409,
+      "grad_norm": 2.9533019313460445,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 0.8753,
+      "step": 17
+    },
+    {
+      "epoch": 0.0728744939271255,
+      "grad_norm": 2.62545729854209,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 0.8466,
+      "step": 18
+    },
+    {
+      "epoch": 0.07692307692307693,
+      "grad_norm": 2.2920613859548298,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 0.8403,
+      "step": 19
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 1.87528677113942,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.8192,
+      "step": 20
+    },
+    {
+      "epoch": 0.08502024291497975,
+      "grad_norm": 1.5001067167337512,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 0.8084,
+      "step": 21
+    },
+    {
+      "epoch": 0.08906882591093117,
+      "grad_norm": 1.6745805069402575,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 0.8214,
+      "step": 22
+    },
+    {
+      "epoch": 0.0931174089068826,
+      "grad_norm": 1.9183529871664637,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 0.7792,
+      "step": 23
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.6722354915516686,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.7638,
+      "step": 24
+    },
+    {
+      "epoch": 0.10121457489878542,
+      "grad_norm": 1.4817425529175645,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.7961,
+      "step": 25
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 1.1368061375856902,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 0.7995,
+      "step": 26
+    },
+    {
+      "epoch": 0.10931174089068826,
+      "grad_norm": 1.3683363364608845,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 0.7666,
+      "step": 27
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 1.42089514122316,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.7644,
+      "step": 28
+    },
+    {
+      "epoch": 0.11740890688259109,
+      "grad_norm": 1.1356925435460277,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 0.7732,
+      "step": 29
+    },
+    {
+      "epoch": 0.1214574898785425,
+      "grad_norm": 0.9639709398925581,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.7551,
+      "step": 30
+    },
+    {
+      "epoch": 0.12550607287449392,
+      "grad_norm": 1.0248842708438597,
+      "learning_rate": 1e-05,
+      "loss": 0.7854,
+      "step": 31
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 1.0610865558771068,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.7738,
+      "step": 32
+    },
+    {
+      "epoch": 0.13360323886639677,
+      "grad_norm": 0.8264626048295632,
+      "learning_rate": 1.0645161290322582e-05,
+      "loss": 0.754,
+      "step": 33
+    },
+    {
+      "epoch": 0.13765182186234817,
+      "grad_norm": 0.7835334536077863,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 0.728,
+      "step": 34
+    },
+    {
+      "epoch": 0.1417004048582996,
+      "grad_norm": 0.8717940597928319,
+      "learning_rate": 1.1290322580645164e-05,
+      "loss": 0.737,
+      "step": 35
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 0.9377863437740397,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.728,
+      "step": 36
+    },
+    {
+      "epoch": 0.14979757085020243,
+      "grad_norm": 0.7287674883543903,
+      "learning_rate": 1.1935483870967743e-05,
+      "loss": 0.739,
+      "step": 37
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.8677269431726605,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 0.7334,
+      "step": 38
+    },
+    {
+      "epoch": 0.15789473684210525,
+      "grad_norm": 0.7629354205799459,
+      "learning_rate": 1.2580645161290324e-05,
+      "loss": 0.7283,
+      "step": 39
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 0.6360816438758752,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.7178,
+      "step": 40
+    },
+    {
+      "epoch": 0.1659919028340081,
+      "grad_norm": 0.767271388333272,
+      "learning_rate": 1.3225806451612903e-05,
+      "loss": 0.7435,
+      "step": 41
+    },
+    {
+      "epoch": 0.1700404858299595,
+      "grad_norm": 0.6984611530165209,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 0.7128,
+      "step": 42
+    },
+    {
+      "epoch": 0.17408906882591094,
+      "grad_norm": 0.5861615149686802,
+      "learning_rate": 1.3870967741935486e-05,
+      "loss": 0.7079,
+      "step": 43
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 0.6356670998979882,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.7569,
+      "step": 44
+    },
+    {
+      "epoch": 0.18218623481781376,
+      "grad_norm": 0.5480378152195359,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.7331,
+      "step": 45
+    },
+    {
+      "epoch": 0.1862348178137652,
+      "grad_norm": 0.6669237967121695,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 0.7338,
+      "step": 46
+    },
+    {
+      "epoch": 0.1902834008097166,
+      "grad_norm": 0.5534117038244993,
+      "learning_rate": 1.5161290322580646e-05,
+      "loss": 0.7073,
+      "step": 47
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 0.5721959817397152,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.7099,
+      "step": 48
+    },
+    {
+      "epoch": 0.19838056680161945,
+      "grad_norm": 0.5523322735967875,
+      "learning_rate": 1.5806451612903226e-05,
+      "loss": 0.6791,
+      "step": 49
+    },
+    {
+      "epoch": 0.20242914979757085,
+      "grad_norm": 0.7151665693512191,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.7286,
+      "step": 50
+    },
+    {
+      "epoch": 0.20647773279352227,
+      "grad_norm": 0.6117384454231076,
+      "learning_rate": 1.6451612903225807e-05,
+      "loss": 0.7231,
+      "step": 51
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.6594508480780212,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.711,
+      "step": 52
+    },
+    {
+      "epoch": 0.2145748987854251,
+      "grad_norm": 0.7176498968280578,
+      "learning_rate": 1.7096774193548388e-05,
+      "loss": 0.6931,
+      "step": 53
+    },
+    {
+      "epoch": 0.21862348178137653,
+      "grad_norm": 0.5734210537696224,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 0.7164,
+      "step": 54
+    },
+    {
+      "epoch": 0.22267206477732793,
+      "grad_norm": 0.657026382590817,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.7104,
+      "step": 55
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 1.5579662617203884,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.7675,
+      "step": 56
+    },
+    {
+      "epoch": 0.23076923076923078,
+      "grad_norm": 0.6258172823920116,
+      "learning_rate": 1.838709677419355e-05,
+      "loss": 0.7035,
+      "step": 57
+    },
+    {
+      "epoch": 0.23481781376518218,
+      "grad_norm": 0.6184433333026408,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 0.7051,
+      "step": 58
+    },
+    {
+      "epoch": 0.2388663967611336,
+      "grad_norm": 0.539580828316945,
+      "learning_rate": 1.903225806451613e-05,
+      "loss": 0.7105,
+      "step": 59
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 0.5918423093915179,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.7027,
+      "step": 60
+    },
+    {
+      "epoch": 0.24696356275303644,
+      "grad_norm": 0.5986388278567428,
+      "learning_rate": 1.967741935483871e-05,
+      "loss": 0.71,
+      "step": 61
+    },
+    {
+      "epoch": 0.25101214574898784,
+      "grad_norm": 0.6459077020675369,
+      "learning_rate": 2e-05,
+      "loss": 0.6836,
+      "step": 62
+    },
+    {
+      "epoch": 0.2550607287449393,
+      "grad_norm": 0.6623403122426573,
+      "learning_rate": 2.0322580645161292e-05,
+      "loss": 0.6923,
+      "step": 63
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 0.6825100221527812,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.6991,
+      "step": 64
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 0.6851346457754744,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.7178,
+      "step": 65
+    },
+    {
+      "epoch": 0.26720647773279355,
+      "grad_norm": 0.5872229061678019,
+      "learning_rate": 2.1290322580645163e-05,
+      "loss": 0.6843,
+      "step": 66
+    },
+    {
+      "epoch": 0.27125506072874495,
+      "grad_norm": 0.6275205973118323,
+      "learning_rate": 2.161290322580645e-05,
+      "loss": 0.6943,
+      "step": 67
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 0.6245262411636283,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.7326,
+      "step": 68
+    },
+    {
+      "epoch": 0.2793522267206478,
+      "grad_norm": 0.7032862490399725,
+      "learning_rate": 2.225806451612903e-05,
+      "loss": 0.703,
+      "step": 69
+    },
+    {
+      "epoch": 0.2834008097165992,
+      "grad_norm": 0.5680853017180928,
+      "learning_rate": 2.2580645161290328e-05,
+      "loss": 0.6936,
+      "step": 70
+    },
+    {
+      "epoch": 0.2874493927125506,
+      "grad_norm": 0.8587043873299046,
+      "learning_rate": 2.2903225806451615e-05,
+      "loss": 0.7078,
+      "step": 71
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 0.6247085521912207,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.7224,
+      "step": 72
+    },
+    {
+      "epoch": 0.29554655870445345,
+      "grad_norm": 0.846523962096058,
+      "learning_rate": 2.3548387096774196e-05,
+      "loss": 0.7095,
+      "step": 73
+    },
+    {
+      "epoch": 0.29959514170040485,
+      "grad_norm": 0.8993812265087744,
+      "learning_rate": 2.3870967741935486e-05,
+      "loss": 0.7096,
+      "step": 74
+    },
+    {
+      "epoch": 0.30364372469635625,
+      "grad_norm": 0.6611544786830991,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.6934,
+      "step": 75
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.6520204673332488,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.7117,
+      "step": 76
+    },
+    {
+      "epoch": 0.3117408906882591,
+      "grad_norm": 0.7476026505155058,
+      "learning_rate": 2.4838709677419358e-05,
+      "loss": 0.7075,
+      "step": 77
+    },
+    {
+      "epoch": 0.3157894736842105,
+      "grad_norm": 0.5832858568850996,
+      "learning_rate": 2.5161290322580648e-05,
+      "loss": 0.6904,
+      "step": 78
+    },
+    {
+      "epoch": 0.31983805668016196,
+      "grad_norm": 0.7251915865338596,
+      "learning_rate": 2.548387096774194e-05,
+      "loss": 0.6739,
+      "step": 79
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 0.7811868755936908,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.6957,
+      "step": 80
+    },
+    {
+      "epoch": 0.32793522267206476,
+      "grad_norm": 0.6845579196917546,
+      "learning_rate": 2.6129032258064516e-05,
+      "loss": 0.6796,
+      "step": 81
+    },
+    {
+      "epoch": 0.3319838056680162,
+      "grad_norm": 0.7461470338115368,
+      "learning_rate": 2.6451612903225806e-05,
+      "loss": 0.7198,
+      "step": 82
+    },
+    {
+      "epoch": 0.3360323886639676,
+      "grad_norm": 0.6860163018484934,
+      "learning_rate": 2.6774193548387097e-05,
+      "loss": 0.6854,
+      "step": 83
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.7418149902966715,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.6993,
+      "step": 84
+    },
+    {
+      "epoch": 0.3441295546558704,
+      "grad_norm": 0.7293124205881484,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.7057,
+      "step": 85
+    },
+    {
+      "epoch": 0.3481781376518219,
+      "grad_norm": 0.755546513449927,
+      "learning_rate": 2.774193548387097e-05,
+      "loss": 0.7009,
+      "step": 86
+    },
+    {
+      "epoch": 0.3522267206477733,
+      "grad_norm": 0.7375580910495049,
+      "learning_rate": 2.8064516129032262e-05,
+      "loss": 0.6493,
+      "step": 87
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.866899728895796,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.7004,
+      "step": 88
+    },
+    {
+      "epoch": 0.3603238866396761,
+      "grad_norm": 0.9943196778063562,
+      "learning_rate": 2.8709677419354843e-05,
+      "loss": 0.6956,
+      "step": 89
+    },
+    {
+      "epoch": 0.3643724696356275,
+      "grad_norm": 0.6954641847770591,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.6821,
+      "step": 90
+    },
+    {
+      "epoch": 0.3684210526315789,
+      "grad_norm": 1.1690497515725462,
+      "learning_rate": 2.9354838709677423e-05,
+      "loss": 0.6832,
+      "step": 91
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 0.6455929679779314,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.6539,
+      "step": 92
+    },
+    {
+      "epoch": 0.3765182186234818,
+      "grad_norm": 0.8013865643867348,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.6659,
+      "step": 93
+    },
+    {
+      "epoch": 0.3805668016194332,
+      "grad_norm": 0.7796525140828724,
+      "learning_rate": 3.032258064516129e-05,
+      "loss": 0.6943,
+      "step": 94
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 0.6547271681507375,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.6798,
+      "step": 95
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.6395182733825817,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.6645,
+      "step": 96
+    },
+    {
+      "epoch": 0.39271255060728744,
+      "grad_norm": 0.6282083339552407,
+      "learning_rate": 3.1290322580645166e-05,
+      "loss": 0.7037,
+      "step": 97
+    },
+    {
+      "epoch": 0.3967611336032389,
+      "grad_norm": 0.8876396048548794,
+      "learning_rate": 3.161290322580645e-05,
+      "loss": 0.6856,
+      "step": 98
+    },
+    {
+      "epoch": 0.4008097165991903,
+      "grad_norm": 0.6571163782946424,
+      "learning_rate": 3.193548387096774e-05,
+      "loss": 0.6594,
+      "step": 99
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 0.8033434748046575,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.6932,
+      "step": 100
+    },
+    {
+      "epoch": 0.4089068825910931,
+      "grad_norm": 0.7395556871986123,
+      "learning_rate": 3.258064516129033e-05,
+      "loss": 0.714,
+      "step": 101
+    },
+    {
+      "epoch": 0.41295546558704455,
+      "grad_norm": 0.744570072639455,
+      "learning_rate": 3.2903225806451614e-05,
+      "loss": 0.6931,
+      "step": 102
+    },
+    {
+      "epoch": 0.41700404858299595,
+      "grad_norm": 0.8514405165081699,
+      "learning_rate": 3.322580645161291e-05,
+      "loss": 0.6887,
+      "step": 103
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 0.7562525571176634,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.687,
+      "step": 104
+    },
+    {
+      "epoch": 0.4251012145748988,
+      "grad_norm": 1.167655215996667,
+      "learning_rate": 3.387096774193549e-05,
+      "loss": 0.6801,
+      "step": 105
+    },
+    {
+      "epoch": 0.4291497975708502,
+      "grad_norm": 0.985185805558799,
+      "learning_rate": 3.4193548387096776e-05,
+      "loss": 0.6931,
+      "step": 106
+    },
+    {
+      "epoch": 0.4331983805668016,
+      "grad_norm": 1.0752000367779424,
+      "learning_rate": 3.451612903225807e-05,
+      "loss": 0.6722,
+      "step": 107
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.7910357220355877,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.6632,
+      "step": 108
+    },
+    {
+      "epoch": 0.44129554655870445,
+      "grad_norm": 1.1355646897656737,
+      "learning_rate": 3.516129032258065e-05,
+      "loss": 0.6824,
+      "step": 109
+    },
+    {
+      "epoch": 0.44534412955465585,
+      "grad_norm": 0.7702109296065348,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.6632,
+      "step": 110
+    },
+    {
+      "epoch": 0.4493927125506073,
+      "grad_norm": 1.361921883632703,
+      "learning_rate": 3.5806451612903225e-05,
+      "loss": 0.6703,
+      "step": 111
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.8470791382022712,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.6453,
+      "step": 112
+    },
+    {
+      "epoch": 0.4574898785425101,
+      "grad_norm": 1.3726774303313287,
+      "learning_rate": 3.6451612903225805e-05,
+      "loss": 0.6766,
+      "step": 113
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.7943916445191549,
+      "learning_rate": 3.67741935483871e-05,
+      "loss": 0.6467,
+      "step": 114
+    },
+    {
+      "epoch": 0.46558704453441296,
+      "grad_norm": 1.0434876825655823,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.6502,
+      "step": 115
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 0.903047604773851,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.6824,
+      "step": 116
+    },
+    {
+      "epoch": 0.47368421052631576,
+      "grad_norm": 0.9003133438618667,
+      "learning_rate": 3.7741935483870974e-05,
+      "loss": 0.6802,
+      "step": 117
+    },
+    {
+      "epoch": 0.4777327935222672,
+      "grad_norm": 0.9750095563539432,
+      "learning_rate": 3.806451612903226e-05,
+      "loss": 0.6503,
+      "step": 118
+    },
+    {
+      "epoch": 0.4817813765182186,
+      "grad_norm": 0.949683817775843,
+      "learning_rate": 3.8387096774193555e-05,
+      "loss": 0.674,
+      "step": 119
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 1.3429658291992699,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.6777,
+      "step": 120
+    },
+    {
+      "epoch": 0.4898785425101215,
+      "grad_norm": 0.9042936018179546,
+      "learning_rate": 3.9032258064516135e-05,
+      "loss": 0.6681,
+      "step": 121
+    },
+    {
+      "epoch": 0.4939271255060729,
+      "grad_norm": 1.2816687656743866,
+      "learning_rate": 3.935483870967742e-05,
+      "loss": 0.6537,
+      "step": 122
+    },
+    {
+      "epoch": 0.4979757085020243,
+      "grad_norm": 0.8991889919712995,
+      "learning_rate": 3.9677419354838716e-05,
+      "loss": 0.6925,
+      "step": 123
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 1.1670590549473845,
+      "learning_rate": 4e-05,
+      "loss": 0.6993,
+      "step": 124
+    },
+    {
+      "epoch": 0.5060728744939271,
+      "grad_norm": 0.9555790502950985,
+      "learning_rate": 3.9999920040266474e-05,
+      "loss": 0.6576,
+      "step": 125
+    },
+    {
+      "epoch": 0.5101214574898786,
+      "grad_norm": 1.2457230158052148,
+      "learning_rate": 3.999968016170525e-05,
+      "loss": 0.6785,
+      "step": 126
+    },
+    {
+      "epoch": 0.5141700404858299,
+      "grad_norm": 0.8620903935086712,
+      "learning_rate": 3.9999280366234394e-05,
+      "loss": 0.6977,
+      "step": 127
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.1707251338518816,
+      "learning_rate": 3.999872065705065e-05,
+      "loss": 0.7006,
+      "step": 128
+    },
+    {
+      "epoch": 0.5222672064777328,
+      "grad_norm": 0.9914638153273833,
+      "learning_rate": 3.999800103862945e-05,
+      "loss": 0.6956,
+      "step": 129
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 1.0686848076337379,
+      "learning_rate": 3.9997121516724834e-05,
+      "loss": 0.6667,
+      "step": 130
+    },
+    {
+      "epoch": 0.5303643724696356,
+      "grad_norm": 1.1548949977379124,
+      "learning_rate": 3.999608209836944e-05,
+      "loss": 0.6686,
+      "step": 131
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 0.8860185245037525,
+      "learning_rate": 3.999488279187443e-05,
+      "loss": 0.677,
+      "step": 132
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 1.3074393848090364,
+      "learning_rate": 3.9993523606829425e-05,
+      "loss": 0.6937,
+      "step": 133
+    },
+    {
+      "epoch": 0.5425101214574899,
+      "grad_norm": 0.8940882728981137,
+      "learning_rate": 3.9992004554102435e-05,
+      "loss": 0.7193,
+      "step": 134
+    },
+    {
+      "epoch": 0.5465587044534413,
+      "grad_norm": 0.8710173552248495,
+      "learning_rate": 3.999032564583976e-05,
+      "loss": 0.699,
+      "step": 135
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 0.7886130230683265,
+      "learning_rate": 3.998848689546591e-05,
+      "loss": 0.689,
+      "step": 136
+    },
+    {
+      "epoch": 0.5546558704453441,
+      "grad_norm": 0.6879239761600101,
+      "learning_rate": 3.9986488317683496e-05,
+      "loss": 0.6711,
+      "step": 137
+    },
+    {
+      "epoch": 0.5587044534412956,
+      "grad_norm": 0.7493282144247371,
+      "learning_rate": 3.9984329928473067e-05,
+      "loss": 0.6802,
+      "step": 138
+    },
+    {
+      "epoch": 0.562753036437247,
+      "grad_norm": 0.8737806752867242,
+      "learning_rate": 3.998201174509306e-05,
+      "loss": 0.6702,
+      "step": 139
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 0.8842582023581711,
+      "learning_rate": 3.997953378607961e-05,
+      "loss": 0.6783,
+      "step": 140
+    },
+    {
+      "epoch": 0.5708502024291497,
+      "grad_norm": 0.7189794840103163,
+      "learning_rate": 3.9976896071246403e-05,
+      "loss": 0.6572,
+      "step": 141
+    },
+    {
+      "epoch": 0.5748987854251012,
+      "grad_norm": 0.833876368459367,
+      "learning_rate": 3.997409862168454e-05,
+      "loss": 0.6527,
+      "step": 142
+    },
+    {
+      "epoch": 0.5789473684210527,
+      "grad_norm": 0.748172937189673,
+      "learning_rate": 3.997114145976237e-05,
+      "loss": 0.6382,
+      "step": 143
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.5469095248545819,
+      "learning_rate": 3.9968024609125245e-05,
+      "loss": 0.6873,
+      "step": 144
+    },
+    {
+      "epoch": 0.5870445344129555,
+      "grad_norm": 0.7942044409543876,
+      "learning_rate": 3.996474809469545e-05,
+      "loss": 0.6876,
+      "step": 145
+    },
+    {
+      "epoch": 0.5910931174089069,
+      "grad_norm": 0.7087320639967432,
+      "learning_rate": 3.996131194267189e-05,
+      "loss": 0.6853,
+      "step": 146
+    },
+    {
+      "epoch": 0.5951417004048583,
+      "grad_norm": 0.6435242675613878,
+      "learning_rate": 3.995771618052995e-05,
+      "loss": 0.6601,
+      "step": 147
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 0.9875436914016895,
+      "learning_rate": 3.9953960837021256e-05,
+      "loss": 0.6826,
+      "step": 148
+    },
+    {
+      "epoch": 0.6032388663967612,
+      "grad_norm": 0.7786654364857014,
+      "learning_rate": 3.995004594217343e-05,
+      "loss": 0.6972,
+      "step": 149
+    },
+    {
+      "epoch": 0.6072874493927125,
+      "grad_norm": 0.730100347030268,
+      "learning_rate": 3.994597152728987e-05,
+      "loss": 0.6492,
+      "step": 150
+    },
+    {
+      "epoch": 0.611336032388664,
+      "grad_norm": 0.611882599081241,
+      "learning_rate": 3.994173762494947e-05,
+      "loss": 0.6646,
+      "step": 151
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.6831061675147548,
+      "learning_rate": 3.9937344269006424e-05,
+      "loss": 0.6778,
+      "step": 152
+    },
+    {
+      "epoch": 0.6194331983805668,
+      "grad_norm": 0.5951093118753646,
+      "learning_rate": 3.993279149458988e-05,
+      "loss": 0.6822,
+      "step": 153
+    },
+    {
+      "epoch": 0.6234817813765182,
+      "grad_norm": 0.6118127326556406,
+      "learning_rate": 3.9928079338103695e-05,
+      "loss": 0.6618,
+      "step": 154
+    },
+    {
+      "epoch": 0.6275303643724697,
+      "grad_norm": 0.5375503223245393,
+      "learning_rate": 3.992320783722616e-05,
+      "loss": 0.6844,
+      "step": 155
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 0.5652760174508046,
+      "learning_rate": 3.991817703090965e-05,
+      "loss": 0.6939,
+      "step": 156
+    },
+    {
+      "epoch": 0.6356275303643725,
+      "grad_norm": 0.6192362626294375,
+      "learning_rate": 3.991298695938038e-05,
+      "loss": 0.6513,
+      "step": 157
+    },
+    {
+      "epoch": 0.6396761133603239,
+      "grad_norm": 0.6036311989805399,
+      "learning_rate": 3.9907637664138e-05,
+      "loss": 0.6699,
+      "step": 158
+    },
+    {
+      "epoch": 0.6437246963562753,
+      "grad_norm": 0.7484267277848907,
+      "learning_rate": 3.990212918795534e-05,
+      "loss": 0.6873,
+      "step": 159
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.734018986984322,
+      "learning_rate": 3.9896461574878036e-05,
+      "loss": 0.6546,
+      "step": 160
+    },
+    {
+      "epoch": 0.6518218623481782,
+      "grad_norm": 0.6313721025338035,
+      "learning_rate": 3.989063487022416e-05,
+      "loss": 0.6781,
+      "step": 161
+    },
+    {
+      "epoch": 0.6558704453441295,
+      "grad_norm": 0.7032454870904765,
+      "learning_rate": 3.9884649120583903e-05,
+      "loss": 0.6495,
+      "step": 162
+    },
+    {
+      "epoch": 0.659919028340081,
+      "grad_norm": 0.5756015566620266,
+      "learning_rate": 3.987850437381915e-05,
+      "loss": 0.6591,
+      "step": 163
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.5701025893693944,
+      "learning_rate": 3.987220067906313e-05,
+      "loss": 0.6725,
+      "step": 164
+    },
+    {
+      "epoch": 0.6680161943319838,
+      "grad_norm": 0.5837016461300196,
+      "learning_rate": 3.986573808672002e-05,
+      "loss": 0.6615,
+      "step": 165
+    },
+    {
+      "epoch": 0.6720647773279352,
+      "grad_norm": 0.6846796415770191,
+      "learning_rate": 3.9859116648464534e-05,
+      "loss": 0.654,
+      "step": 166
+    },
+    {
+      "epoch": 0.6761133603238867,
+      "grad_norm": 0.515239996348152,
+      "learning_rate": 3.985233641724152e-05,
+      "loss": 0.6645,
+      "step": 167
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.6487780484332643,
+      "learning_rate": 3.984539744726554e-05,
+      "loss": 0.6313,
+      "step": 168
+    },
+    {
+      "epoch": 0.6842105263157895,
+      "grad_norm": 0.6089352785349043,
+      "learning_rate": 3.983829979402038e-05,
+      "loss": 0.6974,
+      "step": 169
+    },
+    {
+      "epoch": 0.6882591093117408,
+      "grad_norm": 0.6058893499317228,
+      "learning_rate": 3.983104351425871e-05,
+      "loss": 0.6486,
+      "step": 170
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 0.5941991191567287,
+      "learning_rate": 3.982362866600154e-05,
+      "loss": 0.6749,
+      "step": 171
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.5935236519947086,
+      "learning_rate": 3.9816055308537806e-05,
+      "loss": 0.6457,
+      "step": 172
+    },
+    {
+      "epoch": 0.7004048582995951,
+      "grad_norm": 0.6065418278508488,
+      "learning_rate": 3.980832350242388e-05,
+      "loss": 0.6494,
+      "step": 173
+    },
+    {
+      "epoch": 0.7044534412955465,
+      "grad_norm": 0.5350147502131865,
+      "learning_rate": 3.980043330948306e-05,
+      "loss": 0.647,
+      "step": 174
+    },
+    {
+      "epoch": 0.708502024291498,
+      "grad_norm": 0.7127564708108989,
+      "learning_rate": 3.9792384792805124e-05,
+      "loss": 0.6544,
+      "step": 175
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.6132769294966509,
+      "learning_rate": 3.978417801674579e-05,
+      "loss": 0.6794,
+      "step": 176
+    },
+    {
+      "epoch": 0.7165991902834008,
+      "grad_norm": 0.605688367735808,
+      "learning_rate": 3.9775813046926244e-05,
+      "loss": 0.6829,
+      "step": 177
+    },
+    {
+      "epoch": 0.7206477732793523,
+      "grad_norm": 0.7142000144291976,
+      "learning_rate": 3.976728995023254e-05,
+      "loss": 0.6669,
+      "step": 178
+    },
+    {
+      "epoch": 0.7246963562753036,
+      "grad_norm": 0.610627956652062,
+      "learning_rate": 3.975860879481514e-05,
+      "loss": 0.6584,
+      "step": 179
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 0.6467431046512053,
+      "learning_rate": 3.974976965008833e-05,
+      "loss": 0.647,
+      "step": 180
+    },
+    {
+      "epoch": 0.7327935222672065,
+      "grad_norm": 0.5718013941564709,
+      "learning_rate": 3.9740772586729674e-05,
+      "loss": 0.6542,
+      "step": 181
+    },
+    {
+      "epoch": 0.7368421052631579,
+      "grad_norm": 0.6026253357948874,
+      "learning_rate": 3.973161767667946e-05,
+      "loss": 0.6752,
+      "step": 182
+    },
+    {
+      "epoch": 0.7408906882591093,
+      "grad_norm": 0.591934413051255,
+      "learning_rate": 3.972230499314009e-05,
+      "loss": 0.6824,
+      "step": 183
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 0.5020240561835012,
+      "learning_rate": 3.9712834610575545e-05,
+      "loss": 0.661,
+      "step": 184
+    },
+    {
+      "epoch": 0.7489878542510121,
+      "grad_norm": 0.6313019305175185,
+      "learning_rate": 3.9703206604710746e-05,
+      "loss": 0.6732,
+      "step": 185
+    },
+    {
+      "epoch": 0.7530364372469636,
+      "grad_norm": 0.5307374933772544,
+      "learning_rate": 3.969342105253098e-05,
+      "loss": 0.6734,
+      "step": 186
+    },
+    {
+      "epoch": 0.757085020242915,
+      "grad_norm": 0.5820574170762799,
+      "learning_rate": 3.9683478032281246e-05,
+      "loss": 0.6634,
+      "step": 187
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.5770459374175093,
+      "learning_rate": 3.967337762346568e-05,
+      "loss": 0.6869,
+      "step": 188
+    },
+    {
+      "epoch": 0.7651821862348178,
+      "grad_norm": 0.4894539349023939,
+      "learning_rate": 3.9663119906846885e-05,
+      "loss": 0.6604,
+      "step": 189
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.638904500604752,
+      "learning_rate": 3.965270496444528e-05,
+      "loss": 0.6731,
+      "step": 190
+    },
+    {
+      "epoch": 0.7732793522267206,
+      "grad_norm": 0.5056391432622739,
+      "learning_rate": 3.9642132879538465e-05,
+      "loss": 0.6553,
+      "step": 191
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 0.5465736516860621,
+      "learning_rate": 3.9631403736660566e-05,
+      "loss": 0.6334,
+      "step": 192
+    },
+    {
+      "epoch": 0.7813765182186235,
+      "grad_norm": 0.5253697309592258,
+      "learning_rate": 3.962051762160151e-05,
+      "loss": 0.6742,
+      "step": 193
+    },
+    {
+      "epoch": 0.7854251012145749,
+      "grad_norm": 0.615957845487749,
+      "learning_rate": 3.960947462140638e-05,
+      "loss": 0.6683,
+      "step": 194
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 0.6736087659669611,
+      "learning_rate": 3.959827482437472e-05,
+      "loss": 0.6633,
+      "step": 195
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.5301755950311339,
+      "learning_rate": 3.958691832005981e-05,
+      "loss": 0.6561,
+      "step": 196
+    },
+    {
+      "epoch": 0.7975708502024291,
+      "grad_norm": 0.7303989442424451,
+      "learning_rate": 3.9575405199267944e-05,
+      "loss": 0.6743,
+      "step": 197
+    },
+    {
+      "epoch": 0.8016194331983806,
+      "grad_norm": 0.6058034554236861,
+      "learning_rate": 3.956373555405774e-05,
+      "loss": 0.6783,
+      "step": 198
+    },
+    {
+      "epoch": 0.805668016194332,
+      "grad_norm": 0.6786078573265087,
+      "learning_rate": 3.955190947773937e-05,
+      "loss": 0.7011,
+      "step": 199
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.6388703225752048,
+      "learning_rate": 3.9539927064873824e-05,
+      "loss": 0.6734,
+      "step": 200
+    },
+    {
+      "epoch": 0.8137651821862348,
+      "grad_norm": 0.5898070423800219,
+      "learning_rate": 3.952778841127214e-05,
+      "loss": 0.6839,
+      "step": 201
+    },
+    {
+      "epoch": 0.8178137651821862,
+      "grad_norm": 0.6465037750287846,
+      "learning_rate": 3.95154936139947e-05,
+      "loss": 0.6835,
+      "step": 202
+    },
+    {
+      "epoch": 0.8218623481781376,
+      "grad_norm": 0.5383988017967014,
+      "learning_rate": 3.950304277135034e-05,
+      "loss": 0.6501,
+      "step": 203
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 0.6168030893079044,
+      "learning_rate": 3.94904359828957e-05,
+      "loss": 0.6714,
+      "step": 204
+    },
+    {
+      "epoch": 0.8299595141700404,
+      "grad_norm": 0.5958396476457989,
+      "learning_rate": 3.94776733494343e-05,
+      "loss": 0.6927,
+      "step": 205
+    },
+    {
+      "epoch": 0.8340080971659919,
+      "grad_norm": 0.5763823995425548,
+      "learning_rate": 3.9464754973015826e-05,
+      "loss": 0.6592,
+      "step": 206
+    },
+    {
+      "epoch": 0.8380566801619433,
+      "grad_norm": 0.6846216778736759,
+      "learning_rate": 3.9451680956935275e-05,
+      "loss": 0.6965,
+      "step": 207
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.6290879885452106,
+      "learning_rate": 3.943845140573213e-05,
+      "loss": 0.65,
+      "step": 208
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 0.6537880849205943,
+      "learning_rate": 3.942506642518952e-05,
+      "loss": 0.6528,
+      "step": 209
+    },
+    {
+      "epoch": 0.8502024291497976,
+      "grad_norm": 0.5826275756364923,
+      "learning_rate": 3.9411526122333396e-05,
+      "loss": 0.6858,
+      "step": 210
+    },
+    {
+      "epoch": 0.854251012145749,
+      "grad_norm": 0.659680239248074,
+      "learning_rate": 3.939783060543167e-05,
+      "loss": 0.6388,
+      "step": 211
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 0.5203417704361203,
+      "learning_rate": 3.938397998399333e-05,
+      "loss": 0.6495,
+      "step": 212
+    },
+    {
+      "epoch": 0.8623481781376519,
+      "grad_norm": 0.49434376360408055,
+      "learning_rate": 3.936997436876756e-05,
+      "loss": 0.6446,
+      "step": 213
+    },
+    {
+      "epoch": 0.8663967611336032,
+      "grad_norm": 0.5074184930551908,
+      "learning_rate": 3.935581387174289e-05,
+      "loss": 0.6534,
+      "step": 214
+    },
+    {
+      "epoch": 0.8704453441295547,
+      "grad_norm": 0.6003706279458842,
+      "learning_rate": 3.9341498606146295e-05,
+      "loss": 0.636,
+      "step": 215
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.5196371686807054,
+      "learning_rate": 3.932702868644224e-05,
+      "loss": 0.6655,
+      "step": 216
+    },
+    {
+      "epoch": 0.8785425101214575,
+      "grad_norm": 0.5108580910873722,
+      "learning_rate": 3.931240422833181e-05,
+      "loss": 0.6497,
+      "step": 217
+    },
+    {
+      "epoch": 0.8825910931174089,
+      "grad_norm": 0.47024040839675046,
+      "learning_rate": 3.9297625348751806e-05,
+      "loss": 0.6728,
+      "step": 218
+    },
+    {
+      "epoch": 0.8866396761133604,
+      "grad_norm": 0.44872874183841,
+      "learning_rate": 3.9282692165873744e-05,
+      "loss": 0.6242,
+      "step": 219
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.4981990220499471,
+      "learning_rate": 3.9267604799102945e-05,
+      "loss": 0.6213,
+      "step": 220
+    },
+    {
+      "epoch": 0.8947368421052632,
+      "grad_norm": 0.5006317381411369,
+      "learning_rate": 3.9252363369077615e-05,
+      "loss": 0.6482,
+      "step": 221
+    },
+    {
+      "epoch": 0.8987854251012146,
+      "grad_norm": 0.49684993259616794,
+      "learning_rate": 3.9236967997667796e-05,
+      "loss": 0.6465,
+      "step": 222
+    },
+    {
+      "epoch": 0.902834008097166,
+      "grad_norm": 0.49931311119766525,
+      "learning_rate": 3.9221418807974496e-05,
+      "loss": 0.6629,
+      "step": 223
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.5346157996678983,
+      "learning_rate": 3.92057159243286e-05,
+      "loss": 0.6611,
+      "step": 224
+    },
+    {
+      "epoch": 0.9109311740890689,
+      "grad_norm": 0.49628474554885327,
+      "learning_rate": 3.9189859472289956e-05,
+      "loss": 0.6731,
+      "step": 225
+    },
+    {
+      "epoch": 0.9149797570850202,
+      "grad_norm": 0.5823965855955634,
+      "learning_rate": 3.9173849578646324e-05,
+      "loss": 0.6484,
+      "step": 226
+    },
+    {
+      "epoch": 0.9190283400809717,
+      "grad_norm": 0.5565418729089286,
+      "learning_rate": 3.9157686371412395e-05,
+      "loss": 0.6697,
+      "step": 227
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.46925859531569253,
+      "learning_rate": 3.914136997982874e-05,
+      "loss": 0.6922,
+      "step": 228
+    },
+    {
+      "epoch": 0.9271255060728745,
+      "grad_norm": 0.5129084899320188,
+      "learning_rate": 3.912490053436079e-05,
+      "loss": 0.6718,
+      "step": 229
+    },
+    {
+      "epoch": 0.9311740890688259,
+      "grad_norm": 0.5499977858808365,
+      "learning_rate": 3.9108278166697805e-05,
+      "loss": 0.6391,
+      "step": 230
+    },
+    {
+      "epoch": 0.9352226720647774,
+      "grad_norm": 0.4716195145756715,
+      "learning_rate": 3.909150300975178e-05,
+      "loss": 0.6186,
+      "step": 231
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 0.6006938120340912,
+      "learning_rate": 3.907457519765642e-05,
+      "loss": 0.6236,
+      "step": 232
+    },
+    {
+      "epoch": 0.9433198380566802,
+      "grad_norm": 0.5015879899603167,
+      "learning_rate": 3.905749486576607e-05,
+      "loss": 0.6741,
+      "step": 233
+    },
+    {
+      "epoch": 0.9473684210526315,
+      "grad_norm": 0.7065303624629019,
+      "learning_rate": 3.90402621506546e-05,
+      "loss": 0.6575,
+      "step": 234
+    },
+    {
+      "epoch": 0.951417004048583,
+      "grad_norm": 0.4521008054240456,
+      "learning_rate": 3.902287719011435e-05,
+      "loss": 0.6487,
+      "step": 235
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.5491967460081992,
+      "learning_rate": 3.900534012315499e-05,
+      "loss": 0.6138,
+      "step": 236
+    },
+    {
+      "epoch": 0.9595141700404858,
+      "grad_norm": 0.44695493115281204,
+      "learning_rate": 3.8987651090002454e-05,
+      "loss": 0.6149,
+      "step": 237
+    },
+    {
+      "epoch": 0.9635627530364372,
+      "grad_norm": 0.5156348329940745,
+      "learning_rate": 3.896981023209777e-05,
+      "loss": 0.6423,
+      "step": 238
+    },
+    {
+      "epoch": 0.9676113360323887,
+      "grad_norm": 0.46737344583556706,
+      "learning_rate": 3.8951817692095966e-05,
+      "loss": 0.6309,
+      "step": 239
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.5202071654293702,
+      "learning_rate": 3.893367361386491e-05,
+      "loss": 0.6738,
+      "step": 240
+    },
+    {
+      "epoch": 0.9757085020242915,
+      "grad_norm": 0.5165893242795226,
+      "learning_rate": 3.891537814248417e-05,
+      "loss": 0.6513,
+      "step": 241
+    },
+    {
+      "epoch": 0.979757085020243,
+      "grad_norm": 0.5490428496389494,
+      "learning_rate": 3.889693142424384e-05,
+      "loss": 0.6481,
+      "step": 242
+    },
+    {
+      "epoch": 0.9838056680161943,
+      "grad_norm": 0.4789402758398701,
+      "learning_rate": 3.8878333606643405e-05,
+      "loss": 0.628,
+      "step": 243
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.4605456258058425,
+      "learning_rate": 3.885958483839049e-05,
+      "loss": 0.6495,
+      "step": 244
+    },
+    {
+      "epoch": 0.9919028340080972,
+      "grad_norm": 0.4872438211190824,
+      "learning_rate": 3.8840685269399786e-05,
+      "loss": 0.6202,
+      "step": 245
+    },
+    {
+      "epoch": 0.9959514170040485,
+      "grad_norm": 0.45295292656944425,
+      "learning_rate": 3.882163505079171e-05,
+      "loss": 0.6296,
+      "step": 246
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4340332149905063,
+      "learning_rate": 3.880243433489132e-05,
+      "loss": 0.6236,
+      "step": 247
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 0.5951100242433784,
+      "learning_rate": 3.878308327522702e-05,
+      "loss": 0.5383,
+      "step": 248
+    },
+    {
+      "epoch": 1.008097165991903,
+      "grad_norm": 0.46803382490550166,
+      "learning_rate": 3.876358202652936e-05,
+      "loss": 0.5455,
+      "step": 249
+    },
+    {
+      "epoch": 1.0121457489878543,
+      "grad_norm": 0.6094225132902904,
+      "learning_rate": 3.8743930744729835e-05,
+      "loss": 0.5317,
+      "step": 250
+    },
+    {
+      "epoch": 1.0161943319838056,
+      "grad_norm": 0.5580786340059439,
+      "learning_rate": 3.8724129586959535e-05,
+      "loss": 0.552,
+      "step": 251
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.6231698957607505,
+      "learning_rate": 3.870417871154801e-05,
+      "loss": 0.5391,
+      "step": 252
+    },
+    {
+      "epoch": 1.0242914979757085,
+      "grad_norm": 0.5312947869606666,
+      "learning_rate": 3.868407827802192e-05,
+      "loss": 0.5287,
+      "step": 253
+    },
+    {
+      "epoch": 1.0283400809716599,
+      "grad_norm": 0.5363214317444764,
+      "learning_rate": 3.8663828447103804e-05,
+      "loss": 0.5563,
+      "step": 254
+    },
+    {
+      "epoch": 1.0323886639676114,
+      "grad_norm": 0.5935773632898828,
+      "learning_rate": 3.8643429380710765e-05,
+      "loss": 0.5545,
+      "step": 255
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 0.6081609313186858,
+      "learning_rate": 3.862288124195319e-05,
+      "loss": 0.551,
+      "step": 256
+    },
+    {
+      "epoch": 1.040485829959514,
+      "grad_norm": 0.5649889386201637,
+      "learning_rate": 3.8602184195133464e-05,
+      "loss": 0.541,
+      "step": 257
+    },
+    {
+      "epoch": 1.0445344129554657,
+      "grad_norm": 0.5682088466630207,
+      "learning_rate": 3.8581338405744604e-05,
+      "loss": 0.5331,
+      "step": 258
+    },
+    {
+      "epoch": 1.048582995951417,
+      "grad_norm": 0.7500884644608634,
+      "learning_rate": 3.8560344040468996e-05,
+      "loss": 0.5427,
+      "step": 259
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.6457120029341202,
+      "learning_rate": 3.853920126717703e-05,
+      "loss": 0.5413,
+      "step": 260
+    },
+    {
+      "epoch": 1.05668016194332,
+      "grad_norm": 0.547339112020388,
+      "learning_rate": 3.851791025492574e-05,
+      "loss": 0.5346,
+      "step": 261
+    },
+    {
+      "epoch": 1.0607287449392713,
+      "grad_norm": 0.5401293146490694,
+      "learning_rate": 3.8496471173957515e-05,
+      "loss": 0.5402,
+      "step": 262
+    },
+    {
+      "epoch": 1.0647773279352226,
+      "grad_norm": 0.5674222854116019,
+      "learning_rate": 3.847488419569866e-05,
+      "loss": 0.5151,
+      "step": 263
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.519816324503423,
+      "learning_rate": 3.8453149492758077e-05,
+      "loss": 0.5373,
+      "step": 264
+    },
+    {
+      "epoch": 1.0728744939271255,
+      "grad_norm": 0.5774132379586974,
+      "learning_rate": 3.8431267238925886e-05,
+      "loss": 0.535,
+      "step": 265
+    },
+    {
+      "epoch": 1.0769230769230769,
+      "grad_norm": 0.643927542510621,
+      "learning_rate": 3.840923760917199e-05,
+      "loss": 0.5375,
+      "step": 266
+    },
+    {
+      "epoch": 1.0809716599190284,
+      "grad_norm": 0.5069090631453022,
+      "learning_rate": 3.838706077964473e-05,
+      "loss": 0.4977,
+      "step": 267
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.6386395600832856,
+      "learning_rate": 3.836473692766944e-05,
+      "loss": 0.5271,
+      "step": 268
+    },
+    {
+      "epoch": 1.0890688259109311,
+      "grad_norm": 0.6072803372061669,
+      "learning_rate": 3.834226623174705e-05,
+      "loss": 0.5433,
+      "step": 269
+    },
+    {
+      "epoch": 1.0931174089068827,
+      "grad_norm": 0.5267510942723349,
+      "learning_rate": 3.831964887155264e-05,
+      "loss": 0.5381,
+      "step": 270
+    },
+    {
+      "epoch": 1.097165991902834,
+      "grad_norm": 0.5839975478173983,
+      "learning_rate": 3.829688502793403e-05,
+      "loss": 0.5326,
+      "step": 271
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.5668082654406333,
+      "learning_rate": 3.827397488291029e-05,
+      "loss": 0.5479,
+      "step": 272
+    },
+    {
+      "epoch": 1.1052631578947367,
+      "grad_norm": 0.5439832952718544,
+      "learning_rate": 3.8250918619670343e-05,
+      "loss": 0.5407,
+      "step": 273
+    },
+    {
+      "epoch": 1.1093117408906883,
+      "grad_norm": 0.5465441966686786,
+      "learning_rate": 3.822771642257145e-05,
+      "loss": 0.5233,
+      "step": 274
+    },
+    {
+      "epoch": 1.1133603238866396,
+      "grad_norm": 0.5650694320325546,
+      "learning_rate": 3.820436847713776e-05,
+      "loss": 0.5332,
+      "step": 275
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 0.4828138408831306,
+      "learning_rate": 3.8180874970058826e-05,
+      "loss": 0.5489,
+      "step": 276
+    },
+    {
+      "epoch": 1.1214574898785425,
+      "grad_norm": 0.5507112996995756,
+      "learning_rate": 3.8157236089188103e-05,
+      "loss": 0.5257,
+      "step": 277
+    },
+    {
+      "epoch": 1.125506072874494,
+      "grad_norm": 0.6123053360496129,
+      "learning_rate": 3.813345202354145e-05,
+      "loss": 0.5297,
+      "step": 278
+    },
+    {
+      "epoch": 1.1295546558704452,
+      "grad_norm": 0.5106294097825206,
+      "learning_rate": 3.810952296329563e-05,
+      "loss": 0.5305,
+      "step": 279
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.6848212136297112,
+      "learning_rate": 3.808544909978676e-05,
+      "loss": 0.5126,
+      "step": 280
+    },
+    {
+      "epoch": 1.1376518218623481,
+      "grad_norm": 0.5703743042303188,
+      "learning_rate": 3.806123062550882e-05,
+      "loss": 0.5275,
+      "step": 281
+    },
+    {
+      "epoch": 1.1417004048582995,
+      "grad_norm": 0.670379014577319,
+      "learning_rate": 3.803686773411208e-05,
+      "loss": 0.5511,
+      "step": 282
+    },
+    {
+      "epoch": 1.145748987854251,
+      "grad_norm": 0.5979660098978319,
+      "learning_rate": 3.801236062040158e-05,
+      "loss": 0.4985,
+      "step": 283
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.5047799118243487,
+      "learning_rate": 3.798770948033553e-05,
+      "loss": 0.4966,
+      "step": 284
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 0.6323960290445433,
+      "learning_rate": 3.796291451102381e-05,
+      "loss": 0.5467,
+      "step": 285
+    },
+    {
+      "epoch": 1.1578947368421053,
+      "grad_norm": 0.7219387870264938,
+      "learning_rate": 3.793797591072631e-05,
+      "loss": 0.5245,
+      "step": 286
+    },
+    {
+      "epoch": 1.1619433198380567,
+      "grad_norm": 0.6965672819064144,
+      "learning_rate": 3.791289387885144e-05,
+      "loss": 0.5186,
+      "step": 287
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.5263505086344046,
+      "learning_rate": 3.788766861595443e-05,
+      "loss": 0.5183,
+      "step": 288
+    },
+    {
+      "epoch": 1.1700404858299596,
+      "grad_norm": 0.7424980630117355,
+      "learning_rate": 3.7862300323735835e-05,
+      "loss": 0.5221,
+      "step": 289
+    },
+    {
+      "epoch": 1.174089068825911,
+      "grad_norm": 0.5576900443726841,
+      "learning_rate": 3.783678920503983e-05,
+      "loss": 0.5016,
+      "step": 290
+    },
+    {
+      "epoch": 1.1781376518218623,
+      "grad_norm": 0.6361515632219613,
+      "learning_rate": 3.781113546385265e-05,
+      "loss": 0.5591,
+      "step": 291
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.6849632785633419,
+      "learning_rate": 3.7785339305300916e-05,
+      "loss": 0.5537,
+      "step": 292
+    },
+    {
+      "epoch": 1.1862348178137652,
+      "grad_norm": 0.5027099620935679,
+      "learning_rate": 3.7759400935650025e-05,
+      "loss": 0.5508,
+      "step": 293
+    },
+    {
+      "epoch": 1.1902834008097165,
+      "grad_norm": 0.6736500443111056,
+      "learning_rate": 3.77333205623025e-05,
+      "loss": 0.5435,
+      "step": 294
+    },
+    {
+      "epoch": 1.194331983805668,
+      "grad_norm": 0.8434622881498836,
+      "learning_rate": 3.7707098393796295e-05,
+      "loss": 0.5608,
+      "step": 295
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.51520286205695,
+      "learning_rate": 3.7680734639803185e-05,
+      "loss": 0.5346,
+      "step": 296
+    },
+    {
+      "epoch": 1.2024291497975708,
+      "grad_norm": 0.697404353628167,
+      "learning_rate": 3.765422951112704e-05,
+      "loss": 0.5103,
+      "step": 297
+    },
+    {
+      "epoch": 1.2064777327935223,
+      "grad_norm": 0.6519737594100977,
+      "learning_rate": 3.762758321970216e-05,
+      "loss": 0.533,
+      "step": 298
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 0.5936436401946393,
+      "learning_rate": 3.7600795978591584e-05,
+      "loss": 0.5361,
+      "step": 299
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.6215893836094512,
+      "learning_rate": 3.757386800198538e-05,
+      "loss": 0.5408,
+      "step": 300
+    },
+    {
+      "epoch": 1.2186234817813766,
+      "grad_norm": 0.6745303864038391,
+      "learning_rate": 3.7546799505198925e-05,
+      "loss": 0.5469,
+      "step": 301
+    },
+    {
+      "epoch": 1.222672064777328,
+      "grad_norm": 0.5860236951492409,
+      "learning_rate": 3.7519590704671197e-05,
+      "loss": 0.5484,
+      "step": 302
+    },
+    {
+      "epoch": 1.2267206477732793,
+      "grad_norm": 0.7094253600276622,
+      "learning_rate": 3.749224181796305e-05,
+      "loss": 0.5484,
+      "step": 303
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.48312927824069474,
+      "learning_rate": 3.746475306375545e-05,
+      "loss": 0.5357,
+      "step": 304
+    },
+    {
+      "epoch": 1.2348178137651822,
+      "grad_norm": 0.7807998382827743,
+      "learning_rate": 3.743712466184774e-05,
+      "loss": 0.5539,
+      "step": 305
+    },
+    {
+      "epoch": 1.2388663967611335,
+      "grad_norm": 0.6038852301634721,
+      "learning_rate": 3.7409356833155885e-05,
+      "loss": 0.5389,
+      "step": 306
+    },
+    {
+      "epoch": 1.242914979757085,
+      "grad_norm": 0.5082385524717066,
+      "learning_rate": 3.73814497997107e-05,
+      "loss": 0.532,
+      "step": 307
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.6634978859854141,
+      "learning_rate": 3.7353403784656095e-05,
+      "loss": 0.5495,
+      "step": 308
+    },
+    {
+      "epoch": 1.2510121457489878,
+      "grad_norm": 0.5082062175235097,
+      "learning_rate": 3.732521901224724e-05,
+      "loss": 0.531,
+      "step": 309
+    },
+    {
+      "epoch": 1.2550607287449393,
+      "grad_norm": 0.6712962090761976,
+      "learning_rate": 3.7296895707848844e-05,
+      "loss": 0.5364,
+      "step": 310
+    },
+    {
+      "epoch": 1.2591093117408907,
+      "grad_norm": 0.6617686763162655,
+      "learning_rate": 3.7268434097933275e-05,
+      "loss": 0.4957,
+      "step": 311
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.48208086851720594,
+      "learning_rate": 3.723983441007882e-05,
+      "loss": 0.519,
+      "step": 312
+    },
+    {
+      "epoch": 1.2672064777327936,
+      "grad_norm": 0.72926668466645,
+      "learning_rate": 3.721109687296781e-05,
+      "loss": 0.5407,
+      "step": 313
+    },
+    {
+      "epoch": 1.271255060728745,
+      "grad_norm": 0.5431984393169822,
+      "learning_rate": 3.718222171638484e-05,
+      "loss": 0.521,
+      "step": 314
+    },
+    {
+      "epoch": 1.2753036437246963,
+      "grad_norm": 0.5717880777518556,
+      "learning_rate": 3.715320917121488e-05,
+      "loss": 0.5275,
+      "step": 315
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.5603549958314208,
+      "learning_rate": 3.712405946944149e-05,
+      "loss": 0.5409,
+      "step": 316
+    },
+    {
+      "epoch": 1.2834008097165992,
+      "grad_norm": 0.4924366504893752,
+      "learning_rate": 3.709477284414488e-05,
+      "loss": 0.539,
+      "step": 317
+    },
+    {
+      "epoch": 1.2874493927125505,
+      "grad_norm": 0.558601630723711,
+      "learning_rate": 3.7065349529500136e-05,
+      "loss": 0.51,
+      "step": 318
+    },
+    {
+      "epoch": 1.291497975708502,
+      "grad_norm": 0.48826494429805534,
+      "learning_rate": 3.7035789760775305e-05,
+      "loss": 0.5258,
+      "step": 319
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.6210737337594463,
+      "learning_rate": 3.70060937743295e-05,
+      "loss": 0.522,
+      "step": 320
+    },
+    {
+      "epoch": 1.2995951417004048,
+      "grad_norm": 0.49192151872485806,
+      "learning_rate": 3.6976261807611045e-05,
+      "loss": 0.533,
+      "step": 321
+    },
+    {
+      "epoch": 1.3036437246963564,
+      "grad_norm": 0.4875619985682551,
+      "learning_rate": 3.694629409915555e-05,
+      "loss": 0.5187,
+      "step": 322
+    },
+    {
+      "epoch": 1.3076923076923077,
+      "grad_norm": 0.49836991837717864,
+      "learning_rate": 3.691619088858401e-05,
+      "loss": 0.4994,
+      "step": 323
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.5268020909283779,
+      "learning_rate": 3.68859524166009e-05,
+      "loss": 0.5242,
+      "step": 324
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 0.5061232725877503,
+      "learning_rate": 3.6855578924992226e-05,
+      "loss": 0.5242,
+      "step": 325
+    },
+    {
+      "epoch": 1.319838056680162,
+      "grad_norm": 0.46071248049215285,
+      "learning_rate": 3.6825070656623626e-05,
+      "loss": 0.5259,
+      "step": 326
+    },
+    {
+      "epoch": 1.3238866396761133,
+      "grad_norm": 0.4973711033272377,
+      "learning_rate": 3.67944278554384e-05,
+      "loss": 0.52,
+      "step": 327
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.4494975042014831,
+      "learning_rate": 3.676365076645557e-05,
+      "loss": 0.5307,
+      "step": 328
+    },
+    {
+      "epoch": 1.3319838056680162,
+      "grad_norm": 0.523912475305319,
+      "learning_rate": 3.673273963576791e-05,
+      "loss": 0.5351,
+      "step": 329
+    },
+    {
+      "epoch": 1.3360323886639676,
+      "grad_norm": 0.5176831690924304,
+      "learning_rate": 3.670169471054001e-05,
+      "loss": 0.5237,
+      "step": 330
+    },
+    {
+      "epoch": 1.3400809716599191,
+      "grad_norm": 0.5382725087263432,
+      "learning_rate": 3.6670516239006254e-05,
+      "loss": 0.5166,
+      "step": 331
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.5226458334680275,
+      "learning_rate": 3.663920447046888e-05,
+      "loss": 0.5345,
+      "step": 332
+    },
+    {
+      "epoch": 1.3481781376518218,
+      "grad_norm": 0.47827270380619347,
+      "learning_rate": 3.6607759655295954e-05,
+      "loss": 0.5183,
+      "step": 333
+    },
+    {
+      "epoch": 1.3522267206477734,
+      "grad_norm": 0.5007339804364864,
+      "learning_rate": 3.657618204491937e-05,
+      "loss": 0.5392,
+      "step": 334
+    },
+    {
+      "epoch": 1.3562753036437247,
+      "grad_norm": 0.4383982416409244,
+      "learning_rate": 3.6544471891832875e-05,
+      "loss": 0.5427,
+      "step": 335
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 0.4724318120082341,
+      "learning_rate": 3.6512629449589983e-05,
+      "loss": 0.53,
+      "step": 336
+    },
+    {
+      "epoch": 1.3643724696356276,
+      "grad_norm": 0.47489765653965577,
+      "learning_rate": 3.6480654972802045e-05,
+      "loss": 0.539,
+      "step": 337
+    },
+    {
+      "epoch": 1.368421052631579,
+      "grad_norm": 0.5122822458601235,
+      "learning_rate": 3.644854871713611e-05,
+      "loss": 0.5197,
+      "step": 338
+    },
+    {
+      "epoch": 1.3724696356275303,
+      "grad_norm": 0.5729916698266643,
+      "learning_rate": 3.6416310939312924e-05,
+      "loss": 0.5602,
+      "step": 339
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.4348883792052224,
+      "learning_rate": 3.638394189710493e-05,
+      "loss": 0.5245,
+      "step": 340
+    },
+    {
+      "epoch": 1.3805668016194332,
+      "grad_norm": 0.7044938074259033,
+      "learning_rate": 3.635144184933412e-05,
+      "loss": 0.5307,
+      "step": 341
+    },
+    {
+      "epoch": 1.3846153846153846,
+      "grad_norm": 0.5377533633081365,
+      "learning_rate": 3.6318811055869994e-05,
+      "loss": 0.5277,
+      "step": 342
+    },
+    {
+      "epoch": 1.3886639676113361,
+      "grad_norm": 0.5200965699622784,
+      "learning_rate": 3.628604977762752e-05,
+      "loss": 0.5167,
+      "step": 343
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.5316909728762006,
+      "learning_rate": 3.6253158276565004e-05,
+      "loss": 0.5041,
+      "step": 344
+    },
+    {
+      "epoch": 1.3967611336032388,
+      "grad_norm": 0.4723095663194227,
+      "learning_rate": 3.6220136815682016e-05,
+      "loss": 0.5593,
+      "step": 345
+    },
+    {
+      "epoch": 1.4008097165991904,
+      "grad_norm": 0.5174929397990018,
+      "learning_rate": 3.618698565901727e-05,
+      "loss": 0.5534,
+      "step": 346
+    },
+    {
+      "epoch": 1.4048582995951417,
+      "grad_norm": 0.5108027539620703,
+      "learning_rate": 3.615370507164654e-05,
+      "loss": 0.5526,
+      "step": 347
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.4737996425677972,
+      "learning_rate": 3.61202953196805e-05,
+      "loss": 0.5212,
+      "step": 348
+    },
+    {
+      "epoch": 1.4129554655870447,
+      "grad_norm": 0.4691946513620876,
+      "learning_rate": 3.6086756670262654e-05,
+      "loss": 0.5508,
+      "step": 349
+    },
+    {
+      "epoch": 1.417004048582996,
+      "grad_norm": 0.5432911233862397,
+      "learning_rate": 3.6053089391567135e-05,
+      "loss": 0.5473,
+      "step": 350
+    },
+    {
+      "epoch": 1.4210526315789473,
+      "grad_norm": 0.49501180426517494,
+      "learning_rate": 3.601929375279662e-05,
+      "loss": 0.5361,
+      "step": 351
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.5392680880636919,
+      "learning_rate": 3.598537002418012e-05,
+      "loss": 0.5322,
+      "step": 352
+    },
+    {
+      "epoch": 1.4291497975708503,
+      "grad_norm": 0.5342808101242759,
+      "learning_rate": 3.5951318476970885e-05,
+      "loss": 0.5455,
+      "step": 353
+    },
+    {
+      "epoch": 1.4331983805668016,
+      "grad_norm": 0.4915749949659902,
+      "learning_rate": 3.591713938344416e-05,
+      "loss": 0.538,
+      "step": 354
+    },
+    {
+      "epoch": 1.4372469635627532,
+      "grad_norm": 0.46491403185612135,
+      "learning_rate": 3.588283301689507e-05,
+      "loss": 0.524,
+      "step": 355
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.47067929555553195,
+      "learning_rate": 3.5848399651636424e-05,
+      "loss": 0.5288,
+      "step": 356
+    },
+    {
+      "epoch": 1.4453441295546559,
+      "grad_norm": 0.4794749215836955,
+      "learning_rate": 3.5813839562996473e-05,
+      "loss": 0.5333,
+      "step": 357
+    },
+    {
+      "epoch": 1.4493927125506074,
+      "grad_norm": 0.4746251067209958,
+      "learning_rate": 3.5779153027316775e-05,
+      "loss": 0.5287,
+      "step": 358
+    },
+    {
+      "epoch": 1.4534412955465588,
+      "grad_norm": 0.4480134802570351,
+      "learning_rate": 3.574434032194994e-05,
+      "loss": 0.5206,
+      "step": 359
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.5013868130339115,
+      "learning_rate": 3.5709401725257436e-05,
+      "loss": 0.5179,
+      "step": 360
+    },
+    {
+      "epoch": 1.4615384615384617,
+      "grad_norm": 0.4737563950493894,
+      "learning_rate": 3.5674337516607346e-05,
+      "loss": 0.5418,
+      "step": 361
+    },
+    {
+      "epoch": 1.465587044534413,
+      "grad_norm": 0.45356308810008394,
+      "learning_rate": 3.563914797637216e-05,
+      "loss": 0.5188,
+      "step": 362
+    },
+    {
+      "epoch": 1.4696356275303644,
+      "grad_norm": 0.4768737923054209,
+      "learning_rate": 3.560383338592649e-05,
+      "loss": 0.5222,
+      "step": 363
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.4797135659769926,
+      "learning_rate": 3.556839402764487e-05,
+      "loss": 0.5403,
+      "step": 364
+    },
+    {
+      "epoch": 1.4777327935222673,
+      "grad_norm": 0.4615469025038061,
+      "learning_rate": 3.553283018489946e-05,
+      "loss": 0.546,
+      "step": 365
+    },
+    {
+      "epoch": 1.4817813765182186,
+      "grad_norm": 0.48826814571861193,
+      "learning_rate": 3.54971421420578e-05,
+      "loss": 0.5381,
+      "step": 366
+    },
+    {
+      "epoch": 1.48582995951417,
+      "grad_norm": 0.45740978609302463,
+      "learning_rate": 3.5461330184480524e-05,
+      "loss": 0.5398,
+      "step": 367
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.5431212259922498,
+      "learning_rate": 3.54253945985191e-05,
+      "loss": 0.5335,
+      "step": 368
+    },
+    {
+      "epoch": 1.4939271255060729,
+      "grad_norm": 0.4910403641697464,
+      "learning_rate": 3.5389335671513524e-05,
+      "loss": 0.5142,
+      "step": 369
+    },
+    {
+      "epoch": 1.4979757085020242,
+      "grad_norm": 0.5099237178006697,
+      "learning_rate": 3.5353153691789986e-05,
+      "loss": 0.5315,
+      "step": 370
+    },
+    {
+      "epoch": 1.5020242914979756,
+      "grad_norm": 0.4860992129906428,
+      "learning_rate": 3.531684894865866e-05,
+      "loss": 0.5399,
+      "step": 371
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.4822395464381007,
+      "learning_rate": 3.528042173241128e-05,
+      "loss": 0.5607,
+      "step": 372
+    },
+    {
+      "epoch": 1.5101214574898787,
+      "grad_norm": 0.5260992569252276,
+      "learning_rate": 3.524387233431892e-05,
+      "loss": 0.5435,
+      "step": 373
+    },
+    {
+      "epoch": 1.5141700404858298,
+      "grad_norm": 0.5754186536333304,
+      "learning_rate": 3.520720104662958e-05,
+      "loss": 0.5311,
+      "step": 374
+    },
+    {
+      "epoch": 1.5182186234817814,
+      "grad_norm": 0.46476285893812064,
+      "learning_rate": 3.5170408162565904e-05,
+      "loss": 0.5465,
+      "step": 375
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.5395589560177354,
+      "learning_rate": 3.513349397632281e-05,
+      "loss": 0.5324,
+      "step": 376
+    },
+    {
+      "epoch": 1.526315789473684,
+      "grad_norm": 0.4772085901155437,
+      "learning_rate": 3.5096458783065145e-05,
+      "loss": 0.5562,
+      "step": 377
+    },
+    {
+      "epoch": 1.5303643724696356,
+      "grad_norm": 0.5313339181523234,
+      "learning_rate": 3.505930287892533e-05,
+      "loss": 0.551,
+      "step": 378
+    },
+    {
+      "epoch": 1.5344129554655872,
+      "grad_norm": 0.4656486776960399,
+      "learning_rate": 3.502202656100099e-05,
+      "loss": 0.5536,
+      "step": 379
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.4743401158353235,
+      "learning_rate": 3.498463012735256e-05,
+      "loss": 0.5221,
+      "step": 380
+    },
+    {
+      "epoch": 1.54251012145749,
+      "grad_norm": 0.5276231802662612,
+      "learning_rate": 3.494711387700094e-05,
+      "loss": 0.5138,
+      "step": 381
+    },
+    {
+      "epoch": 1.5465587044534415,
+      "grad_norm": 0.5019377787178595,
+      "learning_rate": 3.490947810992505e-05,
+      "loss": 0.5339,
+      "step": 382
+    },
+    {
+      "epoch": 1.5506072874493926,
+      "grad_norm": 1.0682204116657714,
+      "learning_rate": 3.48717231270595e-05,
+      "loss": 0.5716,
+      "step": 383
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.46604551843650627,
+      "learning_rate": 3.483384923029211e-05,
+      "loss": 0.5101,
+      "step": 384
+    },
+    {
+      "epoch": 1.5587044534412957,
+      "grad_norm": 0.5218757379702306,
+      "learning_rate": 3.479585672246156e-05,
+      "loss": 0.5346,
+      "step": 385
+    },
+    {
+      "epoch": 1.5627530364372468,
+      "grad_norm": 0.4405270455171088,
+      "learning_rate": 3.4757745907354924e-05,
+      "loss": 0.5255,
+      "step": 386
+    },
+    {
+      "epoch": 1.5668016194331984,
+      "grad_norm": 0.5027368776284724,
+      "learning_rate": 3.471951708970527e-05,
+      "loss": 0.5254,
+      "step": 387
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.452510510014909,
+      "learning_rate": 3.468117057518921e-05,
+      "loss": 0.5478,
+      "step": 388
+    },
+    {
+      "epoch": 1.574898785425101,
+      "grad_norm": 0.4855386111867917,
+      "learning_rate": 3.4642706670424426e-05,
+      "loss": 0.5365,
+      "step": 389
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 0.44060791235818486,
+      "learning_rate": 3.460412568296731e-05,
+      "loss": 0.5368,
+      "step": 390
+    },
+    {
+      "epoch": 1.582995951417004,
+      "grad_norm": 0.42220662765445227,
+      "learning_rate": 3.456542792131039e-05,
+      "loss": 0.5424,
+      "step": 391
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.40767561908199257,
+      "learning_rate": 3.4526613694879936e-05,
+      "loss": 0.5409,
+      "step": 392
+    },
+    {
+      "epoch": 1.591093117408907,
+      "grad_norm": 0.49951474821568015,
+      "learning_rate": 3.448768331403348e-05,
+      "loss": 0.5328,
+      "step": 393
+    },
+    {
+      "epoch": 1.5951417004048583,
+      "grad_norm": 0.42417108033193907,
+      "learning_rate": 3.44486370900573e-05,
+      "loss": 0.5348,
+      "step": 394
+    },
+    {
+      "epoch": 1.5991902834008096,
+      "grad_norm": 0.5759220817999551,
+      "learning_rate": 3.440947533516396e-05,
+      "loss": 0.5739,
+      "step": 395
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 0.3985808048455658,
+      "learning_rate": 3.4370198362489816e-05,
+      "loss": 0.5211,
+      "step": 396
+    },
+    {
+      "epoch": 1.6072874493927125,
+      "grad_norm": 0.5718906980315263,
+      "learning_rate": 3.43308064860925e-05,
+      "loss": 0.5343,
+      "step": 397
+    },
+    {
+      "epoch": 1.6113360323886639,
+      "grad_norm": 0.4074880680993602,
+      "learning_rate": 3.4291300020948393e-05,
+      "loss": 0.5443,
+      "step": 398
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 0.47928470174602217,
+      "learning_rate": 3.425167928295015e-05,
+      "loss": 0.5309,
+      "step": 399
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 0.4553372202130343,
+      "learning_rate": 3.421194458890411e-05,
+      "loss": 0.5356,
+      "step": 400
+    },
+    {
+      "epoch": 1.623481781376518,
+      "grad_norm": 0.4285842848171338,
+      "learning_rate": 3.4172096256527865e-05,
+      "loss": 0.5444,
+      "step": 401
+    },
+    {
+      "epoch": 1.6275303643724697,
+      "grad_norm": 0.4217660839697807,
+      "learning_rate": 3.4132134604447595e-05,
+      "loss": 0.5362,
+      "step": 402
+    },
+    {
+      "epoch": 1.631578947368421,
+      "grad_norm": 0.41828599328908667,
+      "learning_rate": 3.409205995219561e-05,
+      "loss": 0.55,
+      "step": 403
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.38449351301794954,
+      "learning_rate": 3.4051872620207765e-05,
+      "loss": 0.5134,
+      "step": 404
+    },
+    {
+      "epoch": 1.639676113360324,
+      "grad_norm": 0.48103069694686246,
+      "learning_rate": 3.40115729298209e-05,
+      "loss": 0.5143,
+      "step": 405
+    },
+    {
+      "epoch": 1.6437246963562753,
+      "grad_norm": 0.4343605079172865,
+      "learning_rate": 3.397116120327025e-05,
+      "loss": 0.5435,
+      "step": 406
+    },
+    {
+      "epoch": 1.6477732793522266,
+      "grad_norm": 0.4766994758910087,
+      "learning_rate": 3.3930637763686916e-05,
+      "loss": 0.5161,
+      "step": 407
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.44047706077946885,
+      "learning_rate": 3.389000293509524e-05,
+      "loss": 0.5546,
+      "step": 408
+    },
+    {
+      "epoch": 1.6558704453441295,
+      "grad_norm": 0.4660326391212398,
+      "learning_rate": 3.3849257042410225e-05,
+      "loss": 0.521,
+      "step": 409
+    },
+    {
+      "epoch": 1.6599190283400809,
+      "grad_norm": 0.40303995007330917,
+      "learning_rate": 3.380840041143494e-05,
+      "loss": 0.5358,
+      "step": 410
+    },
+    {
+      "epoch": 1.6639676113360324,
+      "grad_norm": 0.4720504826005394,
+      "learning_rate": 3.376743336885792e-05,
+      "loss": 0.5361,
+      "step": 411
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.4409080194126786,
+      "learning_rate": 3.372635624225055e-05,
+      "loss": 0.5422,
+      "step": 412
+    },
+    {
+      "epoch": 1.6720647773279351,
+      "grad_norm": 0.423340268304851,
+      "learning_rate": 3.3685169360064445e-05,
+      "loss": 0.5363,
+      "step": 413
+    },
+    {
+      "epoch": 1.6761133603238867,
+      "grad_norm": 0.4747419382059854,
+      "learning_rate": 3.3643873051628804e-05,
+      "loss": 0.5557,
+      "step": 414
+    },
+    {
+      "epoch": 1.680161943319838,
+      "grad_norm": 0.42466832527505677,
+      "learning_rate": 3.3602467647147814e-05,
+      "loss": 0.5192,
+      "step": 415
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.5480097106130704,
+      "learning_rate": 3.356095347769799e-05,
+      "loss": 0.5324,
+      "step": 416
+    },
+    {
+      "epoch": 1.688259109311741,
+      "grad_norm": 0.442666036158379,
+      "learning_rate": 3.351933087522552e-05,
+      "loss": 0.5475,
+      "step": 417
+    },
+    {
+      "epoch": 1.6923076923076923,
+      "grad_norm": 0.5288563549364114,
+      "learning_rate": 3.3477600172543626e-05,
+      "loss": 0.5285,
+      "step": 418
+    },
+    {
+      "epoch": 1.6963562753036436,
+      "grad_norm": 0.45075613835980655,
+      "learning_rate": 3.3435761703329894e-05,
+      "loss": 0.5065,
+      "step": 419
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.446017394628175,
+      "learning_rate": 3.3393815802123606e-05,
+      "loss": 0.5213,
+      "step": 420
+    },
+    {
+      "epoch": 1.7044534412955465,
+      "grad_norm": 0.4650427341551755,
+      "learning_rate": 3.335176280432308e-05,
+      "loss": 0.5366,
+      "step": 421
+    },
+    {
+      "epoch": 1.708502024291498,
+      "grad_norm": 0.4939349114164575,
+      "learning_rate": 3.330960304618295e-05,
+      "loss": 0.5254,
+      "step": 422
+    },
+    {
+      "epoch": 1.7125506072874495,
+      "grad_norm": 0.5190421846063528,
+      "learning_rate": 3.326733686481153e-05,
+      "loss": 0.5368,
+      "step": 423
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.40575166223104214,
+      "learning_rate": 3.3224964598168084e-05,
+      "loss": 0.5338,
+      "step": 424
+    },
+    {
+      "epoch": 1.7206477732793521,
+      "grad_norm": 0.44855104960945497,
+      "learning_rate": 3.318248658506012e-05,
+      "loss": 0.5405,
+      "step": 425
+    },
+    {
+      "epoch": 1.7246963562753037,
+      "grad_norm": 0.5261902377198243,
+      "learning_rate": 3.313990316514069e-05,
+      "loss": 0.5547,
+      "step": 426
+    },
+    {
+      "epoch": 1.728744939271255,
+      "grad_norm": 0.42555882016161606,
+      "learning_rate": 3.309721467890571e-05,
+      "loss": 0.5321,
+      "step": 427
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.49613386594983977,
+      "learning_rate": 3.305442146769115e-05,
+      "loss": 0.5348,
+      "step": 428
+    },
+    {
+      "epoch": 1.736842105263158,
+      "grad_norm": 0.3897846667974609,
+      "learning_rate": 3.30115238736704e-05,
+      "loss": 0.5288,
+      "step": 429
+    },
+    {
+      "epoch": 1.7408906882591093,
+      "grad_norm": 0.49493793212625264,
+      "learning_rate": 3.296852223985148e-05,
+      "loss": 0.555,
+      "step": 430
+    },
+    {
+      "epoch": 1.7449392712550607,
+      "grad_norm": 0.44409743455735307,
+      "learning_rate": 3.2925416910074315e-05,
+      "loss": 0.5248,
+      "step": 431
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.4609002027880675,
+      "learning_rate": 3.288220822900796e-05,
+      "loss": 0.5328,
+      "step": 432
+    },
+    {
+      "epoch": 1.7530364372469636,
+      "grad_norm": 0.4165254651296752,
+      "learning_rate": 3.283889654214788e-05,
+      "loss": 0.5384,
+      "step": 433
+    },
+    {
+      "epoch": 1.757085020242915,
+      "grad_norm": 0.4644950821658569,
+      "learning_rate": 3.279548219581317e-05,
+      "loss": 0.5385,
+      "step": 434
+    },
+    {
+      "epoch": 1.7611336032388665,
+      "grad_norm": 0.41995445004173954,
+      "learning_rate": 3.275196553714379e-05,
+      "loss": 0.5181,
+      "step": 435
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.446012524868916,
+      "learning_rate": 3.270834691409778e-05,
+      "loss": 0.5405,
+      "step": 436
+    },
+    {
+      "epoch": 1.7692307692307692,
+      "grad_norm": 0.4041687353183538,
+      "learning_rate": 3.266462667544849e-05,
+      "loss": 0.5353,
+      "step": 437
+    },
+    {
+      "epoch": 1.7732793522267207,
+      "grad_norm": 0.3935113329527631,
+      "learning_rate": 3.262080517078178e-05,
+      "loss": 0.5169,
+      "step": 438
+    },
+    {
+      "epoch": 1.777327935222672,
+      "grad_norm": 0.40563671343855456,
+      "learning_rate": 3.257688275049323e-05,
+      "loss": 0.536,
+      "step": 439
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.43457328789830585,
+      "learning_rate": 3.253285976578535e-05,
+      "loss": 0.5427,
+      "step": 440
+    },
+    {
+      "epoch": 1.785425101214575,
+      "grad_norm": 0.4695809041454866,
+      "learning_rate": 3.2488736568664756e-05,
+      "loss": 0.5291,
+      "step": 441
+    },
+    {
+      "epoch": 1.7894736842105263,
+      "grad_norm": 0.41972682512426773,
+      "learning_rate": 3.244451351193935e-05,
+      "loss": 0.5512,
+      "step": 442
+    },
+    {
+      "epoch": 1.7935222672064777,
+      "grad_norm": 0.4163872416441201,
+      "learning_rate": 3.240019094921551e-05,
+      "loss": 0.5013,
+      "step": 443
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.4617479815824359,
+      "learning_rate": 3.235576923489527e-05,
+      "loss": 0.5283,
+      "step": 444
+    },
+    {
+      "epoch": 1.8016194331983806,
+      "grad_norm": 0.46656618379825016,
+      "learning_rate": 3.2311248724173476e-05,
+      "loss": 0.5204,
+      "step": 445
+    },
+    {
+      "epoch": 1.805668016194332,
+      "grad_norm": 0.48992078049547383,
+      "learning_rate": 3.226662977303494e-05,
+      "loss": 0.5369,
+      "step": 446
+    },
+    {
+      "epoch": 1.8097165991902835,
+      "grad_norm": 0.4647763254795458,
+      "learning_rate": 3.2221912738251626e-05,
+      "loss": 0.5325,
+      "step": 447
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.435881805600712,
+      "learning_rate": 3.2177097977379736e-05,
+      "loss": 0.524,
+      "step": 448
+    },
+    {
+      "epoch": 1.8178137651821862,
+      "grad_norm": 0.4588138265170988,
+      "learning_rate": 3.21321858487569e-05,
+      "loss": 0.5137,
+      "step": 449
+    },
+    {
+      "epoch": 1.8218623481781377,
+      "grad_norm": 0.49094227019470393,
+      "learning_rate": 3.208717671149931e-05,
+      "loss": 0.5205,
+      "step": 450
+    },
+    {
+      "epoch": 1.825910931174089,
+      "grad_norm": 0.4359723888866673,
+      "learning_rate": 3.204207092549882e-05,
+      "loss": 0.5448,
+      "step": 451
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.4444417065990943,
+      "learning_rate": 3.19968688514201e-05,
+      "loss": 0.5404,
+      "step": 452
+    },
+    {
+      "epoch": 1.834008097165992,
+      "grad_norm": 0.42777110674199614,
+      "learning_rate": 3.195157085069774e-05,
+      "loss": 0.5288,
+      "step": 453
+    },
+    {
+      "epoch": 1.8380566801619433,
+      "grad_norm": 0.4485039634551085,
+      "learning_rate": 3.190617728553332e-05,
+      "loss": 0.5272,
+      "step": 454
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 0.41718915064471024,
+      "learning_rate": 3.1860688518892606e-05,
+      "loss": 0.5211,
+      "step": 455
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.44304618045421507,
+      "learning_rate": 3.181510491450254e-05,
+      "loss": 0.5195,
+      "step": 456
+    },
+    {
+      "epoch": 1.8502024291497976,
+      "grad_norm": 0.3997067569921349,
+      "learning_rate": 3.176942683684842e-05,
+      "loss": 0.5222,
+      "step": 457
+    },
+    {
+      "epoch": 1.854251012145749,
+      "grad_norm": 0.4655460351617563,
+      "learning_rate": 3.1723654651170934e-05,
+      "loss": 0.5299,
+      "step": 458
+    },
+    {
+      "epoch": 1.8582995951417005,
+      "grad_norm": 0.42999246518410134,
+      "learning_rate": 3.167778872346327e-05,
+      "loss": 0.542,
+      "step": 459
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.5024271784586793,
+      "learning_rate": 3.163182942046815e-05,
+      "loss": 0.5443,
+      "step": 460
+    },
+    {
+      "epoch": 1.8663967611336032,
+      "grad_norm": 0.4326786438431615,
+      "learning_rate": 3.158577710967494e-05,
+      "loss": 0.5458,
+      "step": 461
+    },
+    {
+      "epoch": 1.8704453441295548,
+      "grad_norm": 0.5058759374186524,
+      "learning_rate": 3.153963215931669e-05,
+      "loss": 0.5329,
+      "step": 462
+    },
+    {
+      "epoch": 1.874493927125506,
+      "grad_norm": 0.4780384842674827,
+      "learning_rate": 3.14933949383672e-05,
+      "loss": 0.529,
+      "step": 463
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.44056487764517255,
+      "learning_rate": 3.144706581653805e-05,
+      "loss": 0.5207,
+      "step": 464
+    },
+    {
+      "epoch": 1.882591093117409,
+      "grad_norm": 0.4628572912159805,
+      "learning_rate": 3.140064516427566e-05,
+      "loss": 0.5121,
+      "step": 465
+    },
+    {
+      "epoch": 1.8866396761133604,
+      "grad_norm": 0.46392156579652216,
+      "learning_rate": 3.135413335275833e-05,
+      "loss": 0.5307,
+      "step": 466
+    },
+    {
+      "epoch": 1.8906882591093117,
+      "grad_norm": 0.4792493873875201,
+      "learning_rate": 3.130753075389327e-05,
+      "loss": 0.5472,
+      "step": 467
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.4632003681461716,
+      "learning_rate": 3.1260837740313625e-05,
+      "loss": 0.537,
+      "step": 468
+    },
+    {
+      "epoch": 1.8987854251012146,
+      "grad_norm": 0.5084506703056955,
+      "learning_rate": 3.121405468537547e-05,
+      "loss": 0.538,
+      "step": 469
+    },
+    {
+      "epoch": 1.902834008097166,
+      "grad_norm": 0.43786513250236253,
+      "learning_rate": 3.116718196315488e-05,
+      "loss": 0.5207,
+      "step": 470
+    },
+    {
+      "epoch": 1.9068825910931175,
+      "grad_norm": 0.46152265698536155,
+      "learning_rate": 3.1120219948444894e-05,
+      "loss": 0.5112,
+      "step": 471
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.5105585523492562,
+      "learning_rate": 3.107316901675252e-05,
+      "loss": 0.5375,
+      "step": 472
+    },
+    {
+      "epoch": 1.9149797570850202,
+      "grad_norm": 0.4624151832809304,
+      "learning_rate": 3.1026029544295745e-05,
+      "loss": 0.5224,
+      "step": 473
+    },
+    {
+      "epoch": 1.9190283400809718,
+      "grad_norm": 0.5005853949946278,
+      "learning_rate": 3.097880190800057e-05,
+      "loss": 0.5224,
+      "step": 474
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 0.4962370915219149,
+      "learning_rate": 3.093148648549788e-05,
+      "loss": 0.5459,
+      "step": 475
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.5399218662550661,
+      "learning_rate": 3.088408365512055e-05,
+      "loss": 0.5173,
+      "step": 476
+    },
+    {
+      "epoch": 1.931174089068826,
+      "grad_norm": 0.4212877853334291,
+      "learning_rate": 3.083659379590034e-05,
+      "loss": 0.5261,
+      "step": 477
+    },
+    {
+      "epoch": 1.9352226720647774,
+      "grad_norm": 0.4842189471683288,
+      "learning_rate": 3.0789017287564913e-05,
+      "loss": 0.5382,
+      "step": 478
+    },
+    {
+      "epoch": 1.9392712550607287,
+      "grad_norm": 0.4499838894691206,
+      "learning_rate": 3.074135451053475e-05,
+      "loss": 0.5235,
+      "step": 479
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.4551219300997529,
+      "learning_rate": 3.069360584592015e-05,
+      "loss": 0.5115,
+      "step": 480
+    },
+    {
+      "epoch": 1.9473684210526314,
+      "grad_norm": 0.46483833840466016,
+      "learning_rate": 3.064577167551816e-05,
+      "loss": 0.5418,
+      "step": 481
+    },
+    {
+      "epoch": 1.951417004048583,
+      "grad_norm": 0.5350476695461692,
+      "learning_rate": 3.059785238180953e-05,
+      "loss": 0.544,
+      "step": 482
+    },
+    {
+      "epoch": 1.9554655870445345,
+      "grad_norm": 0.48644777733467087,
+      "learning_rate": 3.054984834795566e-05,
+      "loss": 0.531,
+      "step": 483
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.534690649603997,
+      "learning_rate": 3.0501759957795537e-05,
+      "loss": 0.5339,
+      "step": 484
+    },
+    {
+      "epoch": 1.9635627530364372,
+      "grad_norm": 0.4982167462760593,
+      "learning_rate": 3.0453587595842627e-05,
+      "loss": 0.5048,
+      "step": 485
+    },
+    {
+      "epoch": 1.9676113360323888,
+      "grad_norm": 0.5128498600055258,
+      "learning_rate": 3.0405331647281856e-05,
+      "loss": 0.5297,
+      "step": 486
+    },
+    {
+      "epoch": 1.97165991902834,
+      "grad_norm": 0.5119602037634,
+      "learning_rate": 3.0356992497966508e-05,
+      "loss": 0.5097,
+      "step": 487
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.487523454257288,
+      "learning_rate": 3.0308570534415137e-05,
+      "loss": 0.5319,
+      "step": 488
+    },
+    {
+      "epoch": 1.979757085020243,
+      "grad_norm": 0.47419986995470875,
+      "learning_rate": 3.0260066143808467e-05,
+      "loss": 0.5462,
+      "step": 489
+    },
+    {
+      "epoch": 1.9838056680161942,
+      "grad_norm": 0.44430547209842625,
+      "learning_rate": 3.021147971398632e-05,
+      "loss": 0.5117,
+      "step": 490
+    },
+    {
+      "epoch": 1.9878542510121457,
+      "grad_norm": 0.46907080688386793,
+      "learning_rate": 3.0162811633444478e-05,
+      "loss": 0.5242,
+      "step": 491
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.4183245669671288,
+      "learning_rate": 3.0114062291331635e-05,
+      "loss": 0.5546,
+      "step": 492
+    },
+    {
+      "epoch": 1.9959514170040484,
+      "grad_norm": 0.5213485084854363,
+      "learning_rate": 3.0065232077446217e-05,
+      "loss": 0.5261,
+      "step": 493
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.4179329815042241,
+      "learning_rate": 3.001632138223332e-05,
+      "loss": 0.5274,
+      "step": 494
+    },
+    {
+      "epoch": 2.0040485829959516,
+      "grad_norm": 0.7126563355882187,
+      "learning_rate": 2.9967330596781565e-05,
+      "loss": 0.3923,
+      "step": 495
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.4666598406706184,
+      "learning_rate": 2.9918260112819952e-05,
+      "loss": 0.3963,
+      "step": 496
+    },
+    {
+      "epoch": 2.0121457489878543,
+      "grad_norm": 1.150196686685127,
+      "learning_rate": 2.9869110322714778e-05,
+      "loss": 0.402,
+      "step": 497
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 0.5995387725064423,
+      "learning_rate": 2.9819881619466447e-05,
+      "loss": 0.3851,
+      "step": 498
+    },
+    {
+      "epoch": 2.020242914979757,
+      "grad_norm": 0.6271500140883626,
+      "learning_rate": 2.9770574396706364e-05,
+      "loss": 0.3891,
+      "step": 499
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.5981054419372982,
+      "learning_rate": 2.972118904869376e-05,
+      "loss": 0.3914,
+      "step": 500
+    },
+    {
+      "epoch": 2.02834008097166,
+      "grad_norm": 0.643740989305513,
+      "learning_rate": 2.9671725970312563e-05,
+      "loss": 0.3927,
+      "step": 501
+    },
+    {
+      "epoch": 2.032388663967611,
+      "grad_norm": 0.49338747151997125,
+      "learning_rate": 2.9622185557068238e-05,
+      "loss": 0.3856,
+      "step": 502
+    },
+    {
+      "epoch": 2.0364372469635628,
+      "grad_norm": 0.6346352663801714,
+      "learning_rate": 2.95725682050846e-05,
+      "loss": 0.3835,
+      "step": 503
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 0.480283421710661,
+      "learning_rate": 2.9522874311100684e-05,
+      "loss": 0.3946,
+      "step": 504
+    },
+    {
+      "epoch": 2.0445344129554655,
+      "grad_norm": 0.5831758692664536,
+      "learning_rate": 2.9473104272467526e-05,
+      "loss": 0.3881,
+      "step": 505
+    },
+    {
+      "epoch": 2.048582995951417,
+      "grad_norm": 0.4887534598122495,
+      "learning_rate": 2.942325848714505e-05,
+      "loss": 0.3735,
+      "step": 506
+    },
+    {
+      "epoch": 2.0526315789473686,
+      "grad_norm": 0.5387801546483969,
+      "learning_rate": 2.9373337353698804e-05,
+      "loss": 0.375,
+      "step": 507
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.5061909057084246,
+      "learning_rate": 2.932334127129686e-05,
+      "loss": 0.4004,
+      "step": 508
+    },
+    {
+      "epoch": 2.0607287449392713,
+      "grad_norm": 0.5517067167488872,
+      "learning_rate": 2.9273270639706547e-05,
+      "loss": 0.3811,
+      "step": 509
+    },
+    {
+      "epoch": 2.064777327935223,
+      "grad_norm": 0.5234995991840693,
+      "learning_rate": 2.922312585929131e-05,
+      "loss": 0.3787,
+      "step": 510
+    },
+    {
+      "epoch": 2.068825910931174,
+      "grad_norm": 0.49070867066165724,
+      "learning_rate": 2.9172907331007478e-05,
+      "loss": 0.3897,
+      "step": 511
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.5263450714303431,
+      "learning_rate": 2.9122615456401053e-05,
+      "loss": 0.3932,
+      "step": 512
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.5193344823378208,
+      "learning_rate": 2.9072250637604535e-05,
+      "loss": 0.366,
+      "step": 513
+    },
+    {
+      "epoch": 2.080971659919028,
+      "grad_norm": 0.49859664876995685,
+      "learning_rate": 2.9021813277333667e-05,
+      "loss": 0.3769,
+      "step": 514
+    },
+    {
+      "epoch": 2.08502024291498,
+      "grad_norm": 0.5604679074251347,
+      "learning_rate": 2.8971303778884244e-05,
+      "loss": 0.3683,
+      "step": 515
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.446236881423059,
+      "learning_rate": 2.8920722546128863e-05,
+      "loss": 0.369,
+      "step": 516
+    },
+    {
+      "epoch": 2.0931174089068825,
+      "grad_norm": 0.4902416756265236,
+      "learning_rate": 2.887006998351372e-05,
+      "loss": 0.3814,
+      "step": 517
+    },
+    {
+      "epoch": 2.097165991902834,
+      "grad_norm": 0.5310740931749722,
+      "learning_rate": 2.881934649605535e-05,
+      "loss": 0.3735,
+      "step": 518
+    },
+    {
+      "epoch": 2.1012145748987856,
+      "grad_norm": 0.47134254456792773,
+      "learning_rate": 2.8768552489337408e-05,
+      "loss": 0.4116,
+      "step": 519
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 0.5155322292141339,
+      "learning_rate": 2.8717688369507425e-05,
+      "loss": 0.3834,
+      "step": 520
+    },
+    {
+      "epoch": 2.1093117408906883,
+      "grad_norm": 0.4290624695132283,
+      "learning_rate": 2.8666754543273535e-05,
+      "loss": 0.405,
+      "step": 521
+    },
+    {
+      "epoch": 2.11336032388664,
+      "grad_norm": 0.4678692986914002,
+      "learning_rate": 2.8615751417901273e-05,
+      "loss": 0.3811,
+      "step": 522
+    },
+    {
+      "epoch": 2.117408906882591,
+      "grad_norm": 0.4011726249913474,
+      "learning_rate": 2.856467940121025e-05,
+      "loss": 0.3805,
+      "step": 523
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.5453591037233225,
+      "learning_rate": 2.8513538901570964e-05,
+      "loss": 0.3608,
+      "step": 524
+    },
+    {
+      "epoch": 2.125506072874494,
+      "grad_norm": 0.42600694848450493,
+      "learning_rate": 2.846233032790149e-05,
+      "loss": 0.3971,
+      "step": 525
+    },
+    {
+      "epoch": 2.1295546558704452,
+      "grad_norm": 0.5008177614326356,
+      "learning_rate": 2.84110540896642e-05,
+      "loss": 0.3829,
+      "step": 526
+    },
+    {
+      "epoch": 2.133603238866397,
+      "grad_norm": 0.4776866557299095,
+      "learning_rate": 2.8359710596862555e-05,
+      "loss": 0.4061,
+      "step": 527
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.4457720430435331,
+      "learning_rate": 2.8308300260037734e-05,
+      "loss": 0.3863,
+      "step": 528
+    },
+    {
+      "epoch": 2.1417004048582995,
+      "grad_norm": 0.4430936157885554,
+      "learning_rate": 2.8256823490265424e-05,
+      "loss": 0.3913,
+      "step": 529
+    },
+    {
+      "epoch": 2.145748987854251,
+      "grad_norm": 0.38282040368187126,
+      "learning_rate": 2.8205280699152515e-05,
+      "loss": 0.3981,
+      "step": 530
+    },
+    {
+      "epoch": 2.1497975708502026,
+      "grad_norm": 0.4264773913437342,
+      "learning_rate": 2.8153672298833777e-05,
+      "loss": 0.3857,
+      "step": 531
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.3990917615232531,
+      "learning_rate": 2.8101998701968615e-05,
+      "loss": 0.3964,
+      "step": 532
+    },
+    {
+      "epoch": 2.1578947368421053,
+      "grad_norm": 0.3908004019291611,
+      "learning_rate": 2.805026032173773e-05,
+      "loss": 0.3901,
+      "step": 533
+    },
+    {
+      "epoch": 2.161943319838057,
+      "grad_norm": 0.411184747008421,
+      "learning_rate": 2.7998457571839832e-05,
+      "loss": 0.3716,
+      "step": 534
+    },
+    {
+      "epoch": 2.165991902834008,
+      "grad_norm": 0.3864473818807419,
+      "learning_rate": 2.7946590866488322e-05,
+      "loss": 0.3876,
+      "step": 535
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.40297754276968883,
+      "learning_rate": 2.7894660620408e-05,
+      "loss": 0.3928,
+      "step": 536
+    },
+    {
+      "epoch": 2.174089068825911,
+      "grad_norm": 0.3985987166034244,
+      "learning_rate": 2.784266724883173e-05,
+      "loss": 0.3937,
+      "step": 537
+    },
+    {
+      "epoch": 2.1781376518218623,
+      "grad_norm": 0.40043717295314174,
+      "learning_rate": 2.7790611167497128e-05,
+      "loss": 0.3962,
+      "step": 538
+    },
+    {
+      "epoch": 2.182186234817814,
+      "grad_norm": 0.4011624142810057,
+      "learning_rate": 2.7738492792643224e-05,
+      "loss": 0.391,
+      "step": 539
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.36656717527090493,
+      "learning_rate": 2.768631254100716e-05,
+      "loss": 0.3928,
+      "step": 540
+    },
+    {
+      "epoch": 2.1902834008097165,
+      "grad_norm": 0.3858123185841993,
+      "learning_rate": 2.7634070829820836e-05,
+      "loss": 0.3874,
+      "step": 541
+    },
+    {
+      "epoch": 2.194331983805668,
+      "grad_norm": 0.38447255021733184,
+      "learning_rate": 2.7581768076807592e-05,
+      "loss": 0.3946,
+      "step": 542
+    },
+    {
+      "epoch": 2.1983805668016196,
+      "grad_norm": 0.3894383123256074,
+      "learning_rate": 2.7529404700178832e-05,
+      "loss": 0.3889,
+      "step": 543
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.4559633145504188,
+      "learning_rate": 2.7476981118630735e-05,
+      "loss": 0.4076,
+      "step": 544
+    },
+    {
+      "epoch": 2.2064777327935223,
+      "grad_norm": 0.4153104234427408,
+      "learning_rate": 2.742449775134085e-05,
+      "loss": 0.3852,
+      "step": 545
+    },
+    {
+      "epoch": 2.2105263157894735,
+      "grad_norm": 0.4228379989164378,
+      "learning_rate": 2.7371955017964788e-05,
+      "loss": 0.3915,
+      "step": 546
+    },
+    {
+      "epoch": 2.214574898785425,
+      "grad_norm": 0.42577873931100324,
+      "learning_rate": 2.7319353338632845e-05,
+      "loss": 0.3841,
+      "step": 547
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.4324825194610688,
+      "learning_rate": 2.7266693133946645e-05,
+      "loss": 0.371,
+      "step": 548
+    },
+    {
+      "epoch": 2.2226720647773277,
+      "grad_norm": 0.4122511746431106,
+      "learning_rate": 2.7213974824975785e-05,
+      "loss": 0.3966,
+      "step": 549
+    },
+    {
+      "epoch": 2.2267206477732793,
+      "grad_norm": 0.41612446184568896,
+      "learning_rate": 2.7161198833254464e-05,
+      "loss": 0.3983,
+      "step": 550
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.4030622098136118,
+      "learning_rate": 2.7108365580778095e-05,
+      "loss": 0.3855,
+      "step": 551
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.441423633379814,
+      "learning_rate": 2.705547548999996e-05,
+      "loss": 0.3924,
+      "step": 552
+    },
+    {
+      "epoch": 2.2388663967611335,
+      "grad_norm": 0.42084139559701617,
+      "learning_rate": 2.7002528983827817e-05,
+      "loss": 0.3953,
+      "step": 553
+    },
+    {
+      "epoch": 2.242914979757085,
+      "grad_norm": 0.4154566171989948,
+      "learning_rate": 2.6949526485620517e-05,
+      "loss": 0.4035,
+      "step": 554
+    },
+    {
+      "epoch": 2.246963562753036,
+      "grad_norm": 0.3870274930353934,
+      "learning_rate": 2.6896468419184623e-05,
+      "loss": 0.3765,
+      "step": 555
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.4034844621310081,
+      "learning_rate": 2.6843355208771024e-05,
+      "loss": 0.3874,
+      "step": 556
+    },
+    {
+      "epoch": 2.2550607287449393,
+      "grad_norm": 0.39564224528251585,
+      "learning_rate": 2.6790187279071536e-05,
+      "loss": 0.3612,
+      "step": 557
+    },
+    {
+      "epoch": 2.2591093117408905,
+      "grad_norm": 0.3924266201462239,
+      "learning_rate": 2.67369650552155e-05,
+      "loss": 0.375,
+      "step": 558
+    },
+    {
+      "epoch": 2.263157894736842,
+      "grad_norm": 0.42581481120565046,
+      "learning_rate": 2.6683688962766407e-05,
+      "loss": 0.3777,
+      "step": 559
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.39220231273223105,
+      "learning_rate": 2.6630359427718468e-05,
+      "loss": 0.3886,
+      "step": 560
+    },
+    {
+      "epoch": 2.2712550607287447,
+      "grad_norm": 0.4189442598946167,
+      "learning_rate": 2.6576976876493225e-05,
+      "loss": 0.384,
+      "step": 561
+    },
+    {
+      "epoch": 2.2753036437246963,
+      "grad_norm": 0.4040825195886819,
+      "learning_rate": 2.652354173593614e-05,
+      "loss": 0.3776,
+      "step": 562
+    },
+    {
+      "epoch": 2.279352226720648,
+      "grad_norm": 0.43631882992379756,
+      "learning_rate": 2.647005443331317e-05,
+      "loss": 0.3854,
+      "step": 563
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.4299295308572062,
+      "learning_rate": 2.6416515396307354e-05,
+      "loss": 0.3693,
+      "step": 564
+    },
+    {
+      "epoch": 2.2874493927125505,
+      "grad_norm": 0.4220958463160412,
+      "learning_rate": 2.636292505301542e-05,
+      "loss": 0.4022,
+      "step": 565
+    },
+    {
+      "epoch": 2.291497975708502,
+      "grad_norm": 0.42196222115099663,
+      "learning_rate": 2.630928383194431e-05,
+      "loss": 0.3808,
+      "step": 566
+    },
+    {
+      "epoch": 2.2955465587044532,
+      "grad_norm": 0.44427272089695663,
+      "learning_rate": 2.625559216200781e-05,
+      "loss": 0.4009,
+      "step": 567
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.38646255622496334,
+      "learning_rate": 2.620185047252308e-05,
+      "loss": 0.4007,
+      "step": 568
+    },
+    {
+      "epoch": 2.3036437246963564,
+      "grad_norm": 0.4396551178081656,
+      "learning_rate": 2.614805919320723e-05,
+      "loss": 0.3813,
+      "step": 569
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 0.40279827421513503,
+      "learning_rate": 2.6094218754173895e-05,
+      "loss": 0.3923,
+      "step": 570
+    },
+    {
+      "epoch": 2.311740890688259,
+      "grad_norm": 0.4316348213786486,
+      "learning_rate": 2.60403295859298e-05,
+      "loss": 0.3841,
+      "step": 571
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.4217187802796645,
+      "learning_rate": 2.5986392119371297e-05,
+      "loss": 0.3727,
+      "step": 572
+    },
+    {
+      "epoch": 2.3198380566801617,
+      "grad_norm": 0.4068459280648724,
+      "learning_rate": 2.5932406785780928e-05,
+      "loss": 0.3855,
+      "step": 573
+    },
+    {
+      "epoch": 2.3238866396761133,
+      "grad_norm": 0.4450915679809029,
+      "learning_rate": 2.5878374016823987e-05,
+      "loss": 0.368,
+      "step": 574
+    },
+    {
+      "epoch": 2.327935222672065,
+      "grad_norm": 0.41386380159725566,
+      "learning_rate": 2.5824294244545045e-05,
+      "loss": 0.3985,
+      "step": 575
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.39734559557919225,
+      "learning_rate": 2.577016790136453e-05,
+      "loss": 0.3901,
+      "step": 576
+    },
+    {
+      "epoch": 2.3360323886639676,
+      "grad_norm": 0.47015018817525583,
+      "learning_rate": 2.5715995420075225e-05,
+      "loss": 0.3823,
+      "step": 577
+    },
+    {
+      "epoch": 2.340080971659919,
+      "grad_norm": 0.3873914085166831,
+      "learning_rate": 2.5661777233838864e-05,
+      "loss": 0.3687,
+      "step": 578
+    },
+    {
+      "epoch": 2.3441295546558703,
+      "grad_norm": 0.47462414431278704,
+      "learning_rate": 2.5607513776182616e-05,
+      "loss": 0.4068,
+      "step": 579
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.398753441077253,
+      "learning_rate": 2.5553205480995633e-05,
+      "loss": 0.3762,
+      "step": 580
+    },
+    {
+      "epoch": 2.3522267206477734,
+      "grad_norm": 0.3912661352372768,
+      "learning_rate": 2.5498852782525604e-05,
+      "loss": 0.3947,
+      "step": 581
+    },
+    {
+      "epoch": 2.3562753036437245,
+      "grad_norm": 0.3986415274534305,
+      "learning_rate": 2.5444456115375252e-05,
+      "loss": 0.3836,
+      "step": 582
+    },
+    {
+      "epoch": 2.360323886639676,
+      "grad_norm": 0.39661910655480925,
+      "learning_rate": 2.539001591449889e-05,
+      "loss": 0.3944,
+      "step": 583
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.3677197927485816,
+      "learning_rate": 2.53355326151989e-05,
+      "loss": 0.387,
+      "step": 584
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 0.4110688587260027,
+      "learning_rate": 2.5281006653122288e-05,
+      "loss": 0.3701,
+      "step": 585
+    },
+    {
+      "epoch": 2.3724696356275303,
+      "grad_norm": 0.39462990392851505,
+      "learning_rate": 2.5226438464257217e-05,
+      "loss": 0.386,
+      "step": 586
+    },
+    {
+      "epoch": 2.376518218623482,
+      "grad_norm": 0.3907608566864554,
+      "learning_rate": 2.5171828484929447e-05,
+      "loss": 0.3972,
+      "step": 587
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.38857831111020275,
+      "learning_rate": 2.5117177151798934e-05,
+      "loss": 0.3704,
+      "step": 588
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.377184334881269,
+      "learning_rate": 2.506248490185627e-05,
+      "loss": 0.3878,
+      "step": 589
+    },
+    {
+      "epoch": 2.388663967611336,
+      "grad_norm": 0.4269487006366709,
+      "learning_rate": 2.500775217241923e-05,
+      "loss": 0.3906,
+      "step": 590
+    },
+    {
+      "epoch": 2.3927125506072873,
+      "grad_norm": 0.4004447637165433,
+      "learning_rate": 2.4952979401129266e-05,
+      "loss": 0.3803,
+      "step": 591
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.39609594960675804,
+      "learning_rate": 2.4898167025948e-05,
+      "loss": 0.3955,
+      "step": 592
+    },
+    {
+      "epoch": 2.4008097165991904,
+      "grad_norm": 0.41261759507255324,
+      "learning_rate": 2.4843315485153703e-05,
+      "loss": 0.3963,
+      "step": 593
+    },
+    {
+      "epoch": 2.4048582995951415,
+      "grad_norm": 0.3939522170209968,
+      "learning_rate": 2.4788425217337856e-05,
+      "loss": 0.3878,
+      "step": 594
+    },
+    {
+      "epoch": 2.408906882591093,
+      "grad_norm": 0.39694068689299616,
+      "learning_rate": 2.4733496661401572e-05,
+      "loss": 0.3843,
+      "step": 595
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.4142503180284743,
+      "learning_rate": 2.4678530256552114e-05,
+      "loss": 0.3798,
+      "step": 596
+    },
+    {
+      "epoch": 2.417004048582996,
+      "grad_norm": 0.39960298715521886,
+      "learning_rate": 2.46235264422994e-05,
+      "loss": 0.3732,
+      "step": 597
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 0.40066950938792256,
+      "learning_rate": 2.4568485658452452e-05,
+      "loss": 0.3861,
+      "step": 598
+    },
+    {
+      "epoch": 2.425101214574899,
+      "grad_norm": 0.41578347190609943,
+      "learning_rate": 2.451340834511592e-05,
+      "loss": 0.3792,
+      "step": 599
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.39620135786065835,
+      "learning_rate": 2.4458294942686528e-05,
+      "loss": 0.3832,
+      "step": 600
+    },
+    {
+      "epoch": 2.4331983805668016,
+      "grad_norm": 0.40221665910262056,
+      "learning_rate": 2.440314589184958e-05,
+      "loss": 0.3902,
+      "step": 601
+    },
+    {
+      "epoch": 2.437246963562753,
+      "grad_norm": 0.40843493301832,
+      "learning_rate": 2.434796163357541e-05,
+      "loss": 0.3895,
+      "step": 602
+    },
+    {
+      "epoch": 2.4412955465587043,
+      "grad_norm": 0.39000069152697875,
+      "learning_rate": 2.429274260911588e-05,
+      "loss": 0.4024,
+      "step": 603
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.3864688662258653,
+      "learning_rate": 2.423748926000084e-05,
+      "loss": 0.3856,
+      "step": 604
+    },
+    {
+      "epoch": 2.4493927125506074,
+      "grad_norm": 0.37069919526011824,
+      "learning_rate": 2.418220202803459e-05,
+      "loss": 0.3866,
+      "step": 605
+    },
+    {
+      "epoch": 2.4534412955465585,
+      "grad_norm": 0.3877255376161335,
+      "learning_rate": 2.412688135529237e-05,
+      "loss": 0.3756,
+      "step": 606
+    },
+    {
+      "epoch": 2.45748987854251,
+      "grad_norm": 0.42323551589132813,
+      "learning_rate": 2.4071527684116807e-05,
+      "loss": 0.4083,
+      "step": 607
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.38362097972911824,
+      "learning_rate": 2.4016141457114372e-05,
+      "loss": 0.3868,
+      "step": 608
+    },
+    {
+      "epoch": 2.465587044534413,
+      "grad_norm": 0.40969088483725746,
+      "learning_rate": 2.396072311715187e-05,
+      "loss": 0.4026,
+      "step": 609
+    },
+    {
+      "epoch": 2.4696356275303644,
+      "grad_norm": 0.42794574246872574,
+      "learning_rate": 2.3905273107352863e-05,
+      "loss": 0.3982,
+      "step": 610
+    },
+    {
+      "epoch": 2.473684210526316,
+      "grad_norm": 2.8430213480025643,
+      "learning_rate": 2.3849791871094157e-05,
+      "loss": 0.3993,
+      "step": 611
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.49675538143297565,
+      "learning_rate": 2.379427985200223e-05,
+      "loss": 0.3735,
+      "step": 612
+    },
+    {
+      "epoch": 2.4817813765182186,
+      "grad_norm": 0.4199658584753726,
+      "learning_rate": 2.373873749394972e-05,
+      "loss": 0.3715,
+      "step": 613
+    },
+    {
+      "epoch": 2.48582995951417,
+      "grad_norm": 0.53484468931254,
+      "learning_rate": 2.3683165241051836e-05,
+      "loss": 0.3733,
+      "step": 614
+    },
+    {
+      "epoch": 2.4898785425101213,
+      "grad_norm": 0.38825592189612285,
+      "learning_rate": 2.3627563537662823e-05,
+      "loss": 0.3807,
+      "step": 615
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.4408659535411732,
+      "learning_rate": 2.3571932828372433e-05,
+      "loss": 0.4041,
+      "step": 616
+    },
+    {
+      "epoch": 2.4979757085020244,
+      "grad_norm": 0.40056340378080796,
+      "learning_rate": 2.351627355800233e-05,
+      "loss": 0.3818,
+      "step": 617
+    },
+    {
+      "epoch": 2.5020242914979756,
+      "grad_norm": 0.44170910679778147,
+      "learning_rate": 2.346058617160256e-05,
+      "loss": 0.404,
+      "step": 618
+    },
+    {
+      "epoch": 2.506072874493927,
+      "grad_norm": 0.4013076809633818,
+      "learning_rate": 2.3404871114447976e-05,
+      "loss": 0.4077,
+      "step": 619
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.46660601971891946,
+      "learning_rate": 2.334912883203469e-05,
+      "loss": 0.4041,
+      "step": 620
+    },
+    {
+      "epoch": 2.51417004048583,
+      "grad_norm": 0.37113617716761127,
+      "learning_rate": 2.3293359770076505e-05,
+      "loss": 0.4153,
+      "step": 621
+    },
+    {
+      "epoch": 2.5182186234817814,
+      "grad_norm": 0.45771588495307886,
+      "learning_rate": 2.323756437450136e-05,
+      "loss": 0.3909,
+      "step": 622
+    },
+    {
+      "epoch": 2.522267206477733,
+      "grad_norm": 0.3831691547613533,
+      "learning_rate": 2.3181743091447754e-05,
+      "loss": 0.3838,
+      "step": 623
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.3929819331208969,
+      "learning_rate": 2.3125896367261164e-05,
+      "loss": 0.3929,
+      "step": 624
+    },
+    {
+      "epoch": 2.5303643724696356,
+      "grad_norm": 0.4033681113679402,
+      "learning_rate": 2.307002464849052e-05,
+      "loss": 0.378,
+      "step": 625
+    },
+    {
+      "epoch": 2.534412955465587,
+      "grad_norm": 0.3820285916256086,
+      "learning_rate": 2.30141283818846e-05,
+      "loss": 0.3852,
+      "step": 626
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.36763052885958636,
+      "learning_rate": 2.2958208014388462e-05,
+      "loss": 0.3745,
+      "step": 627
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.3987974885975454,
+      "learning_rate": 2.2902263993139857e-05,
+      "loss": 0.3984,
+      "step": 628
+    },
+    {
+      "epoch": 2.5465587044534415,
+      "grad_norm": 0.35911216446947897,
+      "learning_rate": 2.2846296765465708e-05,
+      "loss": 0.4019,
+      "step": 629
+    },
+    {
+      "epoch": 2.5506072874493926,
+      "grad_norm": 0.3960206710694736,
+      "learning_rate": 2.2790306778878465e-05,
+      "loss": 0.4041,
+      "step": 630
+    },
+    {
+      "epoch": 2.554655870445344,
+      "grad_norm": 0.39355373595089643,
+      "learning_rate": 2.273429448107257e-05,
+      "loss": 0.3706,
+      "step": 631
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.39489660865911325,
+      "learning_rate": 2.2678260319920868e-05,
+      "loss": 0.4005,
+      "step": 632
+    },
+    {
+      "epoch": 2.562753036437247,
+      "grad_norm": 0.38463897355438026,
+      "learning_rate": 2.262220474347102e-05,
+      "loss": 0.3898,
+      "step": 633
+    },
+    {
+      "epoch": 2.5668016194331984,
+      "grad_norm": 0.3973188476082157,
+      "learning_rate": 2.2566128199941918e-05,
+      "loss": 0.3882,
+      "step": 634
+    },
+    {
+      "epoch": 2.57085020242915,
+      "grad_norm": 0.3801625998536347,
+      "learning_rate": 2.2510031137720104e-05,
+      "loss": 0.3639,
+      "step": 635
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.3939562550071542,
+      "learning_rate": 2.2453914005356206e-05,
+      "loss": 0.3895,
+      "step": 636
+    },
+    {
+      "epoch": 2.5789473684210527,
+      "grad_norm": 0.41654626151463464,
+      "learning_rate": 2.2397777251561302e-05,
+      "loss": 0.3739,
+      "step": 637
+    },
+    {
+      "epoch": 2.582995951417004,
+      "grad_norm": 0.377418889328505,
+      "learning_rate": 2.2341621325203393e-05,
+      "loss": 0.3809,
+      "step": 638
+    },
+    {
+      "epoch": 2.5870445344129553,
+      "grad_norm": 0.40622005685749235,
+      "learning_rate": 2.2285446675303764e-05,
+      "loss": 0.3745,
+      "step": 639
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.36507671731400604,
+      "learning_rate": 2.2229253751033423e-05,
+      "loss": 0.3845,
+      "step": 640
+    },
+    {
+      "epoch": 2.5951417004048585,
+      "grad_norm": 0.3695219546241292,
+      "learning_rate": 2.2173043001709492e-05,
+      "loss": 0.4055,
+      "step": 641
+    },
+    {
+      "epoch": 2.5991902834008096,
+      "grad_norm": 0.38161659044792534,
+      "learning_rate": 2.2116814876791627e-05,
+      "loss": 0.3875,
+      "step": 642
+    },
+    {
+      "epoch": 2.603238866396761,
+      "grad_norm": 0.3927388798732099,
+      "learning_rate": 2.2060569825878413e-05,
+      "loss": 0.4064,
+      "step": 643
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.38035701763401253,
+      "learning_rate": 2.2004308298703777e-05,
+      "loss": 0.3917,
+      "step": 644
+    },
+    {
+      "epoch": 2.611336032388664,
+      "grad_norm": 0.3928441326321771,
+      "learning_rate": 2.1948030745133396e-05,
+      "loss": 0.3725,
+      "step": 645
+    },
+    {
+      "epoch": 2.6153846153846154,
+      "grad_norm": 0.37445545523895857,
+      "learning_rate": 2.1891737615161083e-05,
+      "loss": 0.3892,
+      "step": 646
+    },
+    {
+      "epoch": 2.619433198380567,
+      "grad_norm": 0.4087987060931621,
+      "learning_rate": 2.1835429358905218e-05,
+      "loss": 0.3875,
+      "step": 647
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.38534710303113656,
+      "learning_rate": 2.17791064266051e-05,
+      "loss": 0.3792,
+      "step": 648
+    },
+    {
+      "epoch": 2.6275303643724697,
+      "grad_norm": 0.41065504076835585,
+      "learning_rate": 2.172276926861741e-05,
+      "loss": 0.38,
+      "step": 649
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 0.41445574626795656,
+      "learning_rate": 2.166641833541255e-05,
+      "loss": 0.3796,
+      "step": 650
+    },
+    {
+      "epoch": 2.6356275303643724,
+      "grad_norm": 0.4300650825947367,
+      "learning_rate": 2.161005407757109e-05,
+      "loss": 0.3578,
+      "step": 651
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.3962736696063698,
+      "learning_rate": 2.155367694578013e-05,
+      "loss": 0.3875,
+      "step": 652
+    },
+    {
+      "epoch": 2.6437246963562755,
+      "grad_norm": 0.5097386624396757,
+      "learning_rate": 2.1497287390829712e-05,
+      "loss": 0.3899,
+      "step": 653
+    },
+    {
+      "epoch": 2.6477732793522266,
+      "grad_norm": 0.3942639402127727,
+      "learning_rate": 2.1440885863609208e-05,
+      "loss": 0.3905,
+      "step": 654
+    },
+    {
+      "epoch": 2.651821862348178,
+      "grad_norm": 0.4379455584710177,
+      "learning_rate": 2.1384472815103737e-05,
+      "loss": 0.4091,
+      "step": 655
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.42057549607275163,
+      "learning_rate": 2.1328048696390534e-05,
+      "loss": 0.3895,
+      "step": 656
+    },
+    {
+      "epoch": 2.659919028340081,
+      "grad_norm": 0.42072297115752083,
+      "learning_rate": 2.1271613958635342e-05,
+      "loss": 0.3881,
+      "step": 657
+    },
+    {
+      "epoch": 2.6639676113360324,
+      "grad_norm": 0.40902246399731224,
+      "learning_rate": 2.1215169053088825e-05,
+      "loss": 0.4031,
+      "step": 658
+    },
+    {
+      "epoch": 2.668016194331984,
+      "grad_norm": 0.381305108757983,
+      "learning_rate": 2.1158714431082937e-05,
+      "loss": 0.4053,
+      "step": 659
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.43180442609538167,
+      "learning_rate": 2.110225054402733e-05,
+      "loss": 0.4282,
+      "step": 660
+    },
+    {
+      "epoch": 2.6761133603238867,
+      "grad_norm": 0.3630284274944407,
+      "learning_rate": 2.1045777843405747e-05,
+      "loss": 0.3885,
+      "step": 661
+    },
+    {
+      "epoch": 2.6801619433198383,
+      "grad_norm": 0.41678472918632753,
+      "learning_rate": 2.09892967807724e-05,
+      "loss": 0.3939,
+      "step": 662
+    },
+    {
+      "epoch": 2.6842105263157894,
+      "grad_norm": 0.3896081199746742,
+      "learning_rate": 2.093280780774835e-05,
+      "loss": 0.3884,
+      "step": 663
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.3718078966363521,
+      "learning_rate": 2.087631137601793e-05,
+      "loss": 0.3779,
+      "step": 664
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.3618780562308493,
+      "learning_rate": 2.0819807937325104e-05,
+      "loss": 0.3908,
+      "step": 665
+    },
+    {
+      "epoch": 2.6963562753036436,
+      "grad_norm": 0.3986182318873002,
+      "learning_rate": 2.0763297943469847e-05,
+      "loss": 0.4084,
+      "step": 666
+    },
+    {
+      "epoch": 2.700404858299595,
+      "grad_norm": 0.34589448209675466,
+      "learning_rate": 2.070678184630458e-05,
+      "loss": 0.3857,
+      "step": 667
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.40536979160261905,
+      "learning_rate": 2.06502600977305e-05,
+      "loss": 0.3738,
+      "step": 668
+    },
+    {
+      "epoch": 2.708502024291498,
+      "grad_norm": 0.3955512100052998,
+      "learning_rate": 2.0593733149694008e-05,
+      "loss": 0.3951,
+      "step": 669
+    },
+    {
+      "epoch": 2.7125506072874495,
+      "grad_norm": 0.4150257303078211,
+      "learning_rate": 2.0537201454183074e-05,
+      "loss": 0.3676,
+      "step": 670
+    },
+    {
+      "epoch": 2.716599190283401,
+      "grad_norm": 0.40463247793173884,
+      "learning_rate": 2.0480665463223626e-05,
+      "loss": 0.3877,
+      "step": 671
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.35693462005563764,
+      "learning_rate": 2.042412562887594e-05,
+      "loss": 0.419,
+      "step": 672
+    },
+    {
+      "epoch": 2.7246963562753037,
+      "grad_norm": 0.3636735973857597,
+      "learning_rate": 2.0367582403231033e-05,
+      "loss": 0.3867,
+      "step": 673
+    },
+    {
+      "epoch": 2.7287449392712553,
+      "grad_norm": 0.40080902159380183,
+      "learning_rate": 2.0311036238407023e-05,
+      "loss": 0.3831,
+      "step": 674
+    },
+    {
+      "epoch": 2.7327935222672064,
+      "grad_norm": 0.3598982920260706,
+      "learning_rate": 2.0254487586545534e-05,
+      "loss": 0.3827,
+      "step": 675
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.43178132162690397,
+      "learning_rate": 2.0197936899808088e-05,
+      "loss": 0.3875,
+      "step": 676
+    },
+    {
+      "epoch": 2.7408906882591095,
+      "grad_norm": 0.4010746204053763,
+      "learning_rate": 2.0141384630372466e-05,
+      "loss": 0.3692,
+      "step": 677
+    },
+    {
+      "epoch": 2.7449392712550607,
+      "grad_norm": 0.3943412904355165,
+      "learning_rate": 2.0084831230429106e-05,
+      "loss": 0.3674,
+      "step": 678
+    },
+    {
+      "epoch": 2.748987854251012,
+      "grad_norm": 0.4084358333145342,
+      "learning_rate": 2.0028277152177492e-05,
+      "loss": 0.3785,
+      "step": 679
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.3401330385276392,
+      "learning_rate": 1.9971722847822518e-05,
+      "loss": 0.3912,
+      "step": 680
+    },
+    {
+      "epoch": 2.757085020242915,
+      "grad_norm": 0.3989856937811702,
+      "learning_rate": 1.99151687695709e-05,
+      "loss": 0.4015,
+      "step": 681
+    },
+    {
+      "epoch": 2.7611336032388665,
+      "grad_norm": 0.37892368534465215,
+      "learning_rate": 1.9858615369627537e-05,
+      "loss": 0.4007,
+      "step": 682
+    },
+    {
+      "epoch": 2.765182186234818,
+      "grad_norm": 0.41264530128264176,
+      "learning_rate": 1.9802063100191916e-05,
+      "loss": 0.3849,
+      "step": 683
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.34884002982923373,
+      "learning_rate": 1.974551241345447e-05,
+      "loss": 0.3871,
+      "step": 684
+    },
+    {
+      "epoch": 2.7732793522267207,
+      "grad_norm": 0.3777151392468431,
+      "learning_rate": 1.9688963761592984e-05,
+      "loss": 0.3671,
+      "step": 685
+    },
+    {
+      "epoch": 2.7773279352226723,
+      "grad_norm": 0.35532851853291564,
+      "learning_rate": 1.963241759676897e-05,
+      "loss": 0.3645,
+      "step": 686
+    },
+    {
+      "epoch": 2.7813765182186234,
+      "grad_norm": 0.3652828793730711,
+      "learning_rate": 1.9575874371124066e-05,
+      "loss": 0.3778,
+      "step": 687
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.3584460099460744,
+      "learning_rate": 1.951933453677638e-05,
+      "loss": 0.3671,
+      "step": 688
+    },
+    {
+      "epoch": 2.7894736842105265,
+      "grad_norm": 0.39063617672127504,
+      "learning_rate": 1.9462798545816932e-05,
+      "loss": 0.3969,
+      "step": 689
+    },
+    {
+      "epoch": 2.7935222672064777,
+      "grad_norm": 0.37835570650231287,
+      "learning_rate": 1.9406266850305996e-05,
+      "loss": 0.396,
+      "step": 690
+    },
+    {
+      "epoch": 2.7975708502024292,
+      "grad_norm": 0.3553108864118431,
+      "learning_rate": 1.9349739902269503e-05,
+      "loss": 0.3815,
+      "step": 691
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.37743249570688964,
+      "learning_rate": 1.9293218153695425e-05,
+      "loss": 0.3676,
+      "step": 692
+    },
+    {
+      "epoch": 2.805668016194332,
+      "grad_norm": 0.39671378057540535,
+      "learning_rate": 1.9236702056530153e-05,
+      "loss": 0.4036,
+      "step": 693
+    },
+    {
+      "epoch": 2.8097165991902835,
+      "grad_norm": 0.3450835804771746,
+      "learning_rate": 1.9180192062674907e-05,
+      "loss": 0.3651,
+      "step": 694
+    },
+    {
+      "epoch": 2.813765182186235,
+      "grad_norm": 0.4066273173082974,
+      "learning_rate": 1.9123688623982076e-05,
+      "loss": 0.3786,
+      "step": 695
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.36568982632642244,
+      "learning_rate": 1.906719219225165e-05,
+      "loss": 0.3941,
+      "step": 696
+    },
+    {
+      "epoch": 2.8218623481781377,
+      "grad_norm": 0.3584511397262581,
+      "learning_rate": 1.9010703219227603e-05,
+      "loss": 0.3603,
+      "step": 697
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 0.3769805886414872,
+      "learning_rate": 1.8954222156594256e-05,
+      "loss": 0.3685,
+      "step": 698
+    },
+    {
+      "epoch": 2.8299595141700404,
+      "grad_norm": 0.3679415153035342,
+      "learning_rate": 1.8897749455972673e-05,
+      "loss": 0.3992,
+      "step": 699
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.3595420434910692,
+      "learning_rate": 1.884128556891707e-05,
+      "loss": 0.3811,
+      "step": 700
+    },
+    {
+      "epoch": 2.8380566801619436,
+      "grad_norm": 0.3545718789910015,
+      "learning_rate": 1.878483094691119e-05,
+      "loss": 0.3787,
+      "step": 701
+    },
+    {
+      "epoch": 2.8421052631578947,
+      "grad_norm": 0.3762351784266276,
+      "learning_rate": 1.8728386041364664e-05,
+      "loss": 0.3994,
+      "step": 702
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.37296417175562857,
+      "learning_rate": 1.867195130360947e-05,
+      "loss": 0.3936,
+      "step": 703
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.3595377667168117,
+      "learning_rate": 1.8615527184896263e-05,
+      "loss": 0.3843,
+      "step": 704
+    },
+    {
+      "epoch": 2.854251012145749,
+      "grad_norm": 0.40794521198372463,
+      "learning_rate": 1.8559114136390795e-05,
+      "loss": 0.3737,
+      "step": 705
+    },
+    {
+      "epoch": 2.8582995951417005,
+      "grad_norm": 0.3721139596494975,
+      "learning_rate": 1.8502712609170298e-05,
+      "loss": 0.3781,
+      "step": 706
+    },
+    {
+      "epoch": 2.862348178137652,
+      "grad_norm": 0.388589761262634,
+      "learning_rate": 1.8446323054219876e-05,
+      "loss": 0.3823,
+      "step": 707
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.383406493107686,
+      "learning_rate": 1.838994592242891e-05,
+      "loss": 0.3706,
+      "step": 708
+    },
+    {
+      "epoch": 2.8704453441295548,
+      "grad_norm": 0.3847005333717645,
+      "learning_rate": 1.8333581664587453e-05,
+      "loss": 0.3608,
+      "step": 709
+    },
+    {
+      "epoch": 2.8744939271255063,
+      "grad_norm": 0.3994934893542008,
+      "learning_rate": 1.8277230731382593e-05,
+      "loss": 0.3735,
+      "step": 710
+    },
+    {
+      "epoch": 2.8785425101214575,
+      "grad_norm": 0.3649943083653967,
+      "learning_rate": 1.82208935733949e-05,
+      "loss": 0.3694,
+      "step": 711
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.37310848366539395,
+      "learning_rate": 1.8164570641094793e-05,
+      "loss": 0.3856,
+      "step": 712
+    },
+    {
+      "epoch": 2.8866396761133606,
+      "grad_norm": 0.3624487243495906,
+      "learning_rate": 1.810826238483892e-05,
+      "loss": 0.3991,
+      "step": 713
+    },
+    {
+      "epoch": 2.8906882591093117,
+      "grad_norm": 0.3744908681753284,
+      "learning_rate": 1.8051969254866608e-05,
+      "loss": 0.3825,
+      "step": 714
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 0.37123306617085766,
+      "learning_rate": 1.7995691701296226e-05,
+      "loss": 0.3871,
+      "step": 715
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.35233313435852065,
+      "learning_rate": 1.7939430174121593e-05,
+      "loss": 0.3766,
+      "step": 716
+    },
+    {
+      "epoch": 2.902834008097166,
+      "grad_norm": 0.35769988730924324,
+      "learning_rate": 1.788318512320838e-05,
+      "loss": 0.385,
+      "step": 717
+    },
+    {
+      "epoch": 2.9068825910931175,
+      "grad_norm": 0.35259056652949394,
+      "learning_rate": 1.782695699829051e-05,
+      "loss": 0.374,
+      "step": 718
+    },
+    {
+      "epoch": 2.910931174089069,
+      "grad_norm": 0.38613911317282473,
+      "learning_rate": 1.7770746248966587e-05,
+      "loss": 0.3856,
+      "step": 719
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.33904275163383446,
+      "learning_rate": 1.7714553324696243e-05,
+      "loss": 0.3804,
+      "step": 720
+    },
+    {
+      "epoch": 2.919028340080972,
+      "grad_norm": 0.35140662631677666,
+      "learning_rate": 1.7658378674796614e-05,
+      "loss": 0.3805,
+      "step": 721
+    },
+    {
+      "epoch": 2.9230769230769234,
+      "grad_norm": 0.37852004710808485,
+      "learning_rate": 1.76022227484387e-05,
+      "loss": 0.416,
+      "step": 722
+    },
+    {
+      "epoch": 2.9271255060728745,
+      "grad_norm": 0.3562162489886909,
+      "learning_rate": 1.7546085994643807e-05,
+      "loss": 0.4033,
+      "step": 723
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.4043226391006801,
+      "learning_rate": 1.7489968862279902e-05,
+      "loss": 0.3733,
+      "step": 724
+    },
+    {
+      "epoch": 2.9352226720647776,
+      "grad_norm": 0.37989503911638617,
+      "learning_rate": 1.743387180005809e-05,
+      "loss": 0.3859,
+      "step": 725
+    },
+    {
+      "epoch": 2.9392712550607287,
+      "grad_norm": 0.4378027108044837,
+      "learning_rate": 1.737779525652899e-05,
+      "loss": 0.3771,
+      "step": 726
+    },
+    {
+      "epoch": 2.9433198380566803,
+      "grad_norm": 0.3730245246152584,
+      "learning_rate": 1.732173968007914e-05,
+      "loss": 0.393,
+      "step": 727
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.3860687715548549,
+      "learning_rate": 1.7265705518927437e-05,
+      "loss": 0.3872,
+      "step": 728
+    },
+    {
+      "epoch": 2.951417004048583,
+      "grad_norm": 0.39798375421649074,
+      "learning_rate": 1.7209693221121542e-05,
+      "loss": 0.3837,
+      "step": 729
+    },
+    {
+      "epoch": 2.9554655870445345,
+      "grad_norm": 0.3627249595110388,
+      "learning_rate": 1.7153703234534302e-05,
+      "loss": 0.383,
+      "step": 730
+    },
+    {
+      "epoch": 2.9595141700404857,
+      "grad_norm": 0.3751585865178512,
+      "learning_rate": 1.709773600686015e-05,
+      "loss": 0.3941,
+      "step": 731
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.33099541050957526,
+      "learning_rate": 1.7041791985611548e-05,
+      "loss": 0.3924,
+      "step": 732
+    },
+    {
+      "epoch": 2.967611336032389,
+      "grad_norm": 0.345544190484732,
+      "learning_rate": 1.69858716181154e-05,
+      "loss": 0.4097,
+      "step": 733
+    },
+    {
+      "epoch": 2.97165991902834,
+      "grad_norm": 0.38144750185176024,
+      "learning_rate": 1.692997535150948e-05,
+      "loss": 0.389,
+      "step": 734
+    },
+    {
+      "epoch": 2.9757085020242915,
+      "grad_norm": 0.37093786430240655,
+      "learning_rate": 1.687410363273884e-05,
+      "loss": 0.3944,
+      "step": 735
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.349954933942722,
+      "learning_rate": 1.6818256908552257e-05,
+      "loss": 0.3787,
+      "step": 736
+    },
+    {
+      "epoch": 2.983805668016194,
+      "grad_norm": 0.3683442582561028,
+      "learning_rate": 1.6762435625498646e-05,
+      "loss": 0.3974,
+      "step": 737
+    },
+    {
+      "epoch": 2.9878542510121457,
+      "grad_norm": 0.34593130179121373,
+      "learning_rate": 1.67066402299235e-05,
+      "loss": 0.3686,
+      "step": 738
+    },
+    {
+      "epoch": 2.9919028340080973,
+      "grad_norm": 0.3938022021869564,
+      "learning_rate": 1.6650871167965313e-05,
+      "loss": 0.382,
+      "step": 739
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.3750913065637187,
+      "learning_rate": 1.6595128885552028e-05,
+      "loss": 0.3775,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.34512902092407444,
+      "learning_rate": 1.653941382839745e-05,
+      "loss": 0.3775,
+      "step": 741
+    },
+    {
+      "epoch": 3.0040485829959516,
+      "grad_norm": 0.5899668553625245,
+      "learning_rate": 1.6483726441997673e-05,
+      "loss": 0.2789,
+      "step": 742
+    },
+    {
+      "epoch": 3.0080971659919027,
+      "grad_norm": 0.3616594717505852,
+      "learning_rate": 1.642806717162757e-05,
+      "loss": 0.2872,
+      "step": 743
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.6006310166232499,
+      "learning_rate": 1.637243646233718e-05,
+      "loss": 0.2723,
+      "step": 744
+    },
+    {
+      "epoch": 3.016194331983806,
+      "grad_norm": 0.6540991523446752,
+      "learning_rate": 1.6316834758948174e-05,
+      "loss": 0.2721,
+      "step": 745
+    },
+    {
+      "epoch": 3.020242914979757,
+      "grad_norm": 0.4053554610210997,
+      "learning_rate": 1.6261262506050282e-05,
+      "loss": 0.2785,
+      "step": 746
+    },
+    {
+      "epoch": 3.0242914979757085,
+      "grad_norm": 0.555519626943839,
+      "learning_rate": 1.620572014799777e-05,
+      "loss": 0.2861,
+      "step": 747
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.508395309023496,
+      "learning_rate": 1.6150208128905857e-05,
+      "loss": 0.2725,
+      "step": 748
+    },
+    {
+      "epoch": 3.032388663967611,
+      "grad_norm": 0.39690636281837854,
+      "learning_rate": 1.6094726892647147e-05,
+      "loss": 0.2868,
+      "step": 749
+    },
+    {
+      "epoch": 3.0364372469635628,
+      "grad_norm": 0.5230469760581599,
+      "learning_rate": 1.6039276882848135e-05,
+      "loss": 0.2828,
+      "step": 750
+    },
+    {
+      "epoch": 3.0404858299595143,
+      "grad_norm": 0.4393880885128282,
+      "learning_rate": 1.5983858542885635e-05,
+      "loss": 0.266,
+      "step": 751
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.37896587667762294,
+      "learning_rate": 1.5928472315883203e-05,
+      "loss": 0.2782,
+      "step": 752
+    },
+    {
+      "epoch": 3.048582995951417,
+      "grad_norm": 0.4340055911578125,
+      "learning_rate": 1.5873118644707633e-05,
+      "loss": 0.2811,
+      "step": 753
+    },
+    {
+      "epoch": 3.0526315789473686,
+      "grad_norm": 0.3916475504726916,
+      "learning_rate": 1.5817797971965413e-05,
+      "loss": 0.2689,
+      "step": 754
+    },
+    {
+      "epoch": 3.0566801619433197,
+      "grad_norm": 0.4162910359732616,
+      "learning_rate": 1.576251073999917e-05,
+      "loss": 0.2699,
+      "step": 755
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.4001039700350155,
+      "learning_rate": 1.5707257390884126e-05,
+      "loss": 0.2807,
+      "step": 756
+    },
+    {
+      "epoch": 3.064777327935223,
+      "grad_norm": 0.3813243824780185,
+      "learning_rate": 1.5652038366424595e-05,
+      "loss": 0.2913,
+      "step": 757
+    },
+    {
+      "epoch": 3.068825910931174,
+      "grad_norm": 0.34857132162960763,
+      "learning_rate": 1.5596854108150424e-05,
+      "loss": 0.2722,
+      "step": 758
+    },
+    {
+      "epoch": 3.0728744939271255,
+      "grad_norm": 0.4013320535081263,
+      "learning_rate": 1.5541705057313476e-05,
+      "loss": 0.267,
+      "step": 759
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.3415999773181253,
+      "learning_rate": 1.5486591654884086e-05,
+      "loss": 0.2608,
+      "step": 760
+    },
+    {
+      "epoch": 3.080971659919028,
+      "grad_norm": 0.35439786729954725,
+      "learning_rate": 1.543151434154755e-05,
+      "loss": 0.2828,
+      "step": 761
+    },
+    {
+      "epoch": 3.08502024291498,
+      "grad_norm": 0.35180807661508606,
+      "learning_rate": 1.5376473557700615e-05,
+      "loss": 0.2615,
+      "step": 762
+    },
+    {
+      "epoch": 3.0890688259109313,
+      "grad_norm": 0.35422786015054153,
+      "learning_rate": 1.5321469743447893e-05,
+      "loss": 0.2687,
+      "step": 763
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.3601510942056735,
+      "learning_rate": 1.5266503338598434e-05,
+      "loss": 0.2876,
+      "step": 764
+    },
+    {
+      "epoch": 3.097165991902834,
+      "grad_norm": 0.3271426568598223,
+      "learning_rate": 1.5211574782662149e-05,
+      "loss": 0.2745,
+      "step": 765
+    },
+    {
+      "epoch": 3.1012145748987856,
+      "grad_norm": 0.3351965763647189,
+      "learning_rate": 1.5156684514846307e-05,
+      "loss": 0.269,
+      "step": 766
+    },
+    {
+      "epoch": 3.1052631578947367,
+      "grad_norm": 0.3434521442042505,
+      "learning_rate": 1.5101832974052012e-05,
+      "loss": 0.2868,
+      "step": 767
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.36095112127735346,
+      "learning_rate": 1.5047020598870737e-05,
+      "loss": 0.2805,
+      "step": 768
+    },
+    {
+      "epoch": 3.11336032388664,
+      "grad_norm": 0.31798872348773705,
+      "learning_rate": 1.4992247827580778e-05,
+      "loss": 0.2971,
+      "step": 769
+    },
+    {
+      "epoch": 3.117408906882591,
+      "grad_norm": 0.34850806107550436,
+      "learning_rate": 1.493751509814374e-05,
+      "loss": 0.2861,
+      "step": 770
+    },
+    {
+      "epoch": 3.1214574898785425,
+      "grad_norm": 0.3438040433499023,
+      "learning_rate": 1.4882822848201073e-05,
+      "loss": 0.2619,
+      "step": 771
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.32406893817477744,
+      "learning_rate": 1.4828171515070553e-05,
+      "loss": 0.2482,
+      "step": 772
+    },
+    {
+      "epoch": 3.1295546558704452,
+      "grad_norm": 0.3305567937086873,
+      "learning_rate": 1.4773561535742793e-05,
+      "loss": 0.2787,
+      "step": 773
+    },
+    {
+      "epoch": 3.133603238866397,
+      "grad_norm": 0.33940368518219227,
+      "learning_rate": 1.4718993346877715e-05,
+      "loss": 0.2705,
+      "step": 774
+    },
+    {
+      "epoch": 3.1376518218623484,
+      "grad_norm": 0.3392913771014694,
+      "learning_rate": 1.466446738480111e-05,
+      "loss": 0.2754,
+      "step": 775
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.3242956732982275,
+      "learning_rate": 1.460998408550112e-05,
+      "loss": 0.2763,
+      "step": 776
+    },
+    {
+      "epoch": 3.145748987854251,
+      "grad_norm": 0.33343323202837155,
+      "learning_rate": 1.4555543884624751e-05,
+      "loss": 0.2669,
+      "step": 777
+    },
+    {
+      "epoch": 3.1497975708502026,
+      "grad_norm": 0.3311951016405233,
+      "learning_rate": 1.4501147217474402e-05,
+      "loss": 0.2792,
+      "step": 778
+    },
+    {
+      "epoch": 3.1538461538461537,
+      "grad_norm": 0.31997362993022116,
+      "learning_rate": 1.444679451900437e-05,
+      "loss": 0.2513,
+      "step": 779
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.33884273084289823,
+      "learning_rate": 1.4392486223817397e-05,
+      "loss": 0.2607,
+      "step": 780
+    },
+    {
+      "epoch": 3.161943319838057,
+      "grad_norm": 0.30091449540761495,
+      "learning_rate": 1.4338222766161143e-05,
+      "loss": 0.2648,
+      "step": 781
+    },
+    {
+      "epoch": 3.165991902834008,
+      "grad_norm": 0.3278913144393239,
+      "learning_rate": 1.4284004579924777e-05,
+      "loss": 0.2574,
+      "step": 782
+    },
+    {
+      "epoch": 3.1700404858299596,
+      "grad_norm": 0.346538832543756,
+      "learning_rate": 1.4229832098635479e-05,
+      "loss": 0.2713,
+      "step": 783
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.3257072269661395,
+      "learning_rate": 1.4175705755454963e-05,
+      "loss": 0.2763,
+      "step": 784
+    },
+    {
+      "epoch": 3.1781376518218623,
+      "grad_norm": 0.3226447478583783,
+      "learning_rate": 1.412162598317602e-05,
+      "loss": 0.2714,
+      "step": 785
+    },
+    {
+      "epoch": 3.182186234817814,
+      "grad_norm": 0.35822188422033857,
+      "learning_rate": 1.406759321421907e-05,
+      "loss": 0.2828,
+      "step": 786
+    },
+    {
+      "epoch": 3.1862348178137654,
+      "grad_norm": 0.34536386703641797,
+      "learning_rate": 1.401360788062871e-05,
+      "loss": 0.2829,
+      "step": 787
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.32422201648401006,
+      "learning_rate": 1.3959670414070204e-05,
+      "loss": 0.2909,
+      "step": 788
+    },
+    {
+      "epoch": 3.194331983805668,
+      "grad_norm": 0.3363447422447112,
+      "learning_rate": 1.3905781245826108e-05,
+      "loss": 0.2546,
+      "step": 789
+    },
+    {
+      "epoch": 3.1983805668016196,
+      "grad_norm": 0.34656780540422705,
+      "learning_rate": 1.3851940806792778e-05,
+      "loss": 0.3129,
+      "step": 790
+    },
+    {
+      "epoch": 3.2024291497975708,
+      "grad_norm": 0.33792435293596224,
+      "learning_rate": 1.379814952747693e-05,
+      "loss": 0.297,
+      "step": 791
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.31646912512709785,
+      "learning_rate": 1.3744407837992193e-05,
+      "loss": 0.2631,
+      "step": 792
+    },
+    {
+      "epoch": 3.2105263157894735,
+      "grad_norm": 0.360563420652107,
+      "learning_rate": 1.3690716168055692e-05,
+      "loss": 0.2631,
+      "step": 793
+    },
+    {
+      "epoch": 3.214574898785425,
+      "grad_norm": 0.38367871314152474,
+      "learning_rate": 1.363707494698459e-05,
+      "loss": 0.266,
+      "step": 794
+    },
+    {
+      "epoch": 3.2186234817813766,
+      "grad_norm": 0.3459239883989863,
+      "learning_rate": 1.358348460369265e-05,
+      "loss": 0.2603,
+      "step": 795
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.32813084594109276,
+      "learning_rate": 1.352994556668684e-05,
+      "loss": 0.2614,
+      "step": 796
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 0.3605603112475064,
+      "learning_rate": 1.3476458264063865e-05,
+      "loss": 0.2732,
+      "step": 797
+    },
+    {
+      "epoch": 3.230769230769231,
+      "grad_norm": 0.3600987036792353,
+      "learning_rate": 1.342302312350678e-05,
+      "loss": 0.2625,
+      "step": 798
+    },
+    {
+      "epoch": 3.234817813765182,
+      "grad_norm": 0.3297987942297315,
+      "learning_rate": 1.3369640572281537e-05,
+      "loss": 0.2795,
+      "step": 799
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.3345769394111237,
+      "learning_rate": 1.3316311037233596e-05,
+      "loss": 0.2718,
+      "step": 800
+    },
+    {
+      "epoch": 3.242914979757085,
+      "grad_norm": 0.32495922935016924,
+      "learning_rate": 1.326303494478451e-05,
+      "loss": 0.2797,
+      "step": 801
+    },
+    {
+      "epoch": 3.246963562753036,
+      "grad_norm": 0.3307036667691045,
+      "learning_rate": 1.3209812720928472e-05,
+      "loss": 0.2839,
+      "step": 802
+    },
+    {
+      "epoch": 3.251012145748988,
+      "grad_norm": 0.3351836047218881,
+      "learning_rate": 1.315664479122898e-05,
+      "loss": 0.2809,
+      "step": 803
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.3356812193465489,
+      "learning_rate": 1.3103531580815378e-05,
+      "loss": 0.2848,
+      "step": 804
+    },
+    {
+      "epoch": 3.2591093117408905,
+      "grad_norm": 0.3295248002850665,
+      "learning_rate": 1.3050473514379488e-05,
+      "loss": 0.2864,
+      "step": 805
+    },
+    {
+      "epoch": 3.263157894736842,
+      "grad_norm": 0.3286630488338636,
+      "learning_rate": 1.2997471016172188e-05,
+      "loss": 0.2565,
+      "step": 806
+    },
+    {
+      "epoch": 3.2672064777327936,
+      "grad_norm": 0.33205522172214574,
+      "learning_rate": 1.2944524510000042e-05,
+      "loss": 0.2882,
+      "step": 807
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.3190194931183099,
+      "learning_rate": 1.2891634419221908e-05,
+      "loss": 0.2593,
+      "step": 808
+    },
+    {
+      "epoch": 3.2753036437246963,
+      "grad_norm": 0.3205469850835168,
+      "learning_rate": 1.2838801166745545e-05,
+      "loss": 0.2826,
+      "step": 809
+    },
+    {
+      "epoch": 3.279352226720648,
+      "grad_norm": 0.32240706314769474,
+      "learning_rate": 1.2786025175024216e-05,
+      "loss": 0.2681,
+      "step": 810
+    },
+    {
+      "epoch": 3.283400809716599,
+      "grad_norm": 0.3178658357853268,
+      "learning_rate": 1.2733306866053357e-05,
+      "loss": 0.2819,
+      "step": 811
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.34264276241733166,
+      "learning_rate": 1.2680646661367163e-05,
+      "loss": 0.2669,
+      "step": 812
+    },
+    {
+      "epoch": 3.291497975708502,
+      "grad_norm": 0.32393815138326815,
+      "learning_rate": 1.2628044982035219e-05,
+      "loss": 0.2625,
+      "step": 813
+    },
+    {
+      "epoch": 3.2955465587044532,
+      "grad_norm": 0.3192172444966522,
+      "learning_rate": 1.2575502248659155e-05,
+      "loss": 0.2738,
+      "step": 814
+    },
+    {
+      "epoch": 3.299595141700405,
+      "grad_norm": 0.3081949368471304,
+      "learning_rate": 1.2523018881369269e-05,
+      "loss": 0.2973,
+      "step": 815
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.322420952162966,
+      "learning_rate": 1.2470595299821176e-05,
+      "loss": 0.2683,
+      "step": 816
+    },
+    {
+      "epoch": 3.3076923076923075,
+      "grad_norm": 0.3158887547084371,
+      "learning_rate": 1.2418231923192416e-05,
+      "loss": 0.2726,
+      "step": 817
+    },
+    {
+      "epoch": 3.311740890688259,
+      "grad_norm": 0.3053598764749777,
+      "learning_rate": 1.2365929170179165e-05,
+      "loss": 0.2666,
+      "step": 818
+    },
+    {
+      "epoch": 3.3157894736842106,
+      "grad_norm": 0.3174551497706414,
+      "learning_rate": 1.2313687458992848e-05,
+      "loss": 0.2915,
+      "step": 819
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.290316177400999,
+      "learning_rate": 1.2261507207356785e-05,
+      "loss": 0.2511,
+      "step": 820
+    },
+    {
+      "epoch": 3.3238866396761133,
+      "grad_norm": 0.30941857890741264,
+      "learning_rate": 1.220938883250288e-05,
+      "loss": 0.2489,
+      "step": 821
+    },
+    {
+      "epoch": 3.327935222672065,
+      "grad_norm": 0.3192989890346805,
+      "learning_rate": 1.215733275116827e-05,
+      "loss": 0.2635,
+      "step": 822
+    },
+    {
+      "epoch": 3.331983805668016,
+      "grad_norm": 0.2984599831851027,
+      "learning_rate": 1.2105339379592005e-05,
+      "loss": 0.3028,
+      "step": 823
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.3201781033770257,
+      "learning_rate": 1.2053409133511681e-05,
+      "loss": 0.2845,
+      "step": 824
+    },
+    {
+      "epoch": 3.340080971659919,
+      "grad_norm": 0.2942172824080036,
+      "learning_rate": 1.2001542428160173e-05,
+      "loss": 0.2694,
+      "step": 825
+    },
+    {
+      "epoch": 3.3441295546558703,
+      "grad_norm": 0.31666736239963,
+      "learning_rate": 1.1949739678262278e-05,
+      "loss": 0.2841,
+      "step": 826
+    },
+    {
+      "epoch": 3.348178137651822,
+      "grad_norm": 0.3208960229840278,
+      "learning_rate": 1.189800129803139e-05,
+      "loss": 0.2741,
+      "step": 827
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.31903327585270674,
+      "learning_rate": 1.1846327701166228e-05,
+      "loss": 0.2631,
+      "step": 828
+    },
+    {
+      "epoch": 3.3562753036437245,
+      "grad_norm": 0.307294733367586,
+      "learning_rate": 1.1794719300847493e-05,
+      "loss": 0.2736,
+      "step": 829
+    },
+    {
+      "epoch": 3.360323886639676,
+      "grad_norm": 0.3008378001375845,
+      "learning_rate": 1.1743176509734582e-05,
+      "loss": 0.2822,
+      "step": 830
+    },
+    {
+      "epoch": 3.3643724696356276,
+      "grad_norm": 0.3141971838565364,
+      "learning_rate": 1.1691699739962275e-05,
+      "loss": 0.2661,
+      "step": 831
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.3309203203890598,
+      "learning_rate": 1.1640289403137456e-05,
+      "loss": 0.2693,
+      "step": 832
+    },
+    {
+      "epoch": 3.3724696356275303,
+      "grad_norm": 0.3116669121216112,
+      "learning_rate": 1.1588945910335802e-05,
+      "loss": 0.2726,
+      "step": 833
+    },
+    {
+      "epoch": 3.376518218623482,
+      "grad_norm": 0.3068479794327762,
+      "learning_rate": 1.1537669672098517e-05,
+      "loss": 0.2712,
+      "step": 834
+    },
+    {
+      "epoch": 3.380566801619433,
+      "grad_norm": 0.3466477941080134,
+      "learning_rate": 1.1486461098429036e-05,
+      "loss": 0.2834,
+      "step": 835
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.31572694764544595,
+      "learning_rate": 1.1435320598789747e-05,
+      "loss": 0.257,
+      "step": 836
+    },
+    {
+      "epoch": 3.388663967611336,
+      "grad_norm": 0.30009776500532986,
+      "learning_rate": 1.1384248582098736e-05,
+      "loss": 0.2743,
+      "step": 837
+    },
+    {
+      "epoch": 3.3927125506072873,
+      "grad_norm": 0.3050087889979358,
+      "learning_rate": 1.1333245456726467e-05,
+      "loss": 0.247,
+      "step": 838
+    },
+    {
+      "epoch": 3.396761133603239,
+      "grad_norm": 0.3417794594305218,
+      "learning_rate": 1.128231163049258e-05,
+      "loss": 0.2886,
+      "step": 839
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.3099732545076169,
+      "learning_rate": 1.123144751066259e-05,
+      "loss": 0.2993,
+      "step": 840
+    },
+    {
+      "epoch": 3.4048582995951415,
+      "grad_norm": 0.29792506212153214,
+      "learning_rate": 1.1180653503944657e-05,
+      "loss": 0.29,
+      "step": 841
+    },
+    {
+      "epoch": 3.408906882591093,
+      "grad_norm": 0.3228078213443297,
+      "learning_rate": 1.1129930016486287e-05,
+      "loss": 0.2637,
+      "step": 842
+    },
+    {
+      "epoch": 3.4129554655870447,
+      "grad_norm": 0.31678532684497396,
+      "learning_rate": 1.107927745387114e-05,
+      "loss": 0.278,
+      "step": 843
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.3105355819415204,
+      "learning_rate": 1.1028696221115766e-05,
+      "loss": 0.2856,
+      "step": 844
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 0.3200376290464347,
+      "learning_rate": 1.0978186722666341e-05,
+      "loss": 0.2825,
+      "step": 845
+    },
+    {
+      "epoch": 3.425101214574899,
+      "grad_norm": 0.3325911174843037,
+      "learning_rate": 1.0927749362395475e-05,
+      "loss": 0.2778,
+      "step": 846
+    },
+    {
+      "epoch": 3.42914979757085,
+      "grad_norm": 0.31770529837323896,
+      "learning_rate": 1.0877384543598954e-05,
+      "loss": 0.2711,
+      "step": 847
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.3103891935777322,
+      "learning_rate": 1.0827092668992529e-05,
+      "loss": 0.2612,
+      "step": 848
+    },
+    {
+      "epoch": 3.437246963562753,
+      "grad_norm": 0.305313515909387,
+      "learning_rate": 1.077687414070869e-05,
+      "loss": 0.2721,
+      "step": 849
+    },
+    {
+      "epoch": 3.4412955465587043,
+      "grad_norm": 0.30006960226032,
+      "learning_rate": 1.0726729360293451e-05,
+      "loss": 0.2695,
+      "step": 850
+    },
+    {
+      "epoch": 3.445344129554656,
+      "grad_norm": 0.3071425297486285,
+      "learning_rate": 1.0676658728703147e-05,
+      "loss": 0.2839,
+      "step": 851
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.3096790740394744,
+      "learning_rate": 1.0626662646301198e-05,
+      "loss": 0.2771,
+      "step": 852
+    },
+    {
+      "epoch": 3.4534412955465585,
+      "grad_norm": 0.31449471327460393,
+      "learning_rate": 1.0576741512854958e-05,
+      "loss": 0.2614,
+      "step": 853
+    },
+    {
+      "epoch": 3.45748987854251,
+      "grad_norm": 0.3069207998713171,
+      "learning_rate": 1.052689572753247e-05,
+      "loss": 0.2749,
+      "step": 854
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 0.3009407478690319,
+      "learning_rate": 1.0477125688899328e-05,
+      "loss": 0.2783,
+      "step": 855
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.3338682129420618,
+      "learning_rate": 1.0427431794915406e-05,
+      "loss": 0.265,
+      "step": 856
+    },
+    {
+      "epoch": 3.4696356275303644,
+      "grad_norm": 0.3152281117883245,
+      "learning_rate": 1.0377814442931769e-05,
+      "loss": 0.2683,
+      "step": 857
+    },
+    {
+      "epoch": 3.473684210526316,
+      "grad_norm": 0.29773997985640205,
+      "learning_rate": 1.0328274029687439e-05,
+      "loss": 0.2875,
+      "step": 858
+    },
+    {
+      "epoch": 3.477732793522267,
+      "grad_norm": 0.3146321644808835,
+      "learning_rate": 1.0278810951306253e-05,
+      "loss": 0.2639,
+      "step": 859
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.29274159742091665,
+      "learning_rate": 1.0229425603293647e-05,
+      "loss": 0.2812,
+      "step": 860
+    },
+    {
+      "epoch": 3.48582995951417,
+      "grad_norm": 0.32038408955827347,
+      "learning_rate": 1.0180118380533558e-05,
+      "loss": 0.2317,
+      "step": 861
+    },
+    {
+      "epoch": 3.4898785425101213,
+      "grad_norm": 0.3227589578155879,
+      "learning_rate": 1.0130889677285225e-05,
+      "loss": 0.265,
+      "step": 862
+    },
+    {
+      "epoch": 3.493927125506073,
+      "grad_norm": 0.31690582749111634,
+      "learning_rate": 1.008173988718005e-05,
+      "loss": 0.2806,
+      "step": 863
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.30651867926174037,
+      "learning_rate": 1.0032669403218439e-05,
+      "loss": 0.2806,
+      "step": 864
+    },
+    {
+      "epoch": 3.5020242914979756,
+      "grad_norm": 0.30130131343636063,
+      "learning_rate": 9.983678617766677e-06,
+      "loss": 0.2807,
+      "step": 865
+    },
+    {
+      "epoch": 3.506072874493927,
+      "grad_norm": 0.3112938817010294,
+      "learning_rate": 9.934767922553788e-06,
+      "loss": 0.2695,
+      "step": 866
+    },
+    {
+      "epoch": 3.5101214574898787,
+      "grad_norm": 0.312041485068758,
+      "learning_rate": 9.885937708668371e-06,
+      "loss": 0.2656,
+      "step": 867
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.303806426727902,
+      "learning_rate": 9.837188366555524e-06,
+      "loss": 0.2663,
+      "step": 868
+    },
+    {
+      "epoch": 3.5182186234817814,
+      "grad_norm": 0.3006998158856867,
+      "learning_rate": 9.788520286013691e-06,
+      "loss": 0.3035,
+      "step": 869
+    },
+    {
+      "epoch": 3.522267206477733,
+      "grad_norm": 0.3210605619639771,
+      "learning_rate": 9.739933856191537e-06,
+      "loss": 0.2931,
+      "step": 870
+    },
+    {
+      "epoch": 3.526315789473684,
+      "grad_norm": 0.3152406656841439,
+      "learning_rate": 9.691429465584866e-06,
+      "loss": 0.2604,
+      "step": 871
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.3185103522787802,
+      "learning_rate": 9.643007502033492e-06,
+      "loss": 0.2866,
+      "step": 872
+    },
+    {
+      "epoch": 3.534412955465587,
+      "grad_norm": 0.2950505656363238,
+      "learning_rate": 9.594668352718152e-06,
+      "loss": 0.2779,
+      "step": 873
+    },
+    {
+      "epoch": 3.5384615384615383,
+      "grad_norm": 0.314158180648097,
+      "learning_rate": 9.546412404157383e-06,
+      "loss": 0.2669,
+      "step": 874
+    },
+    {
+      "epoch": 3.54251012145749,
+      "grad_norm": 0.3104234985066183,
+      "learning_rate": 9.498240042204471e-06,
+      "loss": 0.2639,
+      "step": 875
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.3181767927299884,
+      "learning_rate": 9.450151652044339e-06,
+      "loss": 0.2868,
+      "step": 876
+    },
+    {
+      "epoch": 3.5506072874493926,
+      "grad_norm": 0.30786240058823916,
+      "learning_rate": 9.402147618190472e-06,
+      "loss": 0.2687,
+      "step": 877
+    },
+    {
+      "epoch": 3.554655870445344,
+      "grad_norm": 0.31738549683618694,
+      "learning_rate": 9.354228324481841e-06,
+      "loss": 0.2755,
+      "step": 878
+    },
+    {
+      "epoch": 3.5587044534412957,
+      "grad_norm": 0.3012737978366645,
+      "learning_rate": 9.306394154079852e-06,
+      "loss": 0.2814,
+      "step": 879
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.33376111007476733,
+      "learning_rate": 9.258645489465254e-06,
+      "loss": 0.2745,
+      "step": 880
+    },
+    {
+      "epoch": 3.5668016194331984,
+      "grad_norm": 0.30920347534805925,
+      "learning_rate": 9.21098271243509e-06,
+      "loss": 0.2672,
+      "step": 881
+    },
+    {
+      "epoch": 3.57085020242915,
+      "grad_norm": 0.2950057668858617,
+      "learning_rate": 9.163406204099659e-06,
+      "loss": 0.2787,
+      "step": 882
+    },
+    {
+      "epoch": 3.574898785425101,
+      "grad_norm": 0.3143820240788232,
+      "learning_rate": 9.115916344879452e-06,
+      "loss": 0.2684,
+      "step": 883
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.323863769240369,
+      "learning_rate": 9.068513514502126e-06,
+      "loss": 0.2787,
+      "step": 884
+    },
+    {
+      "epoch": 3.582995951417004,
+      "grad_norm": 0.2983753877168157,
+      "learning_rate": 9.021198091999439e-06,
+      "loss": 0.2778,
+      "step": 885
+    },
+    {
+      "epoch": 3.5870445344129553,
+      "grad_norm": 0.30311042768958746,
+      "learning_rate": 8.973970455704251e-06,
+      "loss": 0.2874,
+      "step": 886
+    },
+    {
+      "epoch": 3.591093117408907,
+      "grad_norm": 0.2925363252086867,
+      "learning_rate": 8.926830983247495e-06,
+      "loss": 0.2963,
+      "step": 887
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.31878733236568113,
+      "learning_rate": 8.879780051555116e-06,
+      "loss": 0.2474,
+      "step": 888
+    },
+    {
+      "epoch": 3.5991902834008096,
+      "grad_norm": 0.30712039502479954,
+      "learning_rate": 8.83281803684512e-06,
+      "loss": 0.2542,
+      "step": 889
+    },
+    {
+      "epoch": 3.603238866396761,
+      "grad_norm": 0.3232893364655406,
+      "learning_rate": 8.785945314624532e-06,
+      "loss": 0.2762,
+      "step": 890
+    },
+    {
+      "epoch": 3.6072874493927127,
+      "grad_norm": 0.3129035559150589,
+      "learning_rate": 8.739162259686385e-06,
+      "loss": 0.2772,
+      "step": 891
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.33763910617987414,
+      "learning_rate": 8.692469246106734e-06,
+      "loss": 0.2475,
+      "step": 892
+    },
+    {
+      "epoch": 3.6153846153846154,
+      "grad_norm": 0.3010601219299761,
+      "learning_rate": 8.645866647241677e-06,
+      "loss": 0.2646,
+      "step": 893
+    },
+    {
+      "epoch": 3.619433198380567,
+      "grad_norm": 0.3158171555666817,
+      "learning_rate": 8.599354835724348e-06,
+      "loss": 0.2679,
+      "step": 894
+    },
+    {
+      "epoch": 3.623481781376518,
+      "grad_norm": 0.29908826749139616,
+      "learning_rate": 8.552934183461958e-06,
+      "loss": 0.2817,
+      "step": 895
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.28880783281083333,
+      "learning_rate": 8.506605061632805e-06,
+      "loss": 0.2763,
+      "step": 896
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 0.2839501989307659,
+      "learning_rate": 8.460367840683308e-06,
+      "loss": 0.2743,
+      "step": 897
+    },
+    {
+      "epoch": 3.6356275303643724,
+      "grad_norm": 0.29808892323448866,
+      "learning_rate": 8.414222890325065e-06,
+      "loss": 0.2814,
+      "step": 898
+    },
+    {
+      "epoch": 3.639676113360324,
+      "grad_norm": 0.29616457498543003,
+      "learning_rate": 8.368170579531855e-06,
+      "loss": 0.2826,
+      "step": 899
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.29343755141461325,
+      "learning_rate": 8.322211276536733e-06,
+      "loss": 0.2777,
+      "step": 900
+    },
+    {
+      "epoch": 3.6477732793522266,
+      "grad_norm": 0.2941901836297676,
+      "learning_rate": 8.27634534882907e-06,
+      "loss": 0.2841,
+      "step": 901
+    },
+    {
+      "epoch": 3.651821862348178,
+      "grad_norm": 0.2961260718227875,
+      "learning_rate": 8.230573163151585e-06,
+      "loss": 0.2646,
+      "step": 902
+    },
+    {
+      "epoch": 3.6558704453441297,
+      "grad_norm": 0.29634905985826626,
+      "learning_rate": 8.184895085497466e-06,
+      "loss": 0.2703,
+      "step": 903
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.3071544574675365,
+      "learning_rate": 8.139311481107401e-06,
+      "loss": 0.3011,
+      "step": 904
+    },
+    {
+      "epoch": 3.6639676113360324,
+      "grad_norm": 0.30837439265948674,
+      "learning_rate": 8.093822714466686e-06,
+      "loss": 0.2774,
+      "step": 905
+    },
+    {
+      "epoch": 3.668016194331984,
+      "grad_norm": 0.28910307479686614,
+      "learning_rate": 8.048429149302271e-06,
+      "loss": 0.288,
+      "step": 906
+    },
+    {
+      "epoch": 3.672064777327935,
+      "grad_norm": 0.2958206112515646,
+      "learning_rate": 8.003131148579902e-06,
+      "loss": 0.2775,
+      "step": 907
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.3050554265912037,
+      "learning_rate": 7.957929074501182e-06,
+      "loss": 0.2743,
+      "step": 908
+    },
+    {
+      "epoch": 3.6801619433198383,
+      "grad_norm": 0.297554210952206,
+      "learning_rate": 7.912823288500695e-06,
+      "loss": 0.25,
+      "step": 909
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 0.3147293499100331,
+      "learning_rate": 7.867814151243102e-06,
+      "loss": 0.2515,
+      "step": 910
+    },
+    {
+      "epoch": 3.688259109311741,
+      "grad_norm": 0.31667472939282215,
+      "learning_rate": 7.822902022620267e-06,
+      "loss": 0.284,
+      "step": 911
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.31410759742636557,
+      "learning_rate": 7.778087261748377e-06,
+      "loss": 0.2702,
+      "step": 912
+    },
+    {
+      "epoch": 3.6963562753036436,
+      "grad_norm": 0.3106400918649383,
+      "learning_rate": 7.733370226965056e-06,
+      "loss": 0.2541,
+      "step": 913
+    },
+    {
+      "epoch": 3.700404858299595,
+      "grad_norm": 0.314210484283767,
+      "learning_rate": 7.688751275826527e-06,
+      "loss": 0.2602,
+      "step": 914
+    },
+    {
+      "epoch": 3.7044534412955468,
+      "grad_norm": 0.30237066745063224,
+      "learning_rate": 7.644230765104735e-06,
+      "loss": 0.3047,
+      "step": 915
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.325240508930449,
+      "learning_rate": 7.599809050784502e-06,
+      "loss": 0.257,
+      "step": 916
+    },
+    {
+      "epoch": 3.7125506072874495,
+      "grad_norm": 0.30437866198977426,
+      "learning_rate": 7.555486488060661e-06,
+      "loss": 0.2728,
+      "step": 917
+    },
+    {
+      "epoch": 3.716599190283401,
+      "grad_norm": 0.3119600162964299,
+      "learning_rate": 7.511263431335249e-06,
+      "loss": 0.2806,
+      "step": 918
+    },
+    {
+      "epoch": 3.720647773279352,
+      "grad_norm": 0.32948846949593674,
+      "learning_rate": 7.467140234214656e-06,
+      "loss": 0.2819,
+      "step": 919
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.28952296764245083,
+      "learning_rate": 7.423117249506777e-06,
+      "loss": 0.2737,
+      "step": 920
+    },
+    {
+      "epoch": 3.7287449392712553,
+      "grad_norm": 0.29899466268797037,
+      "learning_rate": 7.379194829218231e-06,
+      "loss": 0.2697,
+      "step": 921
+    },
+    {
+      "epoch": 3.7327935222672064,
+      "grad_norm": 0.31429597903558626,
+      "learning_rate": 7.335373324551518e-06,
+      "loss": 0.256,
+      "step": 922
+    },
+    {
+      "epoch": 3.736842105263158,
+      "grad_norm": 0.3007888144763978,
+      "learning_rate": 7.291653085902224e-06,
+      "loss": 0.2625,
+      "step": 923
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.2968566410384066,
+      "learning_rate": 7.248034462856213e-06,
+      "loss": 0.2588,
+      "step": 924
+    },
+    {
+      "epoch": 3.7449392712550607,
+      "grad_norm": 0.2995126914533164,
+      "learning_rate": 7.20451780418683e-06,
+      "loss": 0.2764,
+      "step": 925
+    },
+    {
+      "epoch": 3.748987854251012,
+      "grad_norm": 0.325940625006179,
+      "learning_rate": 7.161103457852128e-06,
+      "loss": 0.2907,
+      "step": 926
+    },
+    {
+      "epoch": 3.753036437246964,
+      "grad_norm": 0.2927631861271194,
+      "learning_rate": 7.117791770992046e-06,
+      "loss": 0.2691,
+      "step": 927
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.30808138591550144,
+      "learning_rate": 7.074583089925688e-06,
+      "loss": 0.2759,
+      "step": 928
+    },
+    {
+      "epoch": 3.7611336032388665,
+      "grad_norm": 0.2949346659266875,
+      "learning_rate": 7.0314777601485155e-06,
+      "loss": 0.273,
+      "step": 929
+    },
+    {
+      "epoch": 3.765182186234818,
+      "grad_norm": 0.3172640649530825,
+      "learning_rate": 6.988476126329604e-06,
+      "loss": 0.263,
+      "step": 930
+    },
+    {
+      "epoch": 3.769230769230769,
+      "grad_norm": 0.2965274664624804,
+      "learning_rate": 6.945578532308857e-06,
+      "loss": 0.2714,
+      "step": 931
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.295336261160544,
+      "learning_rate": 6.902785321094301e-06,
+      "loss": 0.2857,
+      "step": 932
+    },
+    {
+      "epoch": 3.7773279352226723,
+      "grad_norm": 0.28598742477727024,
+      "learning_rate": 6.8600968348593115e-06,
+      "loss": 0.2687,
+      "step": 933
+    },
+    {
+      "epoch": 3.7813765182186234,
+      "grad_norm": 0.30762063924359984,
+      "learning_rate": 6.817513414939892e-06,
+      "loss": 0.2626,
+      "step": 934
+    },
+    {
+      "epoch": 3.785425101214575,
+      "grad_norm": 0.2974910150799796,
+      "learning_rate": 6.775035401831924e-06,
+      "loss": 0.2726,
+      "step": 935
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.30244130411300185,
+      "learning_rate": 6.732663135188473e-06,
+      "loss": 0.2584,
+      "step": 936
+    },
+    {
+      "epoch": 3.7935222672064777,
+      "grad_norm": 0.30776124858049336,
+      "learning_rate": 6.690396953817055e-06,
+      "loss": 0.2622,
+      "step": 937
+    },
+    {
+      "epoch": 3.7975708502024292,
+      "grad_norm": 0.30600744366636995,
+      "learning_rate": 6.648237195676927e-06,
+      "loss": 0.2653,
+      "step": 938
+    },
+    {
+      "epoch": 3.801619433198381,
+      "grad_norm": 0.30326817230489406,
+      "learning_rate": 6.606184197876395e-06,
+      "loss": 0.2893,
+      "step": 939
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.2833201620537575,
+      "learning_rate": 6.564238296670109e-06,
+      "loss": 0.2687,
+      "step": 940
+    },
+    {
+      "epoch": 3.8097165991902835,
+      "grad_norm": 0.29370406736713667,
+      "learning_rate": 6.52239982745638e-06,
+      "loss": 0.2618,
+      "step": 941
+    },
+    {
+      "epoch": 3.813765182186235,
+      "grad_norm": 0.2997169883979733,
+      "learning_rate": 6.480669124774484e-06,
+      "loss": 0.2622,
+      "step": 942
+    },
+    {
+      "epoch": 3.817813765182186,
+      "grad_norm": 0.29181697942984525,
+      "learning_rate": 6.439046522302013e-06,
+      "loss": 0.2793,
+      "step": 943
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.29586347097455135,
+      "learning_rate": 6.397532352852191e-06,
+      "loss": 0.2739,
+      "step": 944
+    },
+    {
+      "epoch": 3.8259109311740893,
+      "grad_norm": 1.9713705629898892,
+      "learning_rate": 6.356126948371202e-06,
+      "loss": 0.2905,
+      "step": 945
+    },
+    {
+      "epoch": 3.8299595141700404,
+      "grad_norm": 0.32254378527408173,
+      "learning_rate": 6.314830639935559e-06,
+      "loss": 0.2701,
+      "step": 946
+    },
+    {
+      "epoch": 3.834008097165992,
+      "grad_norm": 0.3028782341857832,
+      "learning_rate": 6.273643757749448e-06,
+      "loss": 0.2525,
+      "step": 947
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.299088045487407,
+      "learning_rate": 6.2325666311420856e-06,
+      "loss": 0.2701,
+      "step": 948
+    },
+    {
+      "epoch": 3.8421052631578947,
+      "grad_norm": 0.2920325147293162,
+      "learning_rate": 6.1915995885650675e-06,
+      "loss": 0.2538,
+      "step": 949
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 0.30051960662807575,
+      "learning_rate": 6.150742957589784e-06,
+      "loss": 0.2837,
+      "step": 950
+    },
+    {
+      "epoch": 3.850202429149798,
+      "grad_norm": 0.29315217723334375,
+      "learning_rate": 6.109997064904774e-06,
+      "loss": 0.2706,
+      "step": 951
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.2904602515850323,
+      "learning_rate": 6.069362236313092e-06,
+      "loss": 0.2986,
+      "step": 952
+    },
+    {
+      "epoch": 3.8582995951417005,
+      "grad_norm": 0.28421312271205634,
+      "learning_rate": 6.0288387967297565e-06,
+      "loss": 0.287,
+      "step": 953
+    },
+    {
+      "epoch": 3.862348178137652,
+      "grad_norm": 0.29617949060689586,
+      "learning_rate": 5.988427070179108e-06,
+      "loss": 0.2649,
+      "step": 954
+    },
+    {
+      "epoch": 3.866396761133603,
+      "grad_norm": 0.2900983025021768,
+      "learning_rate": 5.948127379792234e-06,
+      "loss": 0.2713,
+      "step": 955
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.2963303603281304,
+      "learning_rate": 5.90794004780439e-06,
+      "loss": 0.2826,
+      "step": 956
+    },
+    {
+      "epoch": 3.8744939271255063,
+      "grad_norm": 0.3081885139761003,
+      "learning_rate": 5.867865395552405e-06,
+      "loss": 0.263,
+      "step": 957
+    },
+    {
+      "epoch": 3.8785425101214575,
+      "grad_norm": 0.2992111302243632,
+      "learning_rate": 5.8279037434721364e-06,
+      "loss": 0.2651,
+      "step": 958
+    },
+    {
+      "epoch": 3.882591093117409,
+      "grad_norm": 0.2918843687317841,
+      "learning_rate": 5.788055411095892e-06,
+      "loss": 0.2806,
+      "step": 959
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.2828082057707741,
+      "learning_rate": 5.7483207170498625e-06,
+      "loss": 0.291,
+      "step": 960
+    },
+    {
+      "epoch": 3.8906882591093117,
+      "grad_norm": 0.28972107584740686,
+      "learning_rate": 5.708699979051613e-06,
+      "loss": 0.2623,
+      "step": 961
+    },
+    {
+      "epoch": 3.8947368421052633,
+      "grad_norm": 0.30505843738728183,
+      "learning_rate": 5.66919351390751e-06,
+      "loss": 0.2701,
+      "step": 962
+    },
+    {
+      "epoch": 3.898785425101215,
+      "grad_norm": 0.283270789572354,
+      "learning_rate": 5.629801637510188e-06,
+      "loss": 0.2443,
+      "step": 963
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.2896835353830917,
+      "learning_rate": 5.590524664836043e-06,
+      "loss": 0.2906,
+      "step": 964
+    },
+    {
+      "epoch": 3.9068825910931175,
+      "grad_norm": 0.29837673527025155,
+      "learning_rate": 5.551362909942708e-06,
+      "loss": 0.264,
+      "step": 965
+    },
+    {
+      "epoch": 3.910931174089069,
+      "grad_norm": 0.29057673781801924,
+      "learning_rate": 5.51231668596653e-06,
+      "loss": 0.2397,
+      "step": 966
+    },
+    {
+      "epoch": 3.91497975708502,
+      "grad_norm": 0.2974295313765325,
+      "learning_rate": 5.47338630512007e-06,
+      "loss": 0.2513,
+      "step": 967
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.3152063323975208,
+      "learning_rate": 5.434572078689617e-06,
+      "loss": 0.2821,
+      "step": 968
+    },
+    {
+      "epoch": 3.9230769230769234,
+      "grad_norm": 0.2927490704950411,
+      "learning_rate": 5.395874317032694e-06,
+      "loss": 0.2597,
+      "step": 969
+    },
+    {
+      "epoch": 3.9271255060728745,
+      "grad_norm": 0.2973606701273386,
+      "learning_rate": 5.357293329575573e-06,
+      "loss": 0.2499,
+      "step": 970
+    },
+    {
+      "epoch": 3.931174089068826,
+      "grad_norm": 0.3142276544408366,
+      "learning_rate": 5.318829424810797e-06,
+      "loss": 0.2729,
+      "step": 971
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.3077886847312571,
+      "learning_rate": 5.280482910294728e-06,
+      "loss": 0.2405,
+      "step": 972
+    },
+    {
+      "epoch": 3.9392712550607287,
+      "grad_norm": 0.3083931387866309,
+      "learning_rate": 5.242254092645078e-06,
+      "loss": 0.2563,
+      "step": 973
+    },
+    {
+      "epoch": 3.9433198380566803,
+      "grad_norm": 0.28859082668133634,
+      "learning_rate": 5.2041432775384445e-06,
+      "loss": 0.2715,
+      "step": 974
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 0.2993959157567042,
+      "learning_rate": 5.166150769707892e-06,
+      "loss": 0.2783,
+      "step": 975
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.2891695557599385,
+      "learning_rate": 5.128276872940503e-06,
+      "loss": 0.2543,
+      "step": 976
+    },
+    {
+      "epoch": 3.9554655870445345,
+      "grad_norm": 0.30120043186784656,
+      "learning_rate": 5.090521890074953e-06,
+      "loss": 0.2627,
+      "step": 977
+    },
+    {
+      "epoch": 3.9595141700404857,
+      "grad_norm": 0.2955799646517998,
+      "learning_rate": 5.0528861229990675e-06,
+      "loss": 0.2704,
+      "step": 978
+    },
+    {
+      "epoch": 3.9635627530364372,
+      "grad_norm": 0.3184821500407624,
+      "learning_rate": 5.015369872647442e-06,
+      "loss": 0.2713,
+      "step": 979
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.3050441836266399,
+      "learning_rate": 4.9779734389990195e-06,
+      "loss": 0.2491,
+      "step": 980
+    },
+    {
+      "epoch": 3.97165991902834,
+      "grad_norm": 0.2993070529704862,
+      "learning_rate": 4.940697121074675e-06,
+      "loss": 0.273,
+      "step": 981
+    },
+    {
+      "epoch": 3.9757085020242915,
+      "grad_norm": 0.31485106821803027,
+      "learning_rate": 4.903541216934864e-06,
+      "loss": 0.2578,
+      "step": 982
+    },
+    {
+      "epoch": 3.979757085020243,
+      "grad_norm": 0.3113069529091008,
+      "learning_rate": 4.8665060236772e-06,
+      "loss": 0.2841,
+      "step": 983
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.2907990742359794,
+      "learning_rate": 4.829591837434102e-06,
+      "loss": 0.2425,
+      "step": 984
+    },
+    {
+      "epoch": 3.9878542510121457,
+      "grad_norm": 0.30432401260512965,
+      "learning_rate": 4.792798953370424e-06,
+      "loss": 0.2656,
+      "step": 985
+    },
+    {
+      "epoch": 3.9919028340080973,
+      "grad_norm": 0.295412502321952,
+      "learning_rate": 4.756127665681083e-06,
+      "loss": 0.275,
+      "step": 986
+    },
+    {
+      "epoch": 3.9959514170040484,
+      "grad_norm": 0.30208766532466297,
+      "learning_rate": 4.719578267588725e-06,
+      "loss": 0.257,
+      "step": 987
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.2884166650429656,
+      "learning_rate": 4.6831510513413525e-06,
+      "loss": 0.2742,
+      "step": 988
+    },
+    {
+      "epoch": 4.004048582995951,
+      "grad_norm": 0.3752678025569914,
+      "learning_rate": 4.6468463082100176e-06,
+      "loss": 0.2447,
+      "step": 989
+    },
+    {
+      "epoch": 4.008097165991903,
+      "grad_norm": 0.35423956032904913,
+      "learning_rate": 4.610664328486485e-06,
+      "loss": 0.2088,
+      "step": 990
+    },
+    {
+      "epoch": 4.012145748987854,
+      "grad_norm": 0.29525117389312017,
+      "learning_rate": 4.5746054014809e-06,
+      "loss": 0.1994,
+      "step": 991
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 1.2236628498099826,
+      "learning_rate": 4.5386698155194766e-06,
+      "loss": 0.2321,
+      "step": 992
+    },
+    {
+      "epoch": 4.020242914979757,
+      "grad_norm": 0.345023452086892,
+      "learning_rate": 4.502857857942204e-06,
+      "loss": 0.223,
+      "step": 993
+    },
+    {
+      "epoch": 4.0242914979757085,
+      "grad_norm": 0.42701929672626066,
+      "learning_rate": 4.467169815100549e-06,
+      "loss": 0.2211,
+      "step": 994
+    },
+    {
+      "epoch": 4.02834008097166,
+      "grad_norm": 0.36807373350261813,
+      "learning_rate": 4.431605972355137e-06,
+      "loss": 0.1973,
+      "step": 995
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.32481827576909666,
+      "learning_rate": 4.396166614073516e-06,
+      "loss": 0.2282,
+      "step": 996
+    },
+    {
+      "epoch": 4.036437246963563,
+      "grad_norm": 0.31006838225109584,
+      "learning_rate": 4.360852023627846e-06,
+      "loss": 0.2325,
+      "step": 997
+    },
+    {
+      "epoch": 4.040485829959514,
+      "grad_norm": 0.3299858223679757,
+      "learning_rate": 4.325662483392656e-06,
+      "loss": 0.2206,
+      "step": 998
+    },
+    {
+      "epoch": 4.044534412955466,
+      "grad_norm": 0.34122525353432803,
+      "learning_rate": 4.29059827474257e-06,
+      "loss": 0.2056,
+      "step": 999
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.3211693029673518,
+      "learning_rate": 4.255659678050063e-06,
+      "loss": 0.2066,
+      "step": 1000
+    },
+    {
+      "epoch": 4.052631578947368,
+      "grad_norm": 0.2769062535716997,
+      "learning_rate": 4.22084697268323e-06,
+      "loss": 0.221,
+      "step": 1001
+    },
+    {
+      "epoch": 4.05668016194332,
+      "grad_norm": 0.27631521915958424,
+      "learning_rate": 4.186160437003535e-06,
+      "loss": 0.2145,
+      "step": 1002
+    },
+    {
+      "epoch": 4.060728744939271,
+      "grad_norm": 0.30321278663568146,
+      "learning_rate": 4.151600348363582e-06,
+      "loss": 0.2071,
+      "step": 1003
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.3253906192480427,
+      "learning_rate": 4.11716698310493e-06,
+      "loss": 0.231,
+      "step": 1004
+    },
+    {
+      "epoch": 4.068825910931174,
+      "grad_norm": 0.2914612045119375,
+      "learning_rate": 4.082860616555848e-06,
+      "loss": 0.2112,
+      "step": 1005
+    },
+    {
+      "epoch": 4.0728744939271255,
+      "grad_norm": 0.27191431804442945,
+      "learning_rate": 4.048681523029125e-06,
+      "loss": 0.2122,
+      "step": 1006
+    },
+    {
+      "epoch": 4.076923076923077,
+      "grad_norm": 0.27661708342485747,
+      "learning_rate": 4.014629975819882e-06,
+      "loss": 0.2288,
+      "step": 1007
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.27932893636425665,
+      "learning_rate": 3.9807062472033855e-06,
+      "loss": 0.2117,
+      "step": 1008
+    },
+    {
+      "epoch": 4.08502024291498,
+      "grad_norm": 0.2812547493156742,
+      "learning_rate": 3.94691060843287e-06,
+      "loss": 0.2083,
+      "step": 1009
+    },
+    {
+      "epoch": 4.089068825910931,
+      "grad_norm": 0.27458472787179483,
+      "learning_rate": 3.913243329737353e-06,
+      "loss": 0.2119,
+      "step": 1010
+    },
+    {
+      "epoch": 4.093117408906883,
+      "grad_norm": 0.2853311993477863,
+      "learning_rate": 3.879704680319503e-06,
+      "loss": 0.2203,
+      "step": 1011
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.2695588728535248,
+      "learning_rate": 3.846294928353471e-06,
+      "loss": 0.2244,
+      "step": 1012
+    },
+    {
+      "epoch": 4.101214574898785,
+      "grad_norm": 0.2593053921060652,
+      "learning_rate": 3.8130143409827345e-06,
+      "loss": 0.2255,
+      "step": 1013
+    },
+    {
+      "epoch": 4.105263157894737,
+      "grad_norm": 0.26252437493331776,
+      "learning_rate": 3.7798631843179887e-06,
+      "loss": 0.2104,
+      "step": 1014
+    },
+    {
+      "epoch": 4.109311740890688,
+      "grad_norm": 0.27473878125320517,
+      "learning_rate": 3.7468417234349973e-06,
+      "loss": 0.2147,
+      "step": 1015
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.25579948669790714,
+      "learning_rate": 3.7139502223724823e-06,
+      "loss": 0.2095,
+      "step": 1016
+    },
+    {
+      "epoch": 4.117408906882591,
+      "grad_norm": 0.24331392499448604,
+      "learning_rate": 3.6811889441300095e-06,
+      "loss": 0.2061,
+      "step": 1017
+    },
+    {
+      "epoch": 4.1214574898785425,
+      "grad_norm": 0.25695612432853626,
+      "learning_rate": 3.648558150665886e-06,
+      "loss": 0.1947,
+      "step": 1018
+    },
+    {
+      "epoch": 4.125506072874494,
+      "grad_norm": 0.25709049268963935,
+      "learning_rate": 3.616058102895075e-06,
+      "loss": 0.2054,
+      "step": 1019
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.24537854762697667,
+      "learning_rate": 3.583689060687079e-06,
+      "loss": 0.2185,
+      "step": 1020
+    },
+    {
+      "epoch": 4.133603238866397,
+      "grad_norm": 0.26340596079683964,
+      "learning_rate": 3.5514512828639023e-06,
+      "loss": 0.2284,
+      "step": 1021
+    },
+    {
+      "epoch": 4.137651821862348,
+      "grad_norm": 0.24883566738940108,
+      "learning_rate": 3.51934502719796e-06,
+      "loss": 0.221,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1417004048583,
+      "grad_norm": 0.26034212089668557,
+      "learning_rate": 3.4873705504100185e-06,
+      "loss": 0.234,
+      "step": 1023
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.25132861101961984,
+      "learning_rate": 3.4555281081671345e-06,
+      "loss": 0.1982,
+      "step": 1024
+    },
+    {
+      "epoch": 4.149797570850202,
+      "grad_norm": 0.26008994160222043,
+      "learning_rate": 3.4238179550806326e-06,
+      "loss": 0.203,
+      "step": 1025
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.26345263186370305,
+      "learning_rate": 3.3922403447040497e-06,
+      "loss": 0.2144,
+      "step": 1026
+    },
+    {
+      "epoch": 4.157894736842105,
+      "grad_norm": 0.25577420542351803,
+      "learning_rate": 3.3607955295311247e-06,
+      "loss": 0.2044,
+      "step": 1027
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.26092356082759044,
+      "learning_rate": 3.32948376099375e-06,
+      "loss": 0.2107,
+      "step": 1028
+    },
+    {
+      "epoch": 4.165991902834008,
+      "grad_norm": 0.25073604779682457,
+      "learning_rate": 3.2983052894599976e-06,
+      "loss": 0.2104,
+      "step": 1029
+    },
+    {
+      "epoch": 4.17004048582996,
+      "grad_norm": 0.24927469440060024,
+      "learning_rate": 3.2672603642320945e-06,
+      "loss": 0.2186,
+      "step": 1030
+    },
+    {
+      "epoch": 4.174089068825911,
+      "grad_norm": 0.2557981306058831,
+      "learning_rate": 3.236349233544438e-06,
+      "loss": 0.2187,
+      "step": 1031
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.24517497135089886,
+      "learning_rate": 3.205572144561604e-06,
+      "loss": 0.201,
+      "step": 1032
+    },
+    {
+      "epoch": 4.182186234817814,
+      "grad_norm": 0.25307193362368996,
+      "learning_rate": 3.174929343376374e-06,
+      "loss": 0.2297,
+      "step": 1033
+    },
+    {
+      "epoch": 4.186234817813765,
+      "grad_norm": 0.25097757353157907,
+      "learning_rate": 3.1444210750077777e-06,
+      "loss": 0.2123,
+      "step": 1034
+    },
+    {
+      "epoch": 4.190283400809717,
+      "grad_norm": 0.24884008354279163,
+      "learning_rate": 3.1140475833991046e-06,
+      "loss": 0.195,
+      "step": 1035
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.25625412916421997,
+      "learning_rate": 3.0838091114159895e-06,
+      "loss": 0.2083,
+      "step": 1036
+    },
+    {
+      "epoch": 4.198380566801619,
+      "grad_norm": 0.2522426293419729,
+      "learning_rate": 3.0537059008444524e-06,
+      "loss": 0.1848,
+      "step": 1037
+    },
+    {
+      "epoch": 4.202429149797571,
+      "grad_norm": 0.24324917065526844,
+      "learning_rate": 3.0237381923889563e-06,
+      "loss": 0.2071,
+      "step": 1038
+    },
+    {
+      "epoch": 4.206477732793522,
+      "grad_norm": 0.25859258079710656,
+      "learning_rate": 2.9939062256704997e-06,
+      "loss": 0.1961,
+      "step": 1039
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.2506470625281633,
+      "learning_rate": 2.964210239224696e-06,
+      "loss": 0.1967,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2145748987854255,
+      "grad_norm": 0.2422043982200552,
+      "learning_rate": 2.934650470499869e-06,
+      "loss": 0.2256,
+      "step": 1041
+    },
+    {
+      "epoch": 4.218623481781377,
+      "grad_norm": 0.24231125708627424,
+      "learning_rate": 2.905227155855128e-06,
+      "loss": 0.2023,
+      "step": 1042
+    },
+    {
+      "epoch": 4.222672064777328,
+      "grad_norm": 0.25018001783329424,
+      "learning_rate": 2.875940530558519e-06,
+      "loss": 0.2293,
+      "step": 1043
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.2391884069224687,
+      "learning_rate": 2.8467908287851154e-06,
+      "loss": 0.1933,
+      "step": 1044
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 0.2438002595630391,
+      "learning_rate": 2.8177782836151603e-06,
+      "loss": 0.1963,
+      "step": 1045
+    },
+    {
+      "epoch": 4.234817813765182,
+      "grad_norm": 0.24407111874973098,
+      "learning_rate": 2.7889031270321874e-06,
+      "loss": 0.2112,
+      "step": 1046
+    },
+    {
+      "epoch": 4.238866396761134,
+      "grad_norm": 0.25317011697009467,
+      "learning_rate": 2.760165589921184e-06,
+      "loss": 0.2098,
+      "step": 1047
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.23075642189198725,
+      "learning_rate": 2.7315659020667305e-06,
+      "loss": 0.2056,
+      "step": 1048
+    },
+    {
+      "epoch": 4.246963562753036,
+      "grad_norm": 0.23888127053825747,
+      "learning_rate": 2.703104292151164e-06,
+      "loss": 0.1888,
+      "step": 1049
+    },
+    {
+      "epoch": 4.251012145748988,
+      "grad_norm": 0.2563142202350936,
+      "learning_rate": 2.6747809877527607e-06,
+      "loss": 0.2192,
+      "step": 1050
+    },
+    {
+      "epoch": 4.255060728744939,
+      "grad_norm": 0.24576657179433392,
+      "learning_rate": 2.6465962153439082e-06,
+      "loss": 0.2161,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.2526235153557521,
+      "learning_rate": 2.618550200289303e-06,
+      "loss": 0.2135,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2631578947368425,
+      "grad_norm": 0.2441747363952628,
+      "learning_rate": 2.59064316684412e-06,
+      "loss": 0.2234,
+      "step": 1053
+    },
+    {
+      "epoch": 4.267206477732794,
+      "grad_norm": 0.2425169886750406,
+      "learning_rate": 2.562875338152264e-06,
+      "loss": 0.2323,
+      "step": 1054
+    },
+    {
+      "epoch": 4.271255060728745,
+      "grad_norm": 0.22947179866635756,
+      "learning_rate": 2.5352469362445552e-06,
+      "loss": 0.2061,
+      "step": 1055
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.2489130104906103,
+      "learning_rate": 2.5077581820369523e-06,
+      "loss": 0.2144,
+      "step": 1056
+    },
+    {
+      "epoch": 4.279352226720648,
+      "grad_norm": 0.25030858911901893,
+      "learning_rate": 2.480409295328803e-06,
+      "loss": 0.2076,
+      "step": 1057
+    },
+    {
+      "epoch": 4.283400809716599,
+      "grad_norm": 0.25964377772927977,
+      "learning_rate": 2.45320049480108e-06,
+      "loss": 0.2262,
+      "step": 1058
+    },
+    {
+      "epoch": 4.287449392712551,
+      "grad_norm": 0.25365385034199567,
+      "learning_rate": 2.4261319980146293e-06,
+      "loss": 0.2145,
+      "step": 1059
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.24110356664634983,
+      "learning_rate": 2.399204021408421e-06,
+      "loss": 0.2002,
+      "step": 1060
+    },
+    {
+      "epoch": 4.295546558704453,
+      "grad_norm": 0.2496939628939963,
+      "learning_rate": 2.3724167802978395e-06,
+      "loss": 0.1955,
+      "step": 1061
+    },
+    {
+      "epoch": 4.299595141700405,
+      "grad_norm": 0.23856511279985093,
+      "learning_rate": 2.3457704888729625e-06,
+      "loss": 0.1883,
+      "step": 1062
+    },
+    {
+      "epoch": 4.303643724696356,
+      "grad_norm": 0.251786086048144,
+      "learning_rate": 2.3192653601968162e-06,
+      "loss": 0.2088,
+      "step": 1063
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.25126060226877167,
+      "learning_rate": 2.2929016062037057e-06,
+      "loss": 0.1994,
+      "step": 1064
+    },
+    {
+      "epoch": 4.3117408906882595,
+      "grad_norm": 0.24730518357391065,
+      "learning_rate": 2.2666794376975055e-06,
+      "loss": 0.2223,
+      "step": 1065
+    },
+    {
+      "epoch": 4.315789473684211,
+      "grad_norm": 0.24819902279930176,
+      "learning_rate": 2.24059906434998e-06,
+      "loss": 0.195,
+      "step": 1066
+    },
+    {
+      "epoch": 4.319838056680162,
+      "grad_norm": 0.24765832389973969,
+      "learning_rate": 2.214660694699091e-06,
+      "loss": 0.207,
+      "step": 1067
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.24983475451981682,
+      "learning_rate": 2.188864536147355e-06,
+      "loss": 0.2262,
+      "step": 1068
+    },
+    {
+      "epoch": 4.327935222672065,
+      "grad_norm": 0.24724895694276455,
+      "learning_rate": 2.1632107949601757e-06,
+      "loss": 0.2225,
+      "step": 1069
+    },
+    {
+      "epoch": 4.331983805668016,
+      "grad_norm": 0.2482941424396293,
+      "learning_rate": 2.137699676264171e-06,
+      "loss": 0.2297,
+      "step": 1070
+    },
+    {
+      "epoch": 4.336032388663968,
+      "grad_norm": 0.24519973514187562,
+      "learning_rate": 2.1123313840455738e-06,
+      "loss": 0.2203,
+      "step": 1071
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.2571072645423452,
+      "learning_rate": 2.08710612114857e-06,
+      "loss": 0.224,
+      "step": 1072
+    },
+    {
+      "epoch": 4.34412955465587,
+      "grad_norm": 0.247376296285096,
+      "learning_rate": 2.0620240892736953e-06,
+      "loss": 0.1963,
+      "step": 1073
+    },
+    {
+      "epoch": 4.348178137651822,
+      "grad_norm": 0.24500771907868066,
+      "learning_rate": 2.0370854889762e-06,
+      "loss": 0.2103,
+      "step": 1074
+    },
+    {
+      "epoch": 4.352226720647773,
+      "grad_norm": 0.2403974440127442,
+      "learning_rate": 2.0122905196644726e-06,
+      "loss": 0.2016,
+      "step": 1075
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.25018074802433665,
+      "learning_rate": 1.987639379598427e-06,
+      "loss": 0.2444,
+      "step": 1076
+    },
+    {
+      "epoch": 4.3603238866396765,
+      "grad_norm": 0.2426288334712081,
+      "learning_rate": 1.963132265887919e-06,
+      "loss": 0.2101,
+      "step": 1077
+    },
+    {
+      "epoch": 4.364372469635628,
+      "grad_norm": 0.24788082377581608,
+      "learning_rate": 1.9387693744911784e-06,
+      "loss": 0.1928,
+      "step": 1078
+    },
+    {
+      "epoch": 4.368421052631579,
+      "grad_norm": 0.24463921336417258,
+      "learning_rate": 1.9145509002132368e-06,
+      "loss": 0.1862,
+      "step": 1079
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.24780050222950595,
+      "learning_rate": 1.8904770367043723e-06,
+      "loss": 0.2198,
+      "step": 1080
+    },
+    {
+      "epoch": 4.376518218623482,
+      "grad_norm": 0.24763749553394254,
+      "learning_rate": 1.8665479764585504e-06,
+      "loss": 0.2132,
+      "step": 1081
+    },
+    {
+      "epoch": 4.380566801619433,
+      "grad_norm": 0.23899308680404033,
+      "learning_rate": 1.8427639108118977e-06,
+      "loss": 0.1956,
+      "step": 1082
+    },
+    {
+      "epoch": 4.384615384615385,
+      "grad_norm": 0.24513928341536642,
+      "learning_rate": 1.8191250299411755e-06,
+      "loss": 0.209,
+      "step": 1083
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.2615537885572059,
+      "learning_rate": 1.7956315228622446e-06,
+      "loss": 0.1985,
+      "step": 1084
+    },
+    {
+      "epoch": 4.392712550607287,
+      "grad_norm": 0.24541510988760926,
+      "learning_rate": 1.7722835774285553e-06,
+      "loss": 0.2197,
+      "step": 1085
+    },
+    {
+      "epoch": 4.396761133603239,
+      "grad_norm": 0.25748370806131965,
+      "learning_rate": 1.7490813803296624e-06,
+      "loss": 0.211,
+      "step": 1086
+    },
+    {
+      "epoch": 4.40080971659919,
+      "grad_norm": 0.24053155330938414,
+      "learning_rate": 1.726025117089718e-06,
+      "loss": 0.2083,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.25204865641653207,
+      "learning_rate": 1.70311497206598e-06,
+      "loss": 0.2024,
+      "step": 1088
+    },
+    {
+      "epoch": 4.4089068825910935,
+      "grad_norm": 0.2514022069136545,
+      "learning_rate": 1.6803511284473638e-06,
+      "loss": 0.2126,
+      "step": 1089
+    },
+    {
+      "epoch": 4.412955465587045,
+      "grad_norm": 0.24631120261878314,
+      "learning_rate": 1.6577337682529538e-06,
+      "loss": 0.2214,
+      "step": 1090
+    },
+    {
+      "epoch": 4.417004048582996,
+      "grad_norm": 0.24575397658275766,
+      "learning_rate": 1.6352630723305617e-06,
+      "loss": 0.2105,
+      "step": 1091
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.2323366978205727,
+      "learning_rate": 1.612939220355272e-06,
+      "loss": 0.1899,
+      "step": 1092
+    },
+    {
+      "epoch": 4.425101214574899,
+      "grad_norm": 0.24742271122875792,
+      "learning_rate": 1.5907623908280112e-06,
+      "loss": 0.1997,
+      "step": 1093
+    },
+    {
+      "epoch": 4.42914979757085,
+      "grad_norm": 0.2504845142094529,
+      "learning_rate": 1.568732761074121e-06,
+      "loss": 0.2237,
+      "step": 1094
+    },
+    {
+      "epoch": 4.433198380566802,
+      "grad_norm": 0.2378120051432767,
+      "learning_rate": 1.5468505072419237e-06,
+      "loss": 0.2033,
+      "step": 1095
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.247061132497535,
+      "learning_rate": 1.5251158043013448e-06,
+      "loss": 0.2185,
+      "step": 1096
+    },
+    {
+      "epoch": 4.441295546558704,
+      "grad_norm": 0.2373099041710281,
+      "learning_rate": 1.5035288260424885e-06,
+      "loss": 0.2012,
+      "step": 1097
+    },
+    {
+      "epoch": 4.445344129554655,
+      "grad_norm": 0.24267816589936095,
+      "learning_rate": 1.4820897450742623e-06,
+      "loss": 0.224,
+      "step": 1098
+    },
+    {
+      "epoch": 4.449392712550607,
+      "grad_norm": 0.24480131888318496,
+      "learning_rate": 1.4607987328229767e-06,
+      "loss": 0.2098,
+      "step": 1099
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.25426219945414547,
+      "learning_rate": 1.4396559595310056e-06,
+      "loss": 0.2179,
+      "step": 1100
+    },
+    {
+      "epoch": 4.4574898785425106,
+      "grad_norm": 0.24622037957607806,
+      "learning_rate": 1.4186615942553973e-06,
+      "loss": 0.225,
+      "step": 1101
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.23433119269718408,
+      "learning_rate": 1.3978158048665424e-06,
+      "loss": 0.1996,
+      "step": 1102
+    },
+    {
+      "epoch": 4.465587044534413,
+      "grad_norm": 0.2497107858225183,
+      "learning_rate": 1.3771187580468115e-06,
+      "loss": 0.1839,
+      "step": 1103
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.23926073223376612,
+      "learning_rate": 1.3565706192892392e-06,
+      "loss": 0.2207,
+      "step": 1104
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 0.24126437337760445,
+      "learning_rate": 1.3361715528961993e-06,
+      "loss": 0.2065,
+      "step": 1105
+    },
+    {
+      "epoch": 4.477732793522267,
+      "grad_norm": 0.2466132222718118,
+      "learning_rate": 1.3159217219780817e-06,
+      "loss": 0.2099,
+      "step": 1106
+    },
+    {
+      "epoch": 4.481781376518219,
+      "grad_norm": 0.2512150880026158,
+      "learning_rate": 1.2958212884519949e-06,
+      "loss": 0.2027,
+      "step": 1107
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.25708067112204946,
+      "learning_rate": 1.2758704130404675e-06,
+      "loss": 0.2022,
+      "step": 1108
+    },
+    {
+      "epoch": 4.489878542510121,
+      "grad_norm": 0.23976030938943438,
+      "learning_rate": 1.2560692552701738e-06,
+      "loss": 0.1973,
+      "step": 1109
+    },
+    {
+      "epoch": 4.493927125506072,
+      "grad_norm": 0.24321576961581318,
+      "learning_rate": 1.2364179734706382e-06,
+      "loss": 0.2257,
+      "step": 1110
+    },
+    {
+      "epoch": 4.497975708502024,
+      "grad_norm": 0.24089884510909695,
+      "learning_rate": 1.2169167247729852e-06,
+      "loss": 0.2186,
+      "step": 1111
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.24344418670288692,
+      "learning_rate": 1.1975656651086864e-06,
+      "loss": 0.2277,
+      "step": 1112
+    },
+    {
+      "epoch": 4.506072874493928,
+      "grad_norm": 0.24553106770694758,
+      "learning_rate": 1.1783649492082926e-06,
+      "loss": 0.2007,
+      "step": 1113
+    },
+    {
+      "epoch": 4.510121457489879,
+      "grad_norm": 0.2438356659608302,
+      "learning_rate": 1.1593147306002183e-06,
+      "loss": 0.195,
+      "step": 1114
+    },
+    {
+      "epoch": 4.51417004048583,
+      "grad_norm": 0.25022430694879305,
+      "learning_rate": 1.140415161609505e-06,
+      "loss": 0.1993,
+      "step": 1115
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.2595467547122747,
+      "learning_rate": 1.121666393356604e-06,
+      "loss": 0.2317,
+      "step": 1116
+    },
+    {
+      "epoch": 4.522267206477733,
+      "grad_norm": 0.25526884454501647,
+      "learning_rate": 1.1030685757561632e-06,
+      "loss": 0.2042,
+      "step": 1117
+    },
+    {
+      "epoch": 4.526315789473684,
+      "grad_norm": 0.23547596421053177,
+      "learning_rate": 1.0846218575158373e-06,
+      "loss": 0.2046,
+      "step": 1118
+    },
+    {
+      "epoch": 4.530364372469636,
+      "grad_norm": 0.23917059427942172,
+      "learning_rate": 1.066326386135097e-06,
+      "loss": 0.1987,
+      "step": 1119
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.25512574682368866,
+      "learning_rate": 1.0481823079040398e-06,
+      "loss": 0.2289,
+      "step": 1120
+    },
+    {
+      "epoch": 4.538461538461538,
+      "grad_norm": 0.2509705064247709,
+      "learning_rate": 1.0301897679022344e-06,
+      "loss": 0.212,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5425101214574894,
+      "grad_norm": 0.22867719826864713,
+      "learning_rate": 1.0123489099975491e-06,
+      "loss": 0.2288,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5465587044534415,
+      "grad_norm": 0.24312193708308413,
+      "learning_rate": 9.9465987684501e-07,
+      "loss": 0.2084,
+      "step": 1123
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.2624379931643963,
+      "learning_rate": 9.771228098856534e-07,
+      "loss": 0.2213,
+      "step": 1124
+    },
+    {
+      "epoch": 4.554655870445345,
+      "grad_norm": 0.23788036032567067,
+      "learning_rate": 9.597378493454013e-07,
+      "loss": 0.2143,
+      "step": 1125
+    },
+    {
+      "epoch": 4.558704453441296,
+      "grad_norm": 0.24591656331402503,
+      "learning_rate": 9.425051342339353e-07,
+      "loss": 0.212,
+      "step": 1126
+    },
+    {
+      "epoch": 4.562753036437247,
+      "grad_norm": 0.2407009910008996,
+      "learning_rate": 9.254248023435864e-07,
+      "loss": 0.2302,
+      "step": 1127
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.2503740428362791,
+      "learning_rate": 9.084969902482288e-07,
+      "loss": 0.2072,
+      "step": 1128
+    },
+    {
+      "epoch": 4.57085020242915,
+      "grad_norm": 0.23530589803479376,
+      "learning_rate": 8.917218333021993e-07,
+      "loss": 0.2102,
+      "step": 1129
+    },
+    {
+      "epoch": 4.574898785425101,
+      "grad_norm": 0.24658214336188683,
+      "learning_rate": 8.750994656392109e-07,
+      "loss": 0.2052,
+      "step": 1130
+    },
+    {
+      "epoch": 4.578947368421053,
+      "grad_norm": 0.24786794636658907,
+      "learning_rate": 8.58630020171265e-07,
+      "loss": 0.2201,
+      "step": 1131
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.2425571725631908,
+      "learning_rate": 8.423136285876099e-07,
+      "loss": 0.2249,
+      "step": 1132
+    },
+    {
+      "epoch": 4.587044534412955,
+      "grad_norm": 0.24796300093531085,
+      "learning_rate": 8.26150421353682e-07,
+      "loss": 0.2073,
+      "step": 1133
+    },
+    {
+      "epoch": 4.5910931174089065,
+      "grad_norm": 0.2573833824516464,
+      "learning_rate": 8.101405277100549e-07,
+      "loss": 0.2213,
+      "step": 1134
+    },
+    {
+      "epoch": 4.5951417004048585,
+      "grad_norm": 0.24834406931871067,
+      "learning_rate": 7.942840756714077e-07,
+      "loss": 0.1889,
+      "step": 1135
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.23695868190385697,
+      "learning_rate": 7.785811920255093e-07,
+      "loss": 0.2433,
+      "step": 1136
+    },
+    {
+      "epoch": 4.603238866396762,
+      "grad_norm": 0.2541718640835249,
+      "learning_rate": 7.630320023322024e-07,
+      "loss": 0.2035,
+      "step": 1137
+    },
+    {
+      "epoch": 4.607287449392713,
+      "grad_norm": 0.26217991492598497,
+      "learning_rate": 7.476366309223903e-07,
+      "loss": 0.2177,
+      "step": 1138
+    },
+    {
+      "epoch": 4.611336032388664,
+      "grad_norm": 0.24560238964137407,
+      "learning_rate": 7.323952008970537e-07,
+      "loss": 0.2151,
+      "step": 1139
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.2529537541212619,
+      "learning_rate": 7.173078341262618e-07,
+      "loss": 0.212,
+      "step": 1140
+    },
+    {
+      "epoch": 4.619433198380567,
+      "grad_norm": 0.25558919866417307,
+      "learning_rate": 7.023746512481989e-07,
+      "loss": 0.2183,
+      "step": 1141
+    },
+    {
+      "epoch": 4.623481781376518,
+      "grad_norm": 0.23580200079808025,
+      "learning_rate": 6.875957716681902e-07,
+      "loss": 0.2205,
+      "step": 1142
+    },
+    {
+      "epoch": 4.62753036437247,
+      "grad_norm": 0.2422120947179597,
+      "learning_rate": 6.729713135577687e-07,
+      "loss": 0.2138,
+      "step": 1143
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.24545782677477335,
+      "learning_rate": 6.585013938537143e-07,
+      "loss": 0.2129,
+      "step": 1144
+    },
+    {
+      "epoch": 4.635627530364372,
+      "grad_norm": 0.2397944752314604,
+      "learning_rate": 6.441861282571115e-07,
+      "loss": 0.1859,
+      "step": 1145
+    },
+    {
+      "epoch": 4.6396761133603235,
+      "grad_norm": 0.25375074441170986,
+      "learning_rate": 6.300256312324471e-07,
+      "loss": 0.2265,
+      "step": 1146
+    },
+    {
+      "epoch": 4.6437246963562755,
+      "grad_norm": 0.24433880873349043,
+      "learning_rate": 6.160200160066776e-07,
+      "loss": 0.2084,
+      "step": 1147
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.2542344573062365,
+      "learning_rate": 6.021693945683327e-07,
+      "loss": 0.234,
+      "step": 1148
+    },
+    {
+      "epoch": 4.651821862348179,
+      "grad_norm": 0.24852431363428099,
+      "learning_rate": 5.884738776666044e-07,
+      "loss": 0.2159,
+      "step": 1149
+    },
+    {
+      "epoch": 4.65587044534413,
+      "grad_norm": 0.24824198501893646,
+      "learning_rate": 5.749335748104878e-07,
+      "loss": 0.2075,
+      "step": 1150
+    },
+    {
+      "epoch": 4.659919028340081,
+      "grad_norm": 0.24785886273504942,
+      "learning_rate": 5.615485942678778e-07,
+      "loss": 0.201,
+      "step": 1151
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.23119032400631487,
+      "learning_rate": 5.483190430647289e-07,
+      "loss": 0.2164,
+      "step": 1152
+    },
+    {
+      "epoch": 4.668016194331984,
+      "grad_norm": 0.25427187855134853,
+      "learning_rate": 5.352450269841747e-07,
+      "loss": 0.1978,
+      "step": 1153
+    },
+    {
+      "epoch": 4.672064777327935,
+      "grad_norm": 0.23624209307117897,
+      "learning_rate": 5.22326650565701e-07,
+      "loss": 0.2024,
+      "step": 1154
+    },
+    {
+      "epoch": 4.676113360323887,
+      "grad_norm": 0.24603949267366537,
+      "learning_rate": 5.095640171043048e-07,
+      "loss": 0.1967,
+      "step": 1155
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.25744728128725586,
+      "learning_rate": 4.969572286496594e-07,
+      "loss": 0.2193,
+      "step": 1156
+    },
+    {
+      "epoch": 4.684210526315789,
+      "grad_norm": 0.24408453955832424,
+      "learning_rate": 4.845063860053079e-07,
+      "loss": 0.209,
+      "step": 1157
+    },
+    {
+      "epoch": 4.6882591093117405,
+      "grad_norm": 0.24651431357940723,
+      "learning_rate": 4.7221158872785997e-07,
+      "loss": 0.2265,
+      "step": 1158
+    },
+    {
+      "epoch": 4.6923076923076925,
+      "grad_norm": 0.23919158355028633,
+      "learning_rate": 4.600729351261857e-07,
+      "loss": 0.2107,
+      "step": 1159
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.24743336858034234,
+      "learning_rate": 4.4809052226063355e-07,
+      "loss": 0.1926,
+      "step": 1160
+    },
+    {
+      "epoch": 4.700404858299595,
+      "grad_norm": 0.23910390291849493,
+      "learning_rate": 4.3626444594226046e-07,
+      "loss": 0.2419,
+      "step": 1161
+    },
+    {
+      "epoch": 4.704453441295547,
+      "grad_norm": 0.23529972082401027,
+      "learning_rate": 4.245948007320588e-07,
+      "loss": 0.1904,
+      "step": 1162
+    },
+    {
+      "epoch": 4.708502024291498,
+      "grad_norm": 0.23752724271135633,
+      "learning_rate": 4.130816799401971e-07,
+      "loss": 0.1967,
+      "step": 1163
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.24776108165203245,
+      "learning_rate": 4.017251756252827e-07,
+      "loss": 0.1906,
+      "step": 1164
+    },
+    {
+      "epoch": 4.716599190283401,
+      "grad_norm": 0.23672036093914686,
+      "learning_rate": 3.9052537859362473e-07,
+      "loss": 0.1898,
+      "step": 1165
+    },
+    {
+      "epoch": 4.720647773279352,
+      "grad_norm": 0.24555801708677344,
+      "learning_rate": 3.7948237839849687e-07,
+      "loss": 0.2243,
+      "step": 1166
+    },
+    {
+      "epoch": 4.724696356275303,
+      "grad_norm": 0.23800822868094296,
+      "learning_rate": 3.6859626333943797e-07,
+      "loss": 0.2189,
+      "step": 1167
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.25765188272121536,
+      "learning_rate": 3.578671204615347e-07,
+      "loss": 0.1919,
+      "step": 1168
+    },
+    {
+      "epoch": 4.732793522267206,
+      "grad_norm": 0.24548458802302495,
+      "learning_rate": 3.472950355547267e-07,
+      "loss": 0.2088,
+      "step": 1169
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 0.24233269617122613,
+      "learning_rate": 3.3688009315312244e-07,
+      "loss": 0.2175,
+      "step": 1170
+    },
+    {
+      "epoch": 4.7408906882591095,
+      "grad_norm": 0.2372801403735897,
+      "learning_rate": 3.266223765343224e-07,
+      "loss": 0.2035,
+      "step": 1171
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.24438148882158625,
+      "learning_rate": 3.16521967718757e-07,
+      "loss": 0.2254,
+      "step": 1172
+    },
+    {
+      "epoch": 4.748987854251012,
+      "grad_norm": 0.24856441366287568,
+      "learning_rate": 3.0657894746902726e-07,
+      "loss": 0.1887,
+      "step": 1173
+    },
+    {
+      "epoch": 4.753036437246964,
+      "grad_norm": 0.23801520039009308,
+      "learning_rate": 2.9679339528925654e-07,
+      "loss": 0.2054,
+      "step": 1174
+    },
+    {
+      "epoch": 4.757085020242915,
+      "grad_norm": 0.2334232943019928,
+      "learning_rate": 2.871653894244597e-07,
+      "loss": 0.2011,
+      "step": 1175
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.24113804554725088,
+      "learning_rate": 2.7769500685991266e-07,
+      "loss": 0.2464,
+      "step": 1176
+    },
+    {
+      "epoch": 4.765182186234818,
+      "grad_norm": 0.24137399160578385,
+      "learning_rate": 2.683823233205485e-07,
+      "loss": 0.2007,
+      "step": 1177
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.2368371163426128,
+      "learning_rate": 2.592274132703287e-07,
+      "loss": 0.2182,
+      "step": 1178
+    },
+    {
+      "epoch": 4.77327935222672,
+      "grad_norm": 0.24642960045511825,
+      "learning_rate": 2.502303499116754e-07,
+      "loss": 0.1941,
+      "step": 1179
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.23948377574923074,
+      "learning_rate": 2.413912051848666e-07,
+      "loss": 0.2119,
+      "step": 1180
+    },
+    {
+      "epoch": 4.781376518218623,
+      "grad_norm": 0.23827383939730806,
+      "learning_rate": 2.3271004976746613e-07,
+      "loss": 0.2328,
+      "step": 1181
+    },
+    {
+      "epoch": 4.7854251012145745,
+      "grad_norm": 0.24143514843412187,
+      "learning_rate": 2.2418695307376393e-07,
+      "loss": 0.2046,
+      "step": 1182
+    },
+    {
+      "epoch": 4.7894736842105265,
+      "grad_norm": 0.2471547578203054,
+      "learning_rate": 2.1582198325420968e-07,
+      "loss": 0.2087,
+      "step": 1183
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.23096577210382682,
+      "learning_rate": 2.0761520719488447e-07,
+      "loss": 0.1851,
+      "step": 1184
+    },
+    {
+      "epoch": 4.797570850202429,
+      "grad_norm": 0.23572887443809196,
+      "learning_rate": 1.9956669051694798e-07,
+      "loss": 0.207,
+      "step": 1185
+    },
+    {
+      "epoch": 4.801619433198381,
+      "grad_norm": 0.24699613303585488,
+      "learning_rate": 1.916764975761254e-07,
+      "loss": 0.2021,
+      "step": 1186
+    },
+    {
+      "epoch": 4.805668016194332,
+      "grad_norm": 0.2366948891875096,
+      "learning_rate": 1.839446914621923e-07,
+      "loss": 0.222,
+      "step": 1187
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.24745369003553983,
+      "learning_rate": 1.7637133399845962e-07,
+      "loss": 0.2134,
+      "step": 1188
+    },
+    {
+      "epoch": 4.813765182186235,
+      "grad_norm": 0.24473859522047345,
+      "learning_rate": 1.6895648574129397e-07,
+      "loss": 0.2077,
+      "step": 1189
+    },
+    {
+      "epoch": 4.817813765182186,
+      "grad_norm": 0.247704679271935,
+      "learning_rate": 1.6170020597962465e-07,
+      "loss": 0.2317,
+      "step": 1190
+    },
+    {
+      "epoch": 4.821862348178137,
+      "grad_norm": 0.24187240407214383,
+      "learning_rate": 1.5460255273447078e-07,
+      "loss": 0.2157,
+      "step": 1191
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.2347151134823873,
+      "learning_rate": 1.4766358275847715e-07,
+      "loss": 0.2244,
+      "step": 1192
+    },
+    {
+      "epoch": 4.82995951417004,
+      "grad_norm": 0.23474832917194383,
+      "learning_rate": 1.4088335153546573e-07,
+      "loss": 0.2099,
+      "step": 1193
+    },
+    {
+      "epoch": 4.834008097165992,
+      "grad_norm": 0.24196033991647878,
+      "learning_rate": 1.3426191327998496e-07,
+      "loss": 0.1984,
+      "step": 1194
+    },
+    {
+      "epoch": 4.838056680161944,
+      "grad_norm": 0.2485741754004044,
+      "learning_rate": 1.277993209368744e-07,
+      "loss": 0.2107,
+      "step": 1195
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.2449771206555142,
+      "learning_rate": 1.214956261808542e-07,
+      "loss": 0.2201,
+      "step": 1196
+    },
+    {
+      "epoch": 4.846153846153846,
+      "grad_norm": 0.24067554041792838,
+      "learning_rate": 1.1535087941609846e-07,
+      "loss": 0.2048,
+      "step": 1197
+    },
+    {
+      "epoch": 4.850202429149798,
+      "grad_norm": 0.25041115949296006,
+      "learning_rate": 1.0936512977583802e-07,
+      "loss": 0.192,
+      "step": 1198
+    },
+    {
+      "epoch": 4.854251012145749,
+      "grad_norm": 0.24509114615493183,
+      "learning_rate": 1.0353842512196732e-07,
+      "loss": 0.2236,
+      "step": 1199
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.24286214984520196,
+      "learning_rate": 9.787081204466253e-08,
+      "loss": 0.1944,
+      "step": 1200
+    },
+    {
+      "epoch": 4.862348178137652,
+      "grad_norm": 0.23246947896121145,
+      "learning_rate": 9.236233586200627e-08,
+      "loss": 0.1884,
+      "step": 1201
+    },
+    {
+      "epoch": 4.866396761133603,
+      "grad_norm": 0.24961418736841479,
+      "learning_rate": 8.70130406196279e-08,
+      "loss": 0.2187,
+      "step": 1202
+    },
+    {
+      "epoch": 4.870445344129554,
+      "grad_norm": 0.24302417895052694,
+      "learning_rate": 8.182296909035047e-08,
+      "loss": 0.2002,
+      "step": 1203
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.23989776895486356,
+      "learning_rate": 7.679216277384439e-08,
+      "loss": 0.2144,
+      "step": 1204
+    },
+    {
+      "epoch": 4.8785425101214575,
+      "grad_norm": 0.23230052999907422,
+      "learning_rate": 7.19206618963053e-08,
+      "loss": 0.1963,
+      "step": 1205
+    },
+    {
+      "epoch": 4.882591093117409,
+      "grad_norm": 0.24353358066899078,
+      "learning_rate": 6.720850541012347e-08,
+      "loss": 0.2102,
+      "step": 1206
+    },
+    {
+      "epoch": 4.886639676113361,
+      "grad_norm": 0.2305512504659059,
+      "learning_rate": 6.265573099357714e-08,
+      "loss": 0.1957,
+      "step": 1207
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.24086443861081058,
+      "learning_rate": 5.826237505053067e-08,
+      "loss": 0.2125,
+      "step": 1208
+    },
+    {
+      "epoch": 4.894736842105263,
+      "grad_norm": 0.23862337322278301,
+      "learning_rate": 5.4028472710137e-08,
+      "loss": 0.2088,
+      "step": 1209
+    },
+    {
+      "epoch": 4.898785425101215,
+      "grad_norm": 0.24367452304903192,
+      "learning_rate": 4.9954057826571146e-08,
+      "loss": 0.2212,
+      "step": 1210
+    },
+    {
+      "epoch": 4.902834008097166,
+      "grad_norm": 0.24645444424552743,
+      "learning_rate": 4.6039162978743776e-08,
+      "loss": 0.2129,
+      "step": 1211
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.24085860423001124,
+      "learning_rate": 4.228381947004812e-08,
+      "loss": 0.2073,
+      "step": 1212
+    },
+    {
+      "epoch": 4.910931174089069,
+      "grad_norm": 0.239178077441481,
+      "learning_rate": 3.868805732811343e-08,
+      "loss": 0.2031,
+      "step": 1213
+    },
+    {
+      "epoch": 4.91497975708502,
+      "grad_norm": 0.23375607229789905,
+      "learning_rate": 3.525190530455636e-08,
+      "loss": 0.1927,
+      "step": 1214
+    },
+    {
+      "epoch": 4.919028340080971,
+      "grad_norm": 0.2358984661199372,
+      "learning_rate": 3.197539087475887e-08,
+      "loss": 0.2323,
+      "step": 1215
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.2422720560382151,
+      "learning_rate": 2.8858540237641786e-08,
+      "loss": 0.2079,
+      "step": 1216
+    },
+    {
+      "epoch": 4.9271255060728745,
+      "grad_norm": 0.24227807454660544,
+      "learning_rate": 2.5901378315458247e-08,
+      "loss": 0.2144,
+      "step": 1217
+    },
+    {
+      "epoch": 4.931174089068826,
+      "grad_norm": 0.257971349340192,
+      "learning_rate": 2.3103928753598347e-08,
+      "loss": 0.2207,
+      "step": 1218
+    },
+    {
+      "epoch": 4.935222672064778,
+      "grad_norm": 0.23769957640678488,
+      "learning_rate": 2.0466213920393718e-08,
+      "loss": 0.1969,
+      "step": 1219
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.24153234586458852,
+      "learning_rate": 1.7988254906942115e-08,
+      "loss": 0.1909,
+      "step": 1220
+    },
+    {
+      "epoch": 4.94331983805668,
+      "grad_norm": 0.2477259119106197,
+      "learning_rate": 1.5670071526936447e-08,
+      "loss": 0.2088,
+      "step": 1221
+    },
+    {
+      "epoch": 4.947368421052632,
+      "grad_norm": 0.2406368095717694,
+      "learning_rate": 1.3511682316509344e-08,
+      "loss": 0.2052,
+      "step": 1222
+    },
+    {
+      "epoch": 4.951417004048583,
+      "grad_norm": 0.2585558929179502,
+      "learning_rate": 1.1513104534086605e-08,
+      "loss": 0.1877,
+      "step": 1223
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.23194657330631072,
+      "learning_rate": 9.67435416023843e-09,
+      "loss": 0.2143,
+      "step": 1224
+    },
+    {
+      "epoch": 4.959514170040486,
+      "grad_norm": 0.24216315365239396,
+      "learning_rate": 7.995445897566178e-09,
+      "loss": 0.1952,
+      "step": 1225
+    },
+    {
+      "epoch": 4.963562753036437,
+      "grad_norm": 0.23567925418691207,
+      "learning_rate": 6.476393170578021e-09,
+      "loss": 0.2097,
+      "step": 1226
+    },
+    {
+      "epoch": 4.967611336032388,
+      "grad_norm": 0.23835817978215615,
+      "learning_rate": 5.11720812557348e-09,
+      "loss": 0.2194,
+      "step": 1227
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.2423587863800213,
+      "learning_rate": 3.917901630563492e-09,
+      "loss": 0.1967,
+      "step": 1228
+    },
+    {
+      "epoch": 4.9757085020242915,
+      "grad_norm": 0.2475658155838076,
+      "learning_rate": 2.878483275170485e-09,
+      "loss": 0.1873,
+      "step": 1229
+    },
+    {
+      "epoch": 4.979757085020243,
+      "grad_norm": 0.24226447938501128,
+      "learning_rate": 1.998961370555108e-09,
+      "loss": 0.202,
+      "step": 1230
+    },
+    {
+      "epoch": 4.983805668016195,
+      "grad_norm": 0.2407508306922544,
+      "learning_rate": 1.2793429493518361e-09,
+      "loss": 0.2159,
+      "step": 1231
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.24150757063145392,
+      "learning_rate": 7.196337656112384e-10,
+      "loss": 0.2327,
+      "step": 1232
+    },
+    {
+      "epoch": 4.991902834008097,
+      "grad_norm": 0.24631723781889686,
+      "learning_rate": 3.198382947511292e-10,
+      "loss": 0.2183,
+      "step": 1233
+    },
+    {
+      "epoch": 4.995951417004049,
+      "grad_norm": 0.24339887854482692,
+      "learning_rate": 7.995973352770137e-11,
+      "loss": 0.1983,
+      "step": 1234
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.2345544648635956,
+      "learning_rate": 0.0,
+      "loss": 0.2007,
+      "step": 1235
+    },
+    {
+      "epoch": 5.0,
+      "step": 1235,
+      "total_flos": 2.654408818309464e+18,
+      "train_loss": 0.4213724943428387,
+      "train_runtime": 25490.9493,
+      "train_samples_per_second": 6.198,
+      "train_steps_per_second": 0.048
+    }
+  ],
+  "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": 2.654408818309464e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}