{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 1516,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0013192612137203166,
      "grad_norm": 12.233341930238273,
      "learning_rate": 6.578947368421053e-08,
      "loss": 0.3407,
      "step": 1
    },
    {
      "epoch": 0.002638522427440633,
      "grad_norm": 12.816762328416463,
      "learning_rate": 1.3157894736842107e-07,
      "loss": 0.3118,
      "step": 2
    },
    {
      "epoch": 0.00395778364116095,
      "grad_norm": 12.015688553148182,
      "learning_rate": 1.9736842105263157e-07,
      "loss": 0.3131,
      "step": 3
    },
    {
      "epoch": 0.005277044854881266,
      "grad_norm": 13.05850661637904,
      "learning_rate": 2.6315789473684213e-07,
      "loss": 0.3361,
      "step": 4
    },
    {
      "epoch": 0.006596306068601583,
      "grad_norm": 11.88542762458142,
      "learning_rate": 3.2894736842105264e-07,
      "loss": 0.332,
      "step": 5
    },
    {
      "epoch": 0.0079155672823219,
      "grad_norm": 13.124966565305824,
      "learning_rate": 3.9473684210526315e-07,
      "loss": 0.328,
      "step": 6
    },
    {
      "epoch": 0.009234828496042216,
      "grad_norm": 12.075469159428398,
      "learning_rate": 4.605263157894737e-07,
      "loss": 0.3071,
      "step": 7
    },
    {
      "epoch": 0.010554089709762533,
      "grad_norm": 12.356648628552895,
      "learning_rate": 5.263157894736843e-07,
      "loss": 0.3072,
      "step": 8
    },
    {
      "epoch": 0.011873350923482849,
      "grad_norm": 10.352882639373675,
      "learning_rate": 5.921052631578947e-07,
      "loss": 0.261,
      "step": 9
    },
    {
      "epoch": 0.013192612137203167,
      "grad_norm": 10.058615814168975,
      "learning_rate": 6.578947368421053e-07,
      "loss": 0.253,
      "step": 10
    },
    {
      "epoch": 0.014511873350923483,
      "grad_norm": 8.63214503438111,
      "learning_rate": 7.236842105263158e-07,
      "loss": 0.2593,
      "step": 11
    },
    {
      "epoch": 0.0158311345646438,
      "grad_norm": 7.870088141405597,
      "learning_rate": 7.894736842105263e-07,
      "loss": 0.2304,
      "step": 12
    },
    {
      "epoch": 0.017150395778364115,
      "grad_norm": 5.393787971545889,
      "learning_rate": 8.55263157894737e-07,
      "loss": 0.1915,
      "step": 13
    },
    {
      "epoch": 0.018469656992084433,
      "grad_norm": 5.535158104456134,
      "learning_rate": 9.210526315789474e-07,
      "loss": 0.206,
      "step": 14
    },
    {
      "epoch": 0.01978891820580475,
      "grad_norm": 5.476677916959981,
      "learning_rate": 9.86842105263158e-07,
      "loss": 0.236,
      "step": 15
    },
    {
      "epoch": 0.021108179419525065,
      "grad_norm": 4.151444478521054,
      "learning_rate": 1.0526315789473685e-06,
      "loss": 0.1496,
      "step": 16
    },
    {
      "epoch": 0.022427440633245383,
      "grad_norm": 3.0628867506650748,
      "learning_rate": 1.118421052631579e-06,
      "loss": 0.2017,
      "step": 17
    },
    {
      "epoch": 0.023746701846965697,
      "grad_norm": 3.036924011607267,
      "learning_rate": 1.1842105263157894e-06,
      "loss": 0.1449,
      "step": 18
    },
    {
      "epoch": 0.025065963060686015,
      "grad_norm": 2.4996954384635646,
      "learning_rate": 1.25e-06,
      "loss": 0.134,
      "step": 19
    },
    {
      "epoch": 0.026385224274406333,
      "grad_norm": 2.11406547625844,
      "learning_rate": 1.3157894736842106e-06,
      "loss": 0.1119,
      "step": 20
    },
    {
      "epoch": 0.027704485488126648,
      "grad_norm": 1.9596253186522326,
      "learning_rate": 1.3815789473684212e-06,
      "loss": 0.1629,
      "step": 21
    },
    {
      "epoch": 0.029023746701846966,
      "grad_norm": 1.7753792784753228,
      "learning_rate": 1.4473684210526317e-06,
      "loss": 0.139,
      "step": 22
    },
    {
      "epoch": 0.030343007915567283,
      "grad_norm": 2.43386121958396,
      "learning_rate": 1.5131578947368421e-06,
      "loss": 0.1343,
      "step": 23
    },
    {
      "epoch": 0.0316622691292876,
      "grad_norm": 1.8993087578136192,
      "learning_rate": 1.5789473684210526e-06,
      "loss": 0.1151,
      "step": 24
    },
    {
      "epoch": 0.032981530343007916,
      "grad_norm": 3.0583374294765853,
      "learning_rate": 1.6447368421052635e-06,
      "loss": 0.154,
      "step": 25
    },
    {
      "epoch": 0.03430079155672823,
      "grad_norm": 2.253460947598232,
      "learning_rate": 1.710526315789474e-06,
      "loss": 0.142,
      "step": 26
    },
    {
      "epoch": 0.03562005277044855,
      "grad_norm": 2.016611774580363,
      "learning_rate": 1.7763157894736844e-06,
      "loss": 0.125,
      "step": 27
    },
    {
      "epoch": 0.036939313984168866,
      "grad_norm": 2.017120787608575,
      "learning_rate": 1.8421052631578948e-06,
      "loss": 0.1567,
      "step": 28
    },
    {
      "epoch": 0.03825857519788918,
      "grad_norm": 2.1936744617942034,
      "learning_rate": 1.9078947368421057e-06,
      "loss": 0.1257,
      "step": 29
    },
    {
      "epoch": 0.0395778364116095,
      "grad_norm": 1.9830690908180522,
      "learning_rate": 1.973684210526316e-06,
      "loss": 0.1331,
      "step": 30
    },
    {
      "epoch": 0.040897097625329816,
      "grad_norm": 1.5631779172937716,
      "learning_rate": 2.0394736842105266e-06,
      "loss": 0.1178,
      "step": 31
    },
    {
      "epoch": 0.04221635883905013,
      "grad_norm": 1.4861255029401097,
      "learning_rate": 2.105263157894737e-06,
      "loss": 0.0911,
      "step": 32
    },
    {
      "epoch": 0.04353562005277045,
      "grad_norm": 2.4000033930532956,
      "learning_rate": 2.1710526315789475e-06,
      "loss": 0.1316,
      "step": 33
    },
    {
      "epoch": 0.044854881266490766,
      "grad_norm": 1.6157534765776684,
      "learning_rate": 2.236842105263158e-06,
      "loss": 0.0996,
      "step": 34
    },
    {
      "epoch": 0.04617414248021108,
      "grad_norm": 1.6467470674972677,
      "learning_rate": 2.3026315789473684e-06,
      "loss": 0.0848,
      "step": 35
    },
    {
      "epoch": 0.047493403693931395,
      "grad_norm": 1.4693385869948326,
      "learning_rate": 2.368421052631579e-06,
      "loss": 0.111,
      "step": 36
    },
    {
      "epoch": 0.048812664907651716,
      "grad_norm": 2.300813961627334,
      "learning_rate": 2.4342105263157898e-06,
      "loss": 0.1299,
      "step": 37
    },
    {
      "epoch": 0.05013192612137203,
      "grad_norm": 1.6680540076451675,
      "learning_rate": 2.5e-06,
      "loss": 0.0996,
      "step": 38
    },
    {
      "epoch": 0.051451187335092345,
      "grad_norm": 2.314118463144595,
      "learning_rate": 2.565789473684211e-06,
      "loss": 0.1161,
      "step": 39
    },
    {
      "epoch": 0.052770448548812667,
      "grad_norm": 2.6523025026316795,
      "learning_rate": 2.631578947368421e-06,
      "loss": 0.1378,
      "step": 40
    },
    {
      "epoch": 0.05408970976253298,
      "grad_norm": 2.3663275040840426,
      "learning_rate": 2.697368421052632e-06,
      "loss": 0.1456,
      "step": 41
    },
    {
      "epoch": 0.055408970976253295,
      "grad_norm": 1.4303633951040533,
      "learning_rate": 2.7631578947368424e-06,
      "loss": 0.0858,
      "step": 42
    },
    {
      "epoch": 0.05672823218997362,
      "grad_norm": 1.000990164023815,
      "learning_rate": 2.828947368421053e-06,
      "loss": 0.0823,
      "step": 43
    },
    {
      "epoch": 0.05804749340369393,
      "grad_norm": 1.5872577094636717,
      "learning_rate": 2.8947368421052634e-06,
      "loss": 0.0958,
      "step": 44
    },
    {
      "epoch": 0.059366754617414245,
      "grad_norm": 1.7841530289241443,
      "learning_rate": 2.960526315789474e-06,
      "loss": 0.1239,
      "step": 45
    },
    {
      "epoch": 0.06068601583113457,
      "grad_norm": 1.1766620869883984,
      "learning_rate": 3.0263157894736843e-06,
      "loss": 0.0697,
      "step": 46
    },
    {
      "epoch": 0.06200527704485488,
      "grad_norm": 1.2853170690028022,
      "learning_rate": 3.092105263157895e-06,
      "loss": 0.0963,
      "step": 47
    },
    {
      "epoch": 0.0633245382585752,
      "grad_norm": 1.4709634028779943,
      "learning_rate": 3.157894736842105e-06,
      "loss": 0.0915,
      "step": 48
    },
    {
      "epoch": 0.06464379947229551,
      "grad_norm": 1.3745315046203157,
      "learning_rate": 3.223684210526316e-06,
      "loss": 0.0799,
      "step": 49
    },
    {
      "epoch": 0.06596306068601583,
      "grad_norm": 1.4761871536724172,
      "learning_rate": 3.289473684210527e-06,
      "loss": 0.1046,
      "step": 50
    },
    {
      "epoch": 0.06728232189973615,
      "grad_norm": 1.5105153887101839,
      "learning_rate": 3.355263157894737e-06,
      "loss": 0.0853,
      "step": 51
    },
    {
      "epoch": 0.06860158311345646,
      "grad_norm": 1.3535267919848646,
      "learning_rate": 3.421052631578948e-06,
      "loss": 0.1319,
      "step": 52
    },
    {
      "epoch": 0.06992084432717678,
      "grad_norm": 1.6687947303259212,
      "learning_rate": 3.486842105263158e-06,
      "loss": 0.1055,
      "step": 53
    },
    {
      "epoch": 0.0712401055408971,
      "grad_norm": 3.6501802247402027,
      "learning_rate": 3.5526315789473687e-06,
      "loss": 0.1094,
      "step": 54
    },
    {
      "epoch": 0.07255936675461741,
      "grad_norm": 1.3210848086773013,
      "learning_rate": 3.618421052631579e-06,
      "loss": 0.1088,
      "step": 55
    },
    {
      "epoch": 0.07387862796833773,
      "grad_norm": 1.0101400792927315,
      "learning_rate": 3.6842105263157896e-06,
      "loss": 0.0763,
      "step": 56
    },
    {
      "epoch": 0.07519788918205805,
      "grad_norm": 1.3009728579134852,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.0922,
      "step": 57
    },
    {
      "epoch": 0.07651715039577836,
      "grad_norm": 1.5671514737535006,
      "learning_rate": 3.815789473684211e-06,
      "loss": 0.0967,
      "step": 58
    },
    {
      "epoch": 0.07783641160949868,
      "grad_norm": 1.0883160796639246,
      "learning_rate": 3.8815789473684214e-06,
      "loss": 0.1078,
      "step": 59
    },
    {
      "epoch": 0.079155672823219,
      "grad_norm": 2.311711212772536,
      "learning_rate": 3.947368421052632e-06,
      "loss": 0.0964,
      "step": 60
    },
    {
      "epoch": 0.08047493403693931,
      "grad_norm": 1.6181560133721102,
      "learning_rate": 4.013157894736842e-06,
      "loss": 0.093,
      "step": 61
    },
    {
      "epoch": 0.08179419525065963,
      "grad_norm": 1.4364617838539906,
      "learning_rate": 4.078947368421053e-06,
      "loss": 0.0803,
      "step": 62
    },
    {
      "epoch": 0.08311345646437995,
      "grad_norm": 1.49841077642279,
      "learning_rate": 4.144736842105263e-06,
      "loss": 0.105,
      "step": 63
    },
    {
      "epoch": 0.08443271767810026,
      "grad_norm": 1.449891300704642,
      "learning_rate": 4.210526315789474e-06,
      "loss": 0.0702,
      "step": 64
    },
    {
      "epoch": 0.08575197889182058,
      "grad_norm": 1.8407494276748049,
      "learning_rate": 4.276315789473684e-06,
      "loss": 0.1237,
      "step": 65
    },
    {
      "epoch": 0.0870712401055409,
      "grad_norm": 1.4445786143372672,
      "learning_rate": 4.342105263157895e-06,
      "loss": 0.0861,
      "step": 66
    },
    {
      "epoch": 0.08839050131926121,
      "grad_norm": 1.5820793398923942,
      "learning_rate": 4.407894736842105e-06,
      "loss": 0.1063,
      "step": 67
    },
    {
      "epoch": 0.08970976253298153,
      "grad_norm": 1.6266178158804763,
      "learning_rate": 4.473684210526316e-06,
      "loss": 0.1281,
      "step": 68
    },
    {
      "epoch": 0.09102902374670185,
      "grad_norm": 1.921098471296673,
      "learning_rate": 4.539473684210527e-06,
      "loss": 0.1121,
      "step": 69
    },
    {
      "epoch": 0.09234828496042216,
      "grad_norm": 1.3652318289138008,
      "learning_rate": 4.605263157894737e-06,
      "loss": 0.1167,
      "step": 70
    },
    {
      "epoch": 0.09366754617414248,
      "grad_norm": 1.9511646011862807,
      "learning_rate": 4.671052631578948e-06,
      "loss": 0.0977,
      "step": 71
    },
    {
      "epoch": 0.09498680738786279,
      "grad_norm": 1.3555845548236822,
      "learning_rate": 4.736842105263158e-06,
      "loss": 0.0911,
      "step": 72
    },
    {
      "epoch": 0.09630606860158311,
      "grad_norm": 2.2259057577872445,
      "learning_rate": 4.802631578947369e-06,
      "loss": 0.1279,
      "step": 73
    },
    {
      "epoch": 0.09762532981530343,
      "grad_norm": 1.3888531130344683,
      "learning_rate": 4.8684210526315795e-06,
      "loss": 0.0784,
      "step": 74
    },
    {
      "epoch": 0.09894459102902374,
      "grad_norm": 1.3960319673373207,
      "learning_rate": 4.9342105263157895e-06,
      "loss": 0.0978,
      "step": 75
    },
    {
      "epoch": 0.10026385224274406,
      "grad_norm": 1.7054291543895856,
      "learning_rate": 5e-06,
      "loss": 0.0984,
      "step": 76
    },
    {
      "epoch": 0.10158311345646438,
      "grad_norm": 1.6458987247042498,
      "learning_rate": 5.0657894736842104e-06,
      "loss": 0.0999,
      "step": 77
    },
    {
      "epoch": 0.10290237467018469,
      "grad_norm": 1.6822123202405193,
      "learning_rate": 5.131578947368422e-06,
      "loss": 0.0922,
      "step": 78
    },
    {
      "epoch": 0.10422163588390501,
      "grad_norm": 1.5932180454716234,
      "learning_rate": 5.197368421052632e-06,
      "loss": 0.1223,
      "step": 79
    },
    {
      "epoch": 0.10554089709762533,
      "grad_norm": 1.8249077406901284,
      "learning_rate": 5.263157894736842e-06,
      "loss": 0.0975,
      "step": 80
    },
    {
      "epoch": 0.10686015831134564,
      "grad_norm": 1.6365604083770662,
      "learning_rate": 5.328947368421054e-06,
      "loss": 0.129,
      "step": 81
    },
    {
      "epoch": 0.10817941952506596,
      "grad_norm": 1.9363957711801008,
      "learning_rate": 5.394736842105264e-06,
      "loss": 0.1022,
      "step": 82
    },
    {
      "epoch": 0.10949868073878628,
      "grad_norm": 1.1568619049418536,
      "learning_rate": 5.460526315789474e-06,
      "loss": 0.0687,
      "step": 83
    },
    {
      "epoch": 0.11081794195250659,
      "grad_norm": 1.5128549314706825,
      "learning_rate": 5.526315789473685e-06,
      "loss": 0.1096,
      "step": 84
    },
    {
      "epoch": 0.11213720316622691,
      "grad_norm": 1.416923068463139,
      "learning_rate": 5.592105263157896e-06,
      "loss": 0.1052,
      "step": 85
    },
    {
      "epoch": 0.11345646437994723,
      "grad_norm": 1.423800222860217,
      "learning_rate": 5.657894736842106e-06,
      "loss": 0.0986,
      "step": 86
    },
    {
      "epoch": 0.11477572559366754,
      "grad_norm": 1.4869568465395704,
      "learning_rate": 5.723684210526316e-06,
      "loss": 0.0906,
      "step": 87
    },
    {
      "epoch": 0.11609498680738786,
      "grad_norm": 1.302701356412605,
      "learning_rate": 5.789473684210527e-06,
      "loss": 0.09,
      "step": 88
    },
    {
      "epoch": 0.11741424802110818,
      "grad_norm": 1.2568159914563075,
      "learning_rate": 5.855263157894738e-06,
      "loss": 0.0988,
      "step": 89
    },
    {
      "epoch": 0.11873350923482849,
      "grad_norm": 1.1819489308351352,
      "learning_rate": 5.921052631578948e-06,
      "loss": 0.0816,
      "step": 90
    },
    {
      "epoch": 0.12005277044854881,
      "grad_norm": 0.9550864379939853,
      "learning_rate": 5.9868421052631585e-06,
      "loss": 0.073,
      "step": 91
    },
    {
      "epoch": 0.12137203166226913,
      "grad_norm": 1.4611228215909369,
      "learning_rate": 6.0526315789473685e-06,
      "loss": 0.0889,
      "step": 92
    },
    {
      "epoch": 0.12269129287598944,
      "grad_norm": 1.0971390981785767,
      "learning_rate": 6.118421052631579e-06,
      "loss": 0.1041,
      "step": 93
    },
    {
      "epoch": 0.12401055408970976,
      "grad_norm": 1.364124444234249,
      "learning_rate": 6.18421052631579e-06,
      "loss": 0.0837,
      "step": 94
    },
    {
      "epoch": 0.12532981530343007,
      "grad_norm": 1.5344985712619694,
      "learning_rate": 6.25e-06,
      "loss": 0.0793,
      "step": 95
    },
    {
      "epoch": 0.1266490765171504,
      "grad_norm": 1.1050557737138007,
      "learning_rate": 6.31578947368421e-06,
      "loss": 0.1012,
      "step": 96
    },
    {
      "epoch": 0.1279683377308707,
      "grad_norm": 1.6313639186233246,
      "learning_rate": 6.381578947368422e-06,
      "loss": 0.0725,
      "step": 97
    },
    {
      "epoch": 0.12928759894459102,
      "grad_norm": 1.0300085400623038,
      "learning_rate": 6.447368421052632e-06,
      "loss": 0.0874,
      "step": 98
    },
    {
      "epoch": 0.13060686015831136,
      "grad_norm": 1.4468801863848761,
      "learning_rate": 6.513157894736842e-06,
      "loss": 0.077,
      "step": 99
    },
    {
      "epoch": 0.13192612137203166,
      "grad_norm": 1.245638712405433,
      "learning_rate": 6.578947368421054e-06,
      "loss": 0.0933,
      "step": 100
    },
    {
      "epoch": 0.13324538258575197,
      "grad_norm": 1.5693953291777372,
      "learning_rate": 6.644736842105264e-06,
      "loss": 0.1102,
      "step": 101
    },
    {
      "epoch": 0.1345646437994723,
      "grad_norm": 1.283076950435831,
      "learning_rate": 6.710526315789474e-06,
      "loss": 0.0988,
      "step": 102
    },
    {
      "epoch": 0.1358839050131926,
      "grad_norm": 1.350859235672517,
      "learning_rate": 6.776315789473686e-06,
      "loss": 0.085,
      "step": 103
    },
    {
      "epoch": 0.13720316622691292,
      "grad_norm": 1.0796020601602057,
      "learning_rate": 6.842105263157896e-06,
      "loss": 0.099,
      "step": 104
    },
    {
      "epoch": 0.13852242744063326,
      "grad_norm": 1.1275227482126249,
      "learning_rate": 6.907894736842106e-06,
      "loss": 0.0797,
      "step": 105
    },
    {
      "epoch": 0.13984168865435356,
      "grad_norm": 1.0832151473779532,
      "learning_rate": 6.973684210526316e-06,
      "loss": 0.0918,
      "step": 106
    },
    {
      "epoch": 0.14116094986807387,
      "grad_norm": 0.7930455753517637,
      "learning_rate": 7.0394736842105274e-06,
      "loss": 0.0644,
      "step": 107
    },
    {
      "epoch": 0.1424802110817942,
      "grad_norm": 1.592218242485485,
      "learning_rate": 7.1052631578947375e-06,
      "loss": 0.1265,
      "step": 108
    },
    {
      "epoch": 0.1437994722955145,
      "grad_norm": 1.3389622863855597,
      "learning_rate": 7.1710526315789475e-06,
      "loss": 0.0718,
      "step": 109
    },
    {
      "epoch": 0.14511873350923482,
      "grad_norm": 1.544338181296543,
      "learning_rate": 7.236842105263158e-06,
      "loss": 0.1003,
      "step": 110
    },
    {
      "epoch": 0.14643799472295516,
      "grad_norm": 1.349874808615941,
      "learning_rate": 7.302631578947369e-06,
      "loss": 0.1069,
      "step": 111
    },
    {
      "epoch": 0.14775725593667546,
      "grad_norm": 1.49277474685825,
      "learning_rate": 7.368421052631579e-06,
      "loss": 0.1045,
      "step": 112
    },
    {
      "epoch": 0.14907651715039577,
      "grad_norm": 1.8999681992305248,
      "learning_rate": 7.43421052631579e-06,
      "loss": 0.1392,
      "step": 113
    },
    {
      "epoch": 0.1503957783641161,
      "grad_norm": 1.3082439341312861,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.0947,
      "step": 114
    },
    {
      "epoch": 0.1517150395778364,
      "grad_norm": 0.8488265412383513,
      "learning_rate": 7.565789473684211e-06,
      "loss": 0.0716,
      "step": 115
    },
    {
      "epoch": 0.15303430079155672,
      "grad_norm": 0.9426471120997849,
      "learning_rate": 7.631578947368423e-06,
      "loss": 0.0793,
      "step": 116
    },
    {
      "epoch": 0.15435356200527706,
      "grad_norm": 0.9809029094053667,
      "learning_rate": 7.697368421052632e-06,
      "loss": 0.0829,
      "step": 117
    },
    {
      "epoch": 0.15567282321899736,
      "grad_norm": 1.2026525607016512,
      "learning_rate": 7.763157894736843e-06,
      "loss": 0.0948,
      "step": 118
    },
    {
      "epoch": 0.15699208443271767,
      "grad_norm": 1.3972003502648243,
      "learning_rate": 7.828947368421054e-06,
      "loss": 0.0802,
      "step": 119
    },
    {
      "epoch": 0.158311345646438,
      "grad_norm": 1.464974484595394,
      "learning_rate": 7.894736842105265e-06,
      "loss": 0.1252,
      "step": 120
    },
    {
      "epoch": 0.15963060686015831,
      "grad_norm": 1.3659579209534767,
      "learning_rate": 7.960526315789474e-06,
      "loss": 0.0882,
      "step": 121
    },
    {
      "epoch": 0.16094986807387862,
      "grad_norm": 0.8868074475530271,
      "learning_rate": 8.026315789473685e-06,
      "loss": 0.066,
      "step": 122
    },
    {
      "epoch": 0.16226912928759896,
      "grad_norm": 1.2621289694628735,
      "learning_rate": 8.092105263157896e-06,
      "loss": 0.0687,
      "step": 123
    },
    {
      "epoch": 0.16358839050131926,
      "grad_norm": 1.6829617306054463,
      "learning_rate": 8.157894736842106e-06,
      "loss": 0.1144,
      "step": 124
    },
    {
      "epoch": 0.16490765171503957,
      "grad_norm": 1.251837581274231,
      "learning_rate": 8.223684210526316e-06,
      "loss": 0.0881,
      "step": 125
    },
    {
      "epoch": 0.1662269129287599,
      "grad_norm": 0.8602941113686816,
      "learning_rate": 8.289473684210526e-06,
      "loss": 0.0743,
      "step": 126
    },
    {
      "epoch": 0.16754617414248021,
      "grad_norm": 1.0273660667252307,
      "learning_rate": 8.355263157894737e-06,
      "loss": 0.08,
      "step": 127
    },
    {
      "epoch": 0.16886543535620052,
      "grad_norm": 2.6198303125098352,
      "learning_rate": 8.421052631578948e-06,
      "loss": 0.0983,
      "step": 128
    },
    {
      "epoch": 0.17018469656992086,
      "grad_norm": 1.0880312381207564,
      "learning_rate": 8.486842105263159e-06,
      "loss": 0.0698,
      "step": 129
    },
    {
      "epoch": 0.17150395778364116,
      "grad_norm": 1.9176206928976167,
      "learning_rate": 8.552631578947368e-06,
      "loss": 0.095,
      "step": 130
    },
    {
      "epoch": 0.17282321899736147,
      "grad_norm": 4.856631037789346,
      "learning_rate": 8.61842105263158e-06,
      "loss": 0.0972,
      "step": 131
    },
    {
      "epoch": 0.1741424802110818,
      "grad_norm": 1.3189375702822577,
      "learning_rate": 8.68421052631579e-06,
      "loss": 0.1083,
      "step": 132
    },
    {
      "epoch": 0.17546174142480211,
      "grad_norm": 1.348920100799652,
      "learning_rate": 8.750000000000001e-06,
      "loss": 0.0801,
      "step": 133
    },
    {
      "epoch": 0.17678100263852242,
      "grad_norm": 1.0662674243220835,
      "learning_rate": 8.81578947368421e-06,
      "loss": 0.085,
      "step": 134
    },
    {
      "epoch": 0.17810026385224276,
      "grad_norm": 1.8683103052234746,
      "learning_rate": 8.881578947368423e-06,
      "loss": 0.0812,
      "step": 135
    },
    {
      "epoch": 0.17941952506596306,
      "grad_norm": 0.9975688194192199,
      "learning_rate": 8.947368421052632e-06,
      "loss": 0.0693,
      "step": 136
    },
    {
      "epoch": 0.18073878627968337,
      "grad_norm": 0.9103667124063366,
      "learning_rate": 9.013157894736843e-06,
      "loss": 0.0751,
      "step": 137
    },
    {
      "epoch": 0.1820580474934037,
      "grad_norm": 1.8372221347381417,
      "learning_rate": 9.078947368421054e-06,
      "loss": 0.1082,
      "step": 138
    },
    {
      "epoch": 0.18337730870712401,
      "grad_norm": 1.0083599090393338,
      "learning_rate": 9.144736842105264e-06,
      "loss": 0.106,
      "step": 139
    },
    {
      "epoch": 0.18469656992084432,
      "grad_norm": 0.9763025849661937,
      "learning_rate": 9.210526315789474e-06,
      "loss": 0.0893,
      "step": 140
    },
    {
      "epoch": 0.18601583113456466,
      "grad_norm": 2.6732971208195955,
      "learning_rate": 9.276315789473686e-06,
      "loss": 0.1015,
      "step": 141
    },
    {
      "epoch": 0.18733509234828497,
      "grad_norm": 1.3827420544593714,
      "learning_rate": 9.342105263157895e-06,
      "loss": 0.1079,
      "step": 142
    },
    {
      "epoch": 0.18865435356200527,
      "grad_norm": 7.163906695650369,
      "learning_rate": 9.407894736842106e-06,
      "loss": 0.1498,
      "step": 143
    },
    {
      "epoch": 0.18997361477572558,
      "grad_norm": 1.363322395020731,
      "learning_rate": 9.473684210526315e-06,
      "loss": 0.0799,
      "step": 144
    },
    {
      "epoch": 0.19129287598944592,
      "grad_norm": 0.9848860707778788,
      "learning_rate": 9.539473684210528e-06,
      "loss": 0.0714,
      "step": 145
    },
    {
      "epoch": 0.19261213720316622,
      "grad_norm": 0.7355035178684772,
      "learning_rate": 9.605263157894737e-06,
      "loss": 0.0639,
      "step": 146
    },
    {
      "epoch": 0.19393139841688653,
      "grad_norm": 0.9708590815947932,
      "learning_rate": 9.671052631578948e-06,
      "loss": 0.0829,
      "step": 147
    },
    {
      "epoch": 0.19525065963060687,
      "grad_norm": 1.3176260473068497,
      "learning_rate": 9.736842105263159e-06,
      "loss": 0.0869,
      "step": 148
    },
    {
      "epoch": 0.19656992084432717,
      "grad_norm": 0.8290463498340741,
      "learning_rate": 9.80263157894737e-06,
      "loss": 0.075,
      "step": 149
    },
    {
      "epoch": 0.19788918205804748,
      "grad_norm": 1.2224418207646102,
      "learning_rate": 9.868421052631579e-06,
      "loss": 0.1049,
      "step": 150
    },
    {
      "epoch": 0.19920844327176782,
      "grad_norm": 0.9383534609793346,
      "learning_rate": 9.93421052631579e-06,
      "loss": 0.0693,
      "step": 151
    },
    {
      "epoch": 0.20052770448548812,
      "grad_norm": 1.157120293265899,
      "learning_rate": 1e-05,
      "loss": 0.1242,
      "step": 152
    },
    {
      "epoch": 0.20184696569920843,
      "grad_norm": 1.3848254542503755,
      "learning_rate": 9.9999867379451e-06,
      "loss": 0.0949,
      "step": 153
    },
    {
      "epoch": 0.20316622691292877,
      "grad_norm": 1.1050363308018347,
      "learning_rate": 9.99994695185075e-06,
      "loss": 0.0681,
      "step": 154
    },
    {
      "epoch": 0.20448548812664907,
      "grad_norm": 1.241266481405317,
      "learning_rate": 9.999880641928008e-06,
      "loss": 0.1013,
      "step": 155
    },
    {
      "epoch": 0.20580474934036938,
      "grad_norm": 1.2502664498399243,
      "learning_rate": 9.999787808528639e-06,
      "loss": 0.0958,
      "step": 156
    },
    {
      "epoch": 0.20712401055408972,
      "grad_norm": 0.8902305137007472,
      "learning_rate": 9.999668452145104e-06,
      "loss": 0.0707,
      "step": 157
    },
    {
      "epoch": 0.20844327176781002,
      "grad_norm": 1.2565435303569972,
      "learning_rate": 9.99952257341057e-06,
      "loss": 0.0855,
      "step": 158
    },
    {
      "epoch": 0.20976253298153033,
      "grad_norm": 1.0108162794824818,
      "learning_rate": 9.999350173098899e-06,
      "loss": 0.0828,
      "step": 159
    },
    {
      "epoch": 0.21108179419525067,
      "grad_norm": 1.1432130463466625,
      "learning_rate": 9.999151252124639e-06,
      "loss": 0.0788,
      "step": 160
    },
    {
      "epoch": 0.21240105540897097,
      "grad_norm": 1.2307082448027162,
      "learning_rate": 9.998925811543034e-06,
      "loss": 0.073,
      "step": 161
    },
    {
      "epoch": 0.21372031662269128,
      "grad_norm": 1.1691071584736277,
      "learning_rate": 9.998673852550007e-06,
      "loss": 0.103,
      "step": 162
    },
    {
      "epoch": 0.21503957783641162,
      "grad_norm": 0.9809626759661192,
      "learning_rate": 9.998395376482152e-06,
      "loss": 0.0682,
      "step": 163
    },
    {
      "epoch": 0.21635883905013192,
      "grad_norm": 0.903420296034603,
      "learning_rate": 9.99809038481674e-06,
      "loss": 0.0756,
      "step": 164
    },
    {
      "epoch": 0.21767810026385223,
      "grad_norm": 1.2846822797227462,
      "learning_rate": 9.997758879171693e-06,
      "loss": 0.0923,
      "step": 165
    },
    {
      "epoch": 0.21899736147757257,
      "grad_norm": 1.0264204749663686,
      "learning_rate": 9.99740086130559e-06,
      "loss": 0.0953,
      "step": 166
    },
    {
      "epoch": 0.22031662269129287,
      "grad_norm": 1.0371481360943675,
      "learning_rate": 9.997016333117655e-06,
      "loss": 0.0652,
      "step": 167
    },
    {
      "epoch": 0.22163588390501318,
      "grad_norm": 1.2651198967501744,
      "learning_rate": 9.996605296647737e-06,
      "loss": 0.1112,
      "step": 168
    },
    {
      "epoch": 0.22295514511873352,
      "grad_norm": 0.9017358797708437,
      "learning_rate": 9.996167754076315e-06,
      "loss": 0.0621,
      "step": 169
    },
    {
      "epoch": 0.22427440633245382,
      "grad_norm": 1.3003546894370093,
      "learning_rate": 9.995703707724474e-06,
      "loss": 0.1046,
      "step": 170
    },
    {
      "epoch": 0.22559366754617413,
      "grad_norm": 1.2408727983936052,
      "learning_rate": 9.995213160053897e-06,
      "loss": 0.0983,
      "step": 171
    },
    {
      "epoch": 0.22691292875989447,
      "grad_norm": 0.8514473822455528,
      "learning_rate": 9.99469611366685e-06,
      "loss": 0.0803,
      "step": 172
    },
    {
      "epoch": 0.22823218997361477,
      "grad_norm": 0.8677470665047625,
      "learning_rate": 9.994152571306174e-06,
      "loss": 0.0669,
      "step": 173
    },
    {
      "epoch": 0.22955145118733508,
      "grad_norm": 1.0781370854799601,
      "learning_rate": 9.993582535855265e-06,
      "loss": 0.0831,
      "step": 174
    },
    {
      "epoch": 0.23087071240105542,
      "grad_norm": 1.3878995049892582,
      "learning_rate": 9.992986010338058e-06,
      "loss": 0.1166,
      "step": 175
    },
    {
      "epoch": 0.23218997361477572,
      "grad_norm": 1.2107428243400042,
      "learning_rate": 9.992362997919016e-06,
      "loss": 0.1049,
      "step": 176
    },
    {
      "epoch": 0.23350923482849603,
      "grad_norm": 0.7146543623664033,
      "learning_rate": 9.991713501903107e-06,
      "loss": 0.0581,
      "step": 177
    },
    {
      "epoch": 0.23482849604221637,
      "grad_norm": 1.05069811104466,
      "learning_rate": 9.991037525735794e-06,
      "loss": 0.0736,
      "step": 178
    },
    {
      "epoch": 0.23614775725593667,
      "grad_norm": 0.8007759874947845,
      "learning_rate": 9.99033507300301e-06,
      "loss": 0.06,
      "step": 179
    },
    {
      "epoch": 0.23746701846965698,
      "grad_norm": 1.79713660450814,
      "learning_rate": 9.98960614743114e-06,
      "loss": 0.1229,
      "step": 180
    },
    {
      "epoch": 0.23878627968337732,
      "grad_norm": 0.912606187507227,
      "learning_rate": 9.988850752887006e-06,
      "loss": 0.0773,
      "step": 181
    },
    {
      "epoch": 0.24010554089709762,
      "grad_norm": 0.9167662097707882,
      "learning_rate": 9.988068893377841e-06,
      "loss": 0.0643,
      "step": 182
    },
    {
      "epoch": 0.24142480211081793,
      "grad_norm": 1.3792149016555213,
      "learning_rate": 9.987260573051268e-06,
      "loss": 0.0754,
      "step": 183
    },
    {
      "epoch": 0.24274406332453827,
      "grad_norm": 1.1294403281169605,
      "learning_rate": 9.986425796195287e-06,
      "loss": 0.1082,
      "step": 184
    },
    {
      "epoch": 0.24406332453825857,
      "grad_norm": 1.375642418993936,
      "learning_rate": 9.985564567238237e-06,
      "loss": 0.0977,
      "step": 185
    },
    {
      "epoch": 0.24538258575197888,
      "grad_norm": 1.1064905057054284,
      "learning_rate": 9.984676890748787e-06,
      "loss": 0.088,
      "step": 186
    },
    {
      "epoch": 0.24670184696569922,
      "grad_norm": 0.9950645396905647,
      "learning_rate": 9.983762771435902e-06,
      "loss": 0.0897,
      "step": 187
    },
    {
      "epoch": 0.24802110817941952,
      "grad_norm": 0.7705270623146346,
      "learning_rate": 9.98282221414882e-06,
      "loss": 0.0749,
      "step": 188
    },
    {
      "epoch": 0.24934036939313983,
      "grad_norm": 2.201638940290167,
      "learning_rate": 9.981855223877032e-06,
      "loss": 0.1012,
      "step": 189
    },
    {
      "epoch": 0.25065963060686014,
      "grad_norm": 1.2712247018236318,
      "learning_rate": 9.98086180575025e-06,
      "loss": 0.0923,
      "step": 190
    },
    {
      "epoch": 0.2519788918205805,
      "grad_norm": 0.6249886222978469,
      "learning_rate": 9.97984196503838e-06,
      "loss": 0.0609,
      "step": 191
    },
    {
      "epoch": 0.2532981530343008,
      "grad_norm": 1.5723383799027848,
      "learning_rate": 9.978795707151492e-06,
      "loss": 0.0854,
      "step": 192
    },
    {
      "epoch": 0.2546174142480211,
      "grad_norm": 0.957499710618973,
      "learning_rate": 9.9777230376398e-06,
      "loss": 0.082,
      "step": 193
    },
    {
      "epoch": 0.2559366754617414,
      "grad_norm": 1.4768620196656903,
      "learning_rate": 9.976623962193627e-06,
      "loss": 0.0923,
      "step": 194
    },
    {
      "epoch": 0.25725593667546176,
      "grad_norm": 1.5999528442765696,
      "learning_rate": 9.975498486643368e-06,
      "loss": 0.0696,
      "step": 195
    },
    {
      "epoch": 0.25857519788918204,
      "grad_norm": 1.235675356279894,
      "learning_rate": 9.974346616959476e-06,
      "loss": 0.0816,
      "step": 196
    },
    {
      "epoch": 0.2598944591029024,
      "grad_norm": 1.1903571766318424,
      "learning_rate": 9.973168359252411e-06,
      "loss": 0.0753,
      "step": 197
    },
    {
      "epoch": 0.2612137203166227,
      "grad_norm": 1.0319046665463454,
      "learning_rate": 9.971963719772621e-06,
      "loss": 0.0802,
      "step": 198
    },
    {
      "epoch": 0.262532981530343,
      "grad_norm": 0.9113306304093047,
      "learning_rate": 9.970732704910504e-06,
      "loss": 0.0768,
      "step": 199
    },
    {
      "epoch": 0.2638522427440633,
      "grad_norm": 1.552482915540632,
      "learning_rate": 9.969475321196374e-06,
      "loss": 0.1028,
      "step": 200
    },
    {
      "epoch": 0.26517150395778366,
      "grad_norm": 0.8279365731599866,
      "learning_rate": 9.968191575300428e-06,
      "loss": 0.0699,
      "step": 201
    },
    {
      "epoch": 0.26649076517150394,
      "grad_norm": 1.023073346353744,
      "learning_rate": 9.966881474032711e-06,
      "loss": 0.0644,
      "step": 202
    },
    {
      "epoch": 0.2678100263852243,
      "grad_norm": 0.8921330416372578,
      "learning_rate": 9.965545024343076e-06,
      "loss": 0.0734,
      "step": 203
    },
    {
      "epoch": 0.2691292875989446,
      "grad_norm": 0.8293631684614154,
      "learning_rate": 9.96418223332115e-06,
      "loss": 0.0866,
      "step": 204
    },
    {
      "epoch": 0.2704485488126649,
      "grad_norm": 1.2587903865990873,
      "learning_rate": 9.962793108196299e-06,
      "loss": 0.0901,
      "step": 205
    },
    {
      "epoch": 0.2717678100263852,
      "grad_norm": 1.2148007735564763,
      "learning_rate": 9.961377656337579e-06,
      "loss": 0.0771,
      "step": 206
    },
    {
      "epoch": 0.27308707124010556,
      "grad_norm": 1.264408003966606,
      "learning_rate": 9.959935885253715e-06,
      "loss": 0.0703,
      "step": 207
    },
    {
      "epoch": 0.27440633245382584,
      "grad_norm": 1.446028223338931,
      "learning_rate": 9.958467802593046e-06,
      "loss": 0.088,
      "step": 208
    },
    {
      "epoch": 0.2757255936675462,
      "grad_norm": 0.7220366729686982,
      "learning_rate": 9.956973416143488e-06,
      "loss": 0.0815,
      "step": 209
    },
    {
      "epoch": 0.2770448548812665,
      "grad_norm": 1.5607728210271763,
      "learning_rate": 9.955452733832493e-06,
      "loss": 0.1029,
      "step": 210
    },
    {
      "epoch": 0.2783641160949868,
      "grad_norm": 1.4020322542800785,
      "learning_rate": 9.953905763727015e-06,
      "loss": 0.0666,
      "step": 211
    },
    {
      "epoch": 0.2796833773087071,
      "grad_norm": 0.7076793659310983,
      "learning_rate": 9.952332514033449e-06,
      "loss": 0.0688,
      "step": 212
    },
    {
      "epoch": 0.28100263852242746,
      "grad_norm": 1.2502152245372904,
      "learning_rate": 9.950732993097608e-06,
      "loss": 0.0955,
      "step": 213
    },
    {
      "epoch": 0.28232189973614774,
      "grad_norm": 0.6863402755600805,
      "learning_rate": 9.949107209404664e-06,
      "loss": 0.0664,
      "step": 214
    },
    {
      "epoch": 0.2836411609498681,
      "grad_norm": 0.9324176689524081,
      "learning_rate": 9.947455171579112e-06,
      "loss": 0.0792,
      "step": 215
    },
    {
      "epoch": 0.2849604221635884,
      "grad_norm": 0.7850777734100792,
      "learning_rate": 9.94577688838472e-06,
      "loss": 0.0811,
      "step": 216
    },
    {
      "epoch": 0.2862796833773087,
      "grad_norm": 0.8058284364958739,
      "learning_rate": 9.944072368724476e-06,
      "loss": 0.0722,
      "step": 217
    },
    {
      "epoch": 0.287598944591029,
      "grad_norm": 0.969649949605035,
      "learning_rate": 9.942341621640558e-06,
      "loss": 0.0755,
      "step": 218
    },
    {
      "epoch": 0.28891820580474936,
      "grad_norm": 0.9927927695753133,
      "learning_rate": 9.940584656314269e-06,
      "loss": 0.0826,
      "step": 219
    },
    {
      "epoch": 0.29023746701846964,
      "grad_norm": 0.7580345939150698,
      "learning_rate": 9.938801482065998e-06,
      "loss": 0.0704,
      "step": 220
    },
    {
      "epoch": 0.29155672823219,
      "grad_norm": 0.5935527535305033,
      "learning_rate": 9.936992108355168e-06,
      "loss": 0.062,
      "step": 221
    },
    {
      "epoch": 0.2928759894459103,
      "grad_norm": 0.7799550966222378,
      "learning_rate": 9.935156544780183e-06,
      "loss": 0.0715,
      "step": 222
    },
    {
      "epoch": 0.2941952506596306,
      "grad_norm": 0.6585324473995096,
      "learning_rate": 9.93329480107838e-06,
      "loss": 0.0644,
      "step": 223
    },
    {
      "epoch": 0.2955145118733509,
      "grad_norm": 0.7062926953294116,
      "learning_rate": 9.93140688712598e-06,
      "loss": 0.0682,
      "step": 224
    },
    {
      "epoch": 0.29683377308707126,
      "grad_norm": 0.6994497926974289,
      "learning_rate": 9.929492812938028e-06,
      "loss": 0.0646,
      "step": 225
    },
    {
      "epoch": 0.29815303430079154,
      "grad_norm": 0.9870519318595599,
      "learning_rate": 9.92755258866835e-06,
      "loss": 0.088,
      "step": 226
    },
    {
      "epoch": 0.2994722955145119,
      "grad_norm": 0.7685429820565252,
      "learning_rate": 9.925586224609489e-06,
      "loss": 0.0533,
      "step": 227
    },
    {
      "epoch": 0.3007915567282322,
      "grad_norm": 0.7643317383642257,
      "learning_rate": 9.923593731192655e-06,
      "loss": 0.0626,
      "step": 228
    },
    {
      "epoch": 0.3021108179419525,
      "grad_norm": 0.6293974246135788,
      "learning_rate": 9.921575118987672e-06,
      "loss": 0.0599,
      "step": 229
    },
    {
      "epoch": 0.3034300791556728,
      "grad_norm": 1.1719158706259716,
      "learning_rate": 9.919530398702917e-06,
      "loss": 0.0766,
      "step": 230
    },
    {
      "epoch": 0.30474934036939316,
      "grad_norm": 1.1836452278232157,
      "learning_rate": 9.917459581185269e-06,
      "loss": 0.1078,
      "step": 231
    },
    {
      "epoch": 0.30606860158311344,
      "grad_norm": 0.7456849787237535,
      "learning_rate": 9.915362677420045e-06,
      "loss": 0.0594,
      "step": 232
    },
    {
      "epoch": 0.3073878627968338,
      "grad_norm": 1.0171580488977425,
      "learning_rate": 9.913239698530947e-06,
      "loss": 0.0764,
      "step": 233
    },
    {
      "epoch": 0.3087071240105541,
      "grad_norm": 0.7216218185691871,
      "learning_rate": 9.911090655779997e-06,
      "loss": 0.056,
      "step": 234
    },
    {
      "epoch": 0.3100263852242744,
      "grad_norm": 1.0205408613448763,
      "learning_rate": 9.90891556056749e-06,
      "loss": 0.0697,
      "step": 235
    },
    {
      "epoch": 0.3113456464379947,
      "grad_norm": 0.6787788419289598,
      "learning_rate": 9.906714424431914e-06,
      "loss": 0.0515,
      "step": 236
    },
    {
      "epoch": 0.31266490765171506,
      "grad_norm": 0.6511400074009881,
      "learning_rate": 9.904487259049907e-06,
      "loss": 0.052,
      "step": 237
    },
    {
      "epoch": 0.31398416886543534,
      "grad_norm": 0.9175979604887738,
      "learning_rate": 9.902234076236182e-06,
      "loss": 0.0712,
      "step": 238
    },
    {
      "epoch": 0.3153034300791557,
      "grad_norm": 1.0786583925042368,
      "learning_rate": 9.899954887943474e-06,
      "loss": 0.0937,
      "step": 239
    },
    {
      "epoch": 0.316622691292876,
      "grad_norm": 1.1109202209669133,
      "learning_rate": 9.897649706262474e-06,
      "loss": 0.0746,
      "step": 240
    },
    {
      "epoch": 0.3179419525065963,
      "grad_norm": 1.344033170501885,
      "learning_rate": 9.895318543421755e-06,
      "loss": 0.1053,
      "step": 241
    },
    {
      "epoch": 0.31926121372031663,
      "grad_norm": 0.6626492354910761,
      "learning_rate": 9.892961411787725e-06,
      "loss": 0.0623,
      "step": 242
    },
    {
      "epoch": 0.32058047493403696,
      "grad_norm": 1.3153644580390524,
      "learning_rate": 9.890578323864546e-06,
      "loss": 0.0823,
      "step": 243
    },
    {
      "epoch": 0.32189973614775724,
      "grad_norm": 0.9897293095755729,
      "learning_rate": 9.888169292294077e-06,
      "loss": 0.0986,
      "step": 244
    },
    {
      "epoch": 0.3232189973614776,
      "grad_norm": 0.9305565118911147,
      "learning_rate": 9.885734329855798e-06,
      "loss": 0.089,
      "step": 245
    },
    {
      "epoch": 0.3245382585751979,
      "grad_norm": 0.9362582559092152,
      "learning_rate": 9.883273449466755e-06,
      "loss": 0.0924,
      "step": 246
    },
    {
      "epoch": 0.3258575197889182,
      "grad_norm": 0.6789391312855251,
      "learning_rate": 9.880786664181477e-06,
      "loss": 0.0625,
      "step": 247
    },
    {
      "epoch": 0.32717678100263853,
      "grad_norm": 0.7753222910069528,
      "learning_rate": 9.87827398719192e-06,
      "loss": 0.072,
      "step": 248
    },
    {
      "epoch": 0.32849604221635886,
      "grad_norm": 0.9497710959608149,
      "learning_rate": 9.875735431827386e-06,
      "loss": 0.1043,
      "step": 249
    },
    {
      "epoch": 0.32981530343007914,
      "grad_norm": 0.9888364262598249,
      "learning_rate": 9.87317101155446e-06,
      "loss": 0.0684,
      "step": 250
    },
    {
      "epoch": 0.3311345646437995,
      "grad_norm": 0.7512692600971639,
      "learning_rate": 9.870580739976936e-06,
      "loss": 0.0771,
      "step": 251
    },
    {
      "epoch": 0.3324538258575198,
      "grad_norm": 1.3729675639562857,
      "learning_rate": 9.867964630835742e-06,
      "loss": 0.1031,
      "step": 252
    },
    {
      "epoch": 0.3337730870712401,
      "grad_norm": 0.8564354149718731,
      "learning_rate": 9.865322698008873e-06,
      "loss": 0.0948,
      "step": 253
    },
    {
      "epoch": 0.33509234828496043,
      "grad_norm": 0.6605521047813258,
      "learning_rate": 9.862654955511309e-06,
      "loss": 0.0592,
      "step": 254
    },
    {
      "epoch": 0.33641160949868076,
      "grad_norm": 0.8601197600270915,
      "learning_rate": 9.859961417494952e-06,
      "loss": 0.0741,
      "step": 255
    },
    {
      "epoch": 0.33773087071240104,
      "grad_norm": 0.9894110282228973,
      "learning_rate": 9.857242098248543e-06,
      "loss": 0.0858,
      "step": 256
    },
    {
      "epoch": 0.3390501319261214,
      "grad_norm": 1.4242673666601409,
      "learning_rate": 9.854497012197581e-06,
      "loss": 0.0827,
      "step": 257
    },
    {
      "epoch": 0.3403693931398417,
      "grad_norm": 1.0206885764543667,
      "learning_rate": 9.851726173904264e-06,
      "loss": 0.0864,
      "step": 258
    },
    {
      "epoch": 0.341688654353562,
      "grad_norm": 0.9123864354174962,
      "learning_rate": 9.848929598067393e-06,
      "loss": 0.0595,
      "step": 259
    },
    {
      "epoch": 0.34300791556728233,
      "grad_norm": 1.0914805753448291,
      "learning_rate": 9.846107299522305e-06,
      "loss": 0.0741,
      "step": 260
    },
    {
      "epoch": 0.34432717678100266,
      "grad_norm": 0.7658161529334118,
      "learning_rate": 9.843259293240793e-06,
      "loss": 0.0553,
      "step": 261
    },
    {
      "epoch": 0.34564643799472294,
      "grad_norm": 1.7776327682313273,
      "learning_rate": 9.840385594331022e-06,
      "loss": 0.0866,
      "step": 262
    },
    {
      "epoch": 0.3469656992084433,
      "grad_norm": 0.8366360091521257,
      "learning_rate": 9.837486218037453e-06,
      "loss": 0.0688,
      "step": 263
    },
    {
      "epoch": 0.3482849604221636,
      "grad_norm": 1.0370760066220694,
      "learning_rate": 9.834561179740763e-06,
      "loss": 0.066,
      "step": 264
    },
    {
      "epoch": 0.3496042216358839,
      "grad_norm": 0.735064894973884,
      "learning_rate": 9.831610494957756e-06,
      "loss": 0.0701,
      "step": 265
    },
    {
      "epoch": 0.35092348284960423,
      "grad_norm": 0.5842610061663901,
      "learning_rate": 9.828634179341292e-06,
      "loss": 0.0609,
      "step": 266
    },
    {
      "epoch": 0.35224274406332456,
      "grad_norm": 0.9027630240588085,
      "learning_rate": 9.825632248680195e-06,
      "loss": 0.0632,
      "step": 267
    },
    {
      "epoch": 0.35356200527704484,
      "grad_norm": 0.9391224059433742,
      "learning_rate": 9.82260471889917e-06,
      "loss": 0.0632,
      "step": 268
    },
    {
      "epoch": 0.3548812664907652,
      "grad_norm": 0.5654562995248813,
      "learning_rate": 9.819551606058729e-06,
      "loss": 0.0731,
      "step": 269
    },
    {
      "epoch": 0.3562005277044855,
      "grad_norm": 0.7487458340440738,
      "learning_rate": 9.816472926355087e-06,
      "loss": 0.0739,
      "step": 270
    },
    {
      "epoch": 0.3575197889182058,
      "grad_norm": 0.951138507364624,
      "learning_rate": 9.813368696120095e-06,
      "loss": 0.0623,
      "step": 271
    },
    {
      "epoch": 0.35883905013192613,
      "grad_norm": 0.5810859250822077,
      "learning_rate": 9.810238931821139e-06,
      "loss": 0.0604,
      "step": 272
    },
    {
      "epoch": 0.36015831134564646,
      "grad_norm": 0.7591306012531617,
      "learning_rate": 9.807083650061063e-06,
      "loss": 0.0717,
      "step": 273
    },
    {
      "epoch": 0.36147757255936674,
      "grad_norm": 0.7553326022985923,
      "learning_rate": 9.803902867578075e-06,
      "loss": 0.0748,
      "step": 274
    },
    {
      "epoch": 0.3627968337730871,
      "grad_norm": 0.8149927472285278,
      "learning_rate": 9.800696601245658e-06,
      "loss": 0.0869,
      "step": 275
    },
    {
      "epoch": 0.3641160949868074,
      "grad_norm": 0.7554080086493412,
      "learning_rate": 9.797464868072489e-06,
      "loss": 0.0882,
      "step": 276
    },
    {
      "epoch": 0.3654353562005277,
      "grad_norm": 0.9805029259857,
      "learning_rate": 9.79420768520233e-06,
      "loss": 0.086,
      "step": 277
    },
    {
      "epoch": 0.36675461741424803,
      "grad_norm": 0.5320046970882769,
      "learning_rate": 9.790925069913962e-06,
      "loss": 0.0556,
      "step": 278
    },
    {
      "epoch": 0.36807387862796836,
      "grad_norm": 1.0328295566370667,
      "learning_rate": 9.787617039621071e-06,
      "loss": 0.0562,
      "step": 279
    },
    {
      "epoch": 0.36939313984168864,
      "grad_norm": 0.7291324833094329,
      "learning_rate": 9.78428361187217e-06,
      "loss": 0.0753,
      "step": 280
    },
    {
      "epoch": 0.370712401055409,
      "grad_norm": 0.8601042782278124,
      "learning_rate": 9.7809248043505e-06,
      "loss": 0.0891,
      "step": 281
    },
    {
      "epoch": 0.3720316622691293,
      "grad_norm": 0.9846921049495289,
      "learning_rate": 9.777540634873939e-06,
      "loss": 0.0742,
      "step": 282
    },
    {
      "epoch": 0.3733509234828496,
      "grad_norm": 0.5770895688506988,
      "learning_rate": 9.774131121394898e-06,
      "loss": 0.0506,
      "step": 283
    },
    {
      "epoch": 0.37467018469656993,
      "grad_norm": 0.7478778631896724,
      "learning_rate": 9.770696282000245e-06,
      "loss": 0.0618,
      "step": 284
    },
    {
      "epoch": 0.3759894459102902,
      "grad_norm": 0.9950610396799793,
      "learning_rate": 9.767236134911188e-06,
      "loss": 0.0978,
      "step": 285
    },
    {
      "epoch": 0.37730870712401055,
      "grad_norm": 0.5646364519415039,
      "learning_rate": 9.763750698483192e-06,
      "loss": 0.0597,
      "step": 286
    },
    {
      "epoch": 0.3786279683377309,
      "grad_norm": 0.9558965606066078,
      "learning_rate": 9.760239991205878e-06,
      "loss": 0.0748,
      "step": 287
    },
    {
      "epoch": 0.37994722955145116,
      "grad_norm": 0.6806130282946865,
      "learning_rate": 9.756704031702919e-06,
      "loss": 0.0665,
      "step": 288
    },
    {
      "epoch": 0.3812664907651715,
      "grad_norm": 1.3837371430943852,
      "learning_rate": 9.753142838731956e-06,
      "loss": 0.0704,
      "step": 289
    },
    {
      "epoch": 0.38258575197889183,
      "grad_norm": 1.3991857061373745,
      "learning_rate": 9.74955643118448e-06,
      "loss": 0.1142,
      "step": 290
    },
    {
      "epoch": 0.3839050131926121,
      "grad_norm": 0.598821404550848,
      "learning_rate": 9.745944828085747e-06,
      "loss": 0.06,
      "step": 291
    },
    {
      "epoch": 0.38522427440633245,
      "grad_norm": 0.8680968549739181,
      "learning_rate": 9.742308048594665e-06,
      "loss": 0.0668,
      "step": 292
    },
    {
      "epoch": 0.3865435356200528,
      "grad_norm": 0.8256824842529236,
      "learning_rate": 9.738646112003705e-06,
      "loss": 0.0786,
      "step": 293
    },
    {
      "epoch": 0.38786279683377306,
      "grad_norm": 1.1696409634800806,
      "learning_rate": 9.734959037738788e-06,
      "loss": 0.0728,
      "step": 294
    },
    {
      "epoch": 0.3891820580474934,
      "grad_norm": 1.027580435542134,
      "learning_rate": 9.731246845359187e-06,
      "loss": 0.0728,
      "step": 295
    },
    {
      "epoch": 0.39050131926121373,
      "grad_norm": 0.6566278180751907,
      "learning_rate": 9.727509554557416e-06,
      "loss": 0.0788,
      "step": 296
    },
    {
      "epoch": 0.391820580474934,
      "grad_norm": 1.085693108648712,
      "learning_rate": 9.723747185159146e-06,
      "loss": 0.0689,
      "step": 297
    },
    {
      "epoch": 0.39313984168865435,
      "grad_norm": 0.8493527436632392,
      "learning_rate": 9.719959757123073e-06,
      "loss": 0.0643,
      "step": 298
    },
    {
      "epoch": 0.3944591029023747,
      "grad_norm": 0.6028556111386613,
      "learning_rate": 9.716147290540826e-06,
      "loss": 0.056,
      "step": 299
    },
    {
      "epoch": 0.39577836411609496,
      "grad_norm": 1.0034076205039018,
      "learning_rate": 9.712309805636863e-06,
      "loss": 0.0764,
      "step": 300
    },
    {
      "epoch": 0.3970976253298153,
      "grad_norm": 0.7798822407009329,
      "learning_rate": 9.708447322768361e-06,
      "loss": 0.0549,
      "step": 301
    },
    {
      "epoch": 0.39841688654353563,
      "grad_norm": 0.6552064477046319,
      "learning_rate": 9.704559862425101e-06,
      "loss": 0.0652,
      "step": 302
    },
    {
      "epoch": 0.3997361477572559,
      "grad_norm": 0.745247993577414,
      "learning_rate": 9.70064744522937e-06,
      "loss": 0.0602,
      "step": 303
    },
    {
      "epoch": 0.40105540897097625,
      "grad_norm": 0.9204575136344225,
      "learning_rate": 9.696710091935842e-06,
      "loss": 0.0702,
      "step": 304
    },
    {
      "epoch": 0.4023746701846966,
      "grad_norm": 1.1638098645253576,
      "learning_rate": 9.692747823431477e-06,
      "loss": 0.0696,
      "step": 305
    },
    {
      "epoch": 0.40369393139841686,
      "grad_norm": 0.796062028286607,
      "learning_rate": 9.688760660735403e-06,
      "loss": 0.0548,
      "step": 306
    },
    {
      "epoch": 0.4050131926121372,
      "grad_norm": 0.7029833454447884,
      "learning_rate": 9.68474862499881e-06,
      "loss": 0.0612,
      "step": 307
    },
    {
      "epoch": 0.40633245382585753,
      "grad_norm": 0.8614661145454562,
      "learning_rate": 9.680711737504832e-06,
      "loss": 0.072,
      "step": 308
    },
    {
      "epoch": 0.4076517150395778,
      "grad_norm": 0.9475143767728101,
      "learning_rate": 9.676650019668437e-06,
      "loss": 0.06,
      "step": 309
    },
    {
      "epoch": 0.40897097625329815,
      "grad_norm": 0.6977230732332008,
      "learning_rate": 9.672563493036318e-06,
      "loss": 0.0668,
      "step": 310
    },
    {
      "epoch": 0.4102902374670185,
      "grad_norm": 0.7219201569812285,
      "learning_rate": 9.668452179286769e-06,
      "loss": 0.069,
      "step": 311
    },
    {
      "epoch": 0.41160949868073876,
      "grad_norm": 0.8876890045434614,
      "learning_rate": 9.664316100229578e-06,
      "loss": 0.0717,
      "step": 312
    },
    {
      "epoch": 0.4129287598944591,
      "grad_norm": 0.740844881980912,
      "learning_rate": 9.66015527780591e-06,
      "loss": 0.0766,
      "step": 313
    },
    {
      "epoch": 0.41424802110817943,
      "grad_norm": 0.4778241518339464,
      "learning_rate": 9.655969734088184e-06,
      "loss": 0.05,
      "step": 314
    },
    {
      "epoch": 0.4155672823218997,
      "grad_norm": 0.7347326135730948,
      "learning_rate": 9.651759491279966e-06,
      "loss": 0.0669,
      "step": 315
    },
    {
      "epoch": 0.41688654353562005,
      "grad_norm": 1.1284858033400227,
      "learning_rate": 9.647524571715843e-06,
      "loss": 0.0758,
      "step": 316
    },
    {
      "epoch": 0.4182058047493404,
      "grad_norm": 0.6981388862851322,
      "learning_rate": 9.643264997861312e-06,
      "loss": 0.0613,
      "step": 317
    },
    {
      "epoch": 0.41952506596306066,
      "grad_norm": 0.6905887533373853,
      "learning_rate": 9.638980792312651e-06,
      "loss": 0.0569,
      "step": 318
    },
    {
      "epoch": 0.420844327176781,
      "grad_norm": 0.6460144104037705,
      "learning_rate": 9.63467197779681e-06,
      "loss": 0.0579,
      "step": 319
    },
    {
      "epoch": 0.42216358839050133,
      "grad_norm": 0.9556398511536974,
      "learning_rate": 9.630338577171282e-06,
      "loss": 0.0707,
      "step": 320
    },
    {
      "epoch": 0.4234828496042216,
      "grad_norm": 0.5522455835999988,
      "learning_rate": 9.625980613423985e-06,
      "loss": 0.055,
      "step": 321
    },
    {
      "epoch": 0.42480211081794195,
      "grad_norm": 1.7072118244107584,
      "learning_rate": 9.621598109673142e-06,
      "loss": 0.062,
      "step": 322
    },
    {
      "epoch": 0.4261213720316623,
      "grad_norm": 1.1522322988168188,
      "learning_rate": 9.617191089167155e-06,
      "loss": 0.0759,
      "step": 323
    },
    {
      "epoch": 0.42744063324538256,
      "grad_norm": 0.8347923880408826,
      "learning_rate": 9.612759575284483e-06,
      "loss": 0.075,
      "step": 324
    },
    {
      "epoch": 0.4287598944591029,
      "grad_norm": 1.313688139966534,
      "learning_rate": 9.608303591533516e-06,
      "loss": 0.0947,
      "step": 325
    },
    {
      "epoch": 0.43007915567282323,
      "grad_norm": 0.6733121421395389,
      "learning_rate": 9.603823161552459e-06,
      "loss": 0.0716,
      "step": 326
    },
    {
      "epoch": 0.4313984168865435,
      "grad_norm": 1.0558426500828677,
      "learning_rate": 9.599318309109191e-06,
      "loss": 0.0523,
      "step": 327
    },
    {
      "epoch": 0.43271767810026385,
      "grad_norm": 0.9638966469393847,
      "learning_rate": 9.594789058101154e-06,
      "loss": 0.0991,
      "step": 328
    },
    {
      "epoch": 0.4340369393139842,
      "grad_norm": 0.7507598600974192,
      "learning_rate": 9.590235432555219e-06,
      "loss": 0.1072,
      "step": 329
    },
    {
      "epoch": 0.43535620052770446,
      "grad_norm": 0.8808166305074641,
      "learning_rate": 9.585657456627557e-06,
      "loss": 0.1099,
      "step": 330
    },
    {
      "epoch": 0.4366754617414248,
      "grad_norm": 0.7479798613785579,
      "learning_rate": 9.581055154603517e-06,
      "loss": 0.082,
      "step": 331
    },
    {
      "epoch": 0.43799472295514513,
      "grad_norm": 0.6138270782142422,
      "learning_rate": 9.57642855089749e-06,
      "loss": 0.0602,
      "step": 332
    },
    {
      "epoch": 0.4393139841688654,
      "grad_norm": 0.5998863088241339,
      "learning_rate": 9.571777670052786e-06,
      "loss": 0.0669,
      "step": 333
    },
    {
      "epoch": 0.44063324538258575,
      "grad_norm": 1.0086143123332068,
      "learning_rate": 9.567102536741501e-06,
      "loss": 0.0615,
      "step": 334
    },
    {
      "epoch": 0.4419525065963061,
      "grad_norm": 0.8530962897236515,
      "learning_rate": 9.562403175764383e-06,
      "loss": 0.069,
      "step": 335
    },
    {
      "epoch": 0.44327176781002636,
      "grad_norm": 1.1309365259307091,
      "learning_rate": 9.557679612050708e-06,
      "loss": 0.0685,
      "step": 336
    },
    {
      "epoch": 0.4445910290237467,
      "grad_norm": 1.0140895037768696,
      "learning_rate": 9.552931870658136e-06,
      "loss": 0.0947,
      "step": 337
    },
    {
      "epoch": 0.44591029023746703,
      "grad_norm": 0.7953631852283257,
      "learning_rate": 9.548159976772593e-06,
      "loss": 0.0845,
      "step": 338
    },
    {
      "epoch": 0.4472295514511873,
      "grad_norm": 0.7841669983334505,
      "learning_rate": 9.543363955708124e-06,
      "loss": 0.0448,
      "step": 339
    },
    {
      "epoch": 0.44854881266490765,
      "grad_norm": 0.695290594399394,
      "learning_rate": 9.538543832906773e-06,
      "loss": 0.0571,
      "step": 340
    },
    {
      "epoch": 0.449868073878628,
      "grad_norm": 0.8762689209221858,
      "learning_rate": 9.533699633938425e-06,
      "loss": 0.0558,
      "step": 341
    },
    {
      "epoch": 0.45118733509234826,
      "grad_norm": 0.6259802212079215,
      "learning_rate": 9.528831384500699e-06,
      "loss": 0.0561,
      "step": 342
    },
    {
      "epoch": 0.4525065963060686,
      "grad_norm": 0.7451459063031873,
      "learning_rate": 9.523939110418789e-06,
      "loss": 0.0827,
      "step": 343
    },
    {
      "epoch": 0.45382585751978893,
      "grad_norm": 0.6913467923736692,
      "learning_rate": 9.519022837645337e-06,
      "loss": 0.0687,
      "step": 344
    },
    {
      "epoch": 0.4551451187335092,
      "grad_norm": 0.7169029001507153,
      "learning_rate": 9.514082592260299e-06,
      "loss": 0.0621,
      "step": 345
    },
    {
      "epoch": 0.45646437994722955,
      "grad_norm": 0.582058305192275,
      "learning_rate": 9.509118400470792e-06,
      "loss": 0.0575,
      "step": 346
    },
    {
      "epoch": 0.4577836411609499,
      "grad_norm": 0.6825579966883129,
      "learning_rate": 9.504130288610972e-06,
      "loss": 0.0662,
      "step": 347
    },
    {
      "epoch": 0.45910290237467016,
      "grad_norm": 0.5876062939595589,
      "learning_rate": 9.499118283141887e-06,
      "loss": 0.0442,
      "step": 348
    },
    {
      "epoch": 0.4604221635883905,
      "grad_norm": 0.7863259636475942,
      "learning_rate": 9.494082410651331e-06,
      "loss": 0.0774,
      "step": 349
    },
    {
      "epoch": 0.46174142480211083,
      "grad_norm": 0.6211367792942487,
      "learning_rate": 9.48902269785371e-06,
      "loss": 0.0653,
      "step": 350
    },
    {
      "epoch": 0.4630606860158311,
      "grad_norm": 0.6740261695203442,
      "learning_rate": 9.4839391715899e-06,
      "loss": 0.0825,
      "step": 351
    },
    {
      "epoch": 0.46437994722955145,
      "grad_norm": 1.043113360246788,
      "learning_rate": 9.478831858827105e-06,
      "loss": 0.0795,
      "step": 352
    },
    {
      "epoch": 0.4656992084432718,
      "grad_norm": 0.7359044485292054,
      "learning_rate": 9.473700786658706e-06,
      "loss": 0.0726,
      "step": 353
    },
    {
      "epoch": 0.46701846965699206,
      "grad_norm": 0.6909110207632071,
      "learning_rate": 9.468545982304132e-06,
      "loss": 0.0731,
      "step": 354
    },
    {
      "epoch": 0.4683377308707124,
      "grad_norm": 0.5814197718154829,
      "learning_rate": 9.463367473108697e-06,
      "loss": 0.0613,
      "step": 355
    },
    {
      "epoch": 0.46965699208443273,
      "grad_norm": 0.9262695641177886,
      "learning_rate": 9.458165286543477e-06,
      "loss": 0.0952,
      "step": 356
    },
    {
      "epoch": 0.470976253298153,
      "grad_norm": 0.6127534132821949,
      "learning_rate": 9.452939450205139e-06,
      "loss": 0.0727,
      "step": 357
    },
    {
      "epoch": 0.47229551451187335,
      "grad_norm": 0.5642022396225125,
      "learning_rate": 9.447689991815819e-06,
      "loss": 0.061,
      "step": 358
    },
    {
      "epoch": 0.4736147757255937,
      "grad_norm": 0.5652422361321967,
      "learning_rate": 9.442416939222956e-06,
      "loss": 0.0659,
      "step": 359
    },
    {
      "epoch": 0.47493403693931396,
      "grad_norm": 0.6652258751925895,
      "learning_rate": 9.437120320399158e-06,
      "loss": 0.0965,
      "step": 360
    },
    {
      "epoch": 0.4762532981530343,
      "grad_norm": 0.6901963164548323,
      "learning_rate": 9.431800163442043e-06,
      "loss": 0.0692,
      "step": 361
    },
    {
      "epoch": 0.47757255936675463,
      "grad_norm": 0.571898610656191,
      "learning_rate": 9.426456496574095e-06,
      "loss": 0.0509,
      "step": 362
    },
    {
      "epoch": 0.4788918205804749,
      "grad_norm": 0.5449257166951276,
      "learning_rate": 9.421089348142519e-06,
      "loss": 0.0411,
      "step": 363
    },
    {
      "epoch": 0.48021108179419525,
      "grad_norm": 0.9021664212076399,
      "learning_rate": 9.41569874661908e-06,
      "loss": 0.0785,
      "step": 364
    },
    {
      "epoch": 0.4815303430079156,
      "grad_norm": 0.8750083598482614,
      "learning_rate": 9.410284720599958e-06,
      "loss": 0.0867,
      "step": 365
    },
    {
      "epoch": 0.48284960422163586,
      "grad_norm": 0.6011018833711832,
      "learning_rate": 9.4048472988056e-06,
      "loss": 0.068,
      "step": 366
    },
    {
      "epoch": 0.4841688654353562,
      "grad_norm": 0.7520079528217786,
      "learning_rate": 9.39938651008056e-06,
      "loss": 0.0822,
      "step": 367
    },
    {
      "epoch": 0.48548812664907653,
      "grad_norm": 0.6250727259259536,
      "learning_rate": 9.393902383393347e-06,
      "loss": 0.0741,
      "step": 368
    },
    {
      "epoch": 0.4868073878627968,
      "grad_norm": 0.6953888116558605,
      "learning_rate": 9.388394947836278e-06,
      "loss": 0.058,
      "step": 369
    },
    {
      "epoch": 0.48812664907651715,
      "grad_norm": 0.7831220569741709,
      "learning_rate": 9.382864232625321e-06,
      "loss": 0.0747,
      "step": 370
    },
    {
      "epoch": 0.4894459102902375,
      "grad_norm": 0.8773155442663596,
      "learning_rate": 9.377310267099932e-06,
      "loss": 0.0943,
      "step": 371
    },
    {
      "epoch": 0.49076517150395776,
      "grad_norm": 0.7336952231337915,
      "learning_rate": 9.371733080722911e-06,
      "loss": 0.0553,
      "step": 372
    },
    {
      "epoch": 0.4920844327176781,
      "grad_norm": 0.8627759942056888,
      "learning_rate": 9.366132703080239e-06,
      "loss": 0.0574,
      "step": 373
    },
    {
      "epoch": 0.49340369393139843,
      "grad_norm": 0.8103532619274963,
      "learning_rate": 9.36050916388092e-06,
      "loss": 0.0554,
      "step": 374
    },
    {
      "epoch": 0.4947229551451187,
      "grad_norm": 0.741887593849594,
      "learning_rate": 9.354862492956831e-06,
      "loss": 0.069,
      "step": 375
    },
    {
      "epoch": 0.49604221635883905,
      "grad_norm": 0.7855112695865729,
      "learning_rate": 9.349192720262556e-06,
      "loss": 0.0661,
      "step": 376
    },
    {
      "epoch": 0.4973614775725594,
      "grad_norm": 0.6866863581677422,
      "learning_rate": 9.343499875875226e-06,
      "loss": 0.0594,
      "step": 377
    },
    {
      "epoch": 0.49868073878627966,
      "grad_norm": 0.8782759603345595,
      "learning_rate": 9.337783989994371e-06,
      "loss": 0.0772,
      "step": 378
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9549895199126721,
      "learning_rate": 9.332045092941747e-06,
      "loss": 0.068,
      "step": 379
    },
    {
      "epoch": 0.5013192612137203,
      "grad_norm": 0.5448660341904454,
      "learning_rate": 9.326283215161177e-06,
      "loss": 0.0667,
      "step": 380
    },
    {
      "epoch": 0.5026385224274407,
      "grad_norm": 0.5985824770826613,
      "learning_rate": 9.320498387218404e-06,
      "loss": 0.0664,
      "step": 381
    },
    {
      "epoch": 0.503957783641161,
      "grad_norm": 0.752830871271958,
      "learning_rate": 9.314690639800906e-06,
      "loss": 0.0645,
      "step": 382
    },
    {
      "epoch": 0.5052770448548812,
      "grad_norm": 0.5477776553953697,
      "learning_rate": 9.308860003717748e-06,
      "loss": 0.0603,
      "step": 383
    },
    {
      "epoch": 0.5065963060686016,
      "grad_norm": 0.6139407974821701,
      "learning_rate": 9.30300650989942e-06,
      "loss": 0.0877,
      "step": 384
    },
    {
      "epoch": 0.5079155672823219,
      "grad_norm": 0.6039560183881748,
      "learning_rate": 9.297130189397661e-06,
      "loss": 0.0519,
      "step": 385
    },
    {
      "epoch": 0.5092348284960422,
      "grad_norm": 0.6664238974836402,
      "learning_rate": 9.291231073385306e-06,
      "loss": 0.0809,
      "step": 386
    },
    {
      "epoch": 0.5105540897097626,
      "grad_norm": 0.6488779162272155,
      "learning_rate": 9.285309193156118e-06,
      "loss": 0.0745,
      "step": 387
    },
    {
      "epoch": 0.5118733509234829,
      "grad_norm": 0.8710463478016437,
      "learning_rate": 9.279364580124615e-06,
      "loss": 0.0653,
      "step": 388
    },
    {
      "epoch": 0.5131926121372031,
      "grad_norm": 0.7282940066695746,
      "learning_rate": 9.273397265825909e-06,
      "loss": 0.0706,
      "step": 389
    },
    {
      "epoch": 0.5145118733509235,
      "grad_norm": 0.6377222079463343,
      "learning_rate": 9.267407281915541e-06,
      "loss": 0.0566,
      "step": 390
    },
    {
      "epoch": 0.5158311345646438,
      "grad_norm": 0.5717894561053849,
      "learning_rate": 9.261394660169311e-06,
      "loss": 0.0481,
      "step": 391
    },
    {
      "epoch": 0.5171503957783641,
      "grad_norm": 0.6051053509035249,
      "learning_rate": 9.255359432483106e-06,
      "loss": 0.042,
      "step": 392
    },
    {
      "epoch": 0.5184696569920845,
      "grad_norm": 1.0265587669123954,
      "learning_rate": 9.249301630872735e-06,
      "loss": 0.0728,
      "step": 393
    },
    {
      "epoch": 0.5197889182058048,
      "grad_norm": 0.9466700337352391,
      "learning_rate": 9.243221287473755e-06,
      "loss": 0.0736,
      "step": 394
    },
    {
      "epoch": 0.521108179419525,
      "grad_norm": 1.1721998085842662,
      "learning_rate": 9.237118434541308e-06,
      "loss": 0.0639,
      "step": 395
    },
    {
      "epoch": 0.5224274406332454,
      "grad_norm": 0.964741992189812,
      "learning_rate": 9.23099310444994e-06,
      "loss": 0.0621,
      "step": 396
    },
    {
      "epoch": 0.5237467018469657,
      "grad_norm": 0.6970255221602126,
      "learning_rate": 9.224845329693434e-06,
      "loss": 0.079,
      "step": 397
    },
    {
      "epoch": 0.525065963060686,
      "grad_norm": 1.3823551560613605,
      "learning_rate": 9.218675142884648e-06,
      "loss": 0.0795,
      "step": 398
    },
    {
      "epoch": 0.5263852242744064,
      "grad_norm": 1.4558190810564504,
      "learning_rate": 9.212482576755318e-06,
      "loss": 0.067,
      "step": 399
    },
    {
      "epoch": 0.5277044854881267,
      "grad_norm": 0.6341626113530675,
      "learning_rate": 9.206267664155906e-06,
      "loss": 0.0647,
      "step": 400
    },
    {
      "epoch": 0.5290237467018469,
      "grad_norm": 0.7316499535743366,
      "learning_rate": 9.20003043805542e-06,
      "loss": 0.0535,
      "step": 401
    },
    {
      "epoch": 0.5303430079155673,
      "grad_norm": 0.7680943343177051,
      "learning_rate": 9.19377093154123e-06,
      "loss": 0.0679,
      "step": 402
    },
    {
      "epoch": 0.5316622691292876,
      "grad_norm": 1.2180641513033668,
      "learning_rate": 9.187489177818907e-06,
      "loss": 0.0716,
      "step": 403
    },
    {
      "epoch": 0.5329815303430079,
      "grad_norm": 0.7701251269926012,
      "learning_rate": 9.181185210212034e-06,
      "loss": 0.0826,
      "step": 404
    },
    {
      "epoch": 0.5343007915567283,
      "grad_norm": 1.0054417805604758,
      "learning_rate": 9.174859062162037e-06,
      "loss": 0.0723,
      "step": 405
    },
    {
      "epoch": 0.5356200527704486,
      "grad_norm": 0.9830147127682102,
      "learning_rate": 9.168510767228008e-06,
      "loss": 0.071,
      "step": 406
    },
    {
      "epoch": 0.5369393139841688,
      "grad_norm": 0.6662121630599811,
      "learning_rate": 9.162140359086515e-06,
      "loss": 0.0562,
      "step": 407
    },
    {
      "epoch": 0.5382585751978892,
      "grad_norm": 0.8657347202295601,
      "learning_rate": 9.155747871531444e-06,
      "loss": 0.0699,
      "step": 408
    },
    {
      "epoch": 0.5395778364116095,
      "grad_norm": 0.8713898850191355,
      "learning_rate": 9.149333338473803e-06,
      "loss": 0.0526,
      "step": 409
    },
    {
      "epoch": 0.5408970976253298,
      "grad_norm": 0.8321596239400251,
      "learning_rate": 9.142896793941546e-06,
      "loss": 0.0743,
      "step": 410
    },
    {
      "epoch": 0.5422163588390502,
      "grad_norm": 0.6268650641625824,
      "learning_rate": 9.136438272079397e-06,
      "loss": 0.0449,
      "step": 411
    },
    {
      "epoch": 0.5435356200527705,
      "grad_norm": 1.0046732601631165,
      "learning_rate": 9.129957807148666e-06,
      "loss": 0.069,
      "step": 412
    },
    {
      "epoch": 0.5448548812664907,
      "grad_norm": 0.7417313448011015,
      "learning_rate": 9.123455433527063e-06,
      "loss": 0.0605,
      "step": 413
    },
    {
      "epoch": 0.5461741424802111,
      "grad_norm": 0.5862198883042264,
      "learning_rate": 9.116931185708523e-06,
      "loss": 0.0565,
      "step": 414
    },
    {
      "epoch": 0.5474934036939314,
      "grad_norm": 0.5663121991718124,
      "learning_rate": 9.110385098303021e-06,
      "loss": 0.0626,
      "step": 415
    },
    {
      "epoch": 0.5488126649076517,
      "grad_norm": 0.513372498718201,
      "learning_rate": 9.103817206036383e-06,
      "loss": 0.0501,
      "step": 416
    },
    {
      "epoch": 0.5501319261213721,
      "grad_norm": 0.7663231048740375,
      "learning_rate": 9.097227543750109e-06,
      "loss": 0.0952,
      "step": 417
    },
    {
      "epoch": 0.5514511873350924,
      "grad_norm": 0.6760552525944711,
      "learning_rate": 9.090616146401183e-06,
      "loss": 0.0577,
      "step": 418
    },
    {
      "epoch": 0.5527704485488126,
      "grad_norm": 0.5317840409418032,
      "learning_rate": 9.083983049061893e-06,
      "loss": 0.0469,
      "step": 419
    },
    {
      "epoch": 0.554089709762533,
      "grad_norm": 0.8851719428502758,
      "learning_rate": 9.077328286919638e-06,
      "loss": 0.0617,
      "step": 420
    },
    {
      "epoch": 0.5554089709762533,
      "grad_norm": 0.7497996032541763,
      "learning_rate": 9.070651895276748e-06,
      "loss": 0.0714,
      "step": 421
    },
    {
      "epoch": 0.5567282321899736,
      "grad_norm": 0.6498492053332823,
      "learning_rate": 9.063953909550289e-06,
      "loss": 0.0538,
      "step": 422
    },
    {
      "epoch": 0.558047493403694,
      "grad_norm": 0.44622429830036786,
      "learning_rate": 9.057234365271886e-06,
      "loss": 0.0521,
      "step": 423
    },
    {
      "epoch": 0.5593667546174143,
      "grad_norm": 0.5512561169864484,
      "learning_rate": 9.050493298087523e-06,
      "loss": 0.054,
      "step": 424
    },
    {
      "epoch": 0.5606860158311345,
      "grad_norm": 0.631404862974141,
      "learning_rate": 9.043730743757362e-06,
      "loss": 0.0764,
      "step": 425
    },
    {
      "epoch": 0.5620052770448549,
      "grad_norm": 0.5870072097930791,
      "learning_rate": 9.036946738155548e-06,
      "loss": 0.0595,
      "step": 426
    },
    {
      "epoch": 0.5633245382585752,
      "grad_norm": 0.5040142704734094,
      "learning_rate": 9.030141317270026e-06,
      "loss": 0.0556,
      "step": 427
    },
    {
      "epoch": 0.5646437994722955,
      "grad_norm": 0.7403758206591379,
      "learning_rate": 9.023314517202341e-06,
      "loss": 0.0655,
      "step": 428
    },
    {
      "epoch": 0.5659630606860159,
      "grad_norm": 0.6611426759437872,
      "learning_rate": 9.016466374167451e-06,
      "loss": 0.0667,
      "step": 429
    },
    {
      "epoch": 0.5672823218997362,
      "grad_norm": 0.9850776098945953,
      "learning_rate": 9.009596924493536e-06,
      "loss": 0.0742,
      "step": 430
    },
    {
      "epoch": 0.5686015831134564,
      "grad_norm": 0.6785243477097369,
      "learning_rate": 9.002706204621802e-06,
      "loss": 0.0619,
      "step": 431
    },
    {
      "epoch": 0.5699208443271768,
      "grad_norm": 0.7190003125330587,
      "learning_rate": 8.995794251106295e-06,
      "loss": 0.0858,
      "step": 432
    },
    {
      "epoch": 0.5712401055408971,
      "grad_norm": 0.779155867242513,
      "learning_rate": 8.988861100613695e-06,
      "loss": 0.0894,
      "step": 433
    },
    {
      "epoch": 0.5725593667546174,
      "grad_norm": 0.821896375590076,
      "learning_rate": 8.98190678992313e-06,
      "loss": 0.0629,
      "step": 434
    },
    {
      "epoch": 0.5738786279683378,
      "grad_norm": 0.5076050481256884,
      "learning_rate": 8.974931355925983e-06,
      "loss": 0.0481,
      "step": 435
    },
    {
      "epoch": 0.575197889182058,
      "grad_norm": 0.5488701803801146,
      "learning_rate": 8.96793483562569e-06,
      "loss": 0.0495,
      "step": 436
    },
    {
      "epoch": 0.5765171503957783,
      "grad_norm": 0.4926498135199434,
      "learning_rate": 8.96091726613754e-06,
      "loss": 0.051,
      "step": 437
    },
    {
      "epoch": 0.5778364116094987,
      "grad_norm": 0.7305868797856492,
      "learning_rate": 8.953878684688492e-06,
      "loss": 0.0781,
      "step": 438
    },
    {
      "epoch": 0.579155672823219,
      "grad_norm": 0.7928116961537923,
      "learning_rate": 8.946819128616973e-06,
      "loss": 0.0798,
      "step": 439
    },
    {
      "epoch": 0.5804749340369393,
      "grad_norm": 0.8287924226908049,
      "learning_rate": 8.939738635372664e-06,
      "loss": 0.05,
      "step": 440
    },
    {
      "epoch": 0.5817941952506597,
      "grad_norm": 0.5086476050371508,
      "learning_rate": 8.932637242516325e-06,
      "loss": 0.0484,
      "step": 441
    },
    {
      "epoch": 0.58311345646438,
      "grad_norm": 0.44896912424993907,
      "learning_rate": 8.92551498771958e-06,
      "loss": 0.0535,
      "step": 442
    },
    {
      "epoch": 0.5844327176781002,
      "grad_norm": 0.7427267049471321,
      "learning_rate": 8.91837190876472e-06,
      "loss": 0.0448,
      "step": 443
    },
    {
      "epoch": 0.5857519788918206,
      "grad_norm": 0.4077238431271433,
      "learning_rate": 8.911208043544513e-06,
      "loss": 0.0404,
      "step": 444
    },
    {
      "epoch": 0.5870712401055409,
      "grad_norm": 0.8279103277542957,
      "learning_rate": 8.904023430061981e-06,
      "loss": 0.0839,
      "step": 445
    },
    {
      "epoch": 0.5883905013192612,
      "grad_norm": 0.5893492543117536,
      "learning_rate": 8.896818106430225e-06,
      "loss": 0.0607,
      "step": 446
    },
    {
      "epoch": 0.5897097625329816,
      "grad_norm": 0.6819067552196123,
      "learning_rate": 8.889592110872203e-06,
      "loss": 0.0674,
      "step": 447
    },
    {
      "epoch": 0.5910290237467019,
      "grad_norm": 0.4468129220866856,
      "learning_rate": 8.882345481720533e-06,
      "loss": 0.0404,
      "step": 448
    },
    {
      "epoch": 0.5923482849604221,
      "grad_norm": 0.595551681922815,
      "learning_rate": 8.875078257417294e-06,
      "loss": 0.0801,
      "step": 449
    },
    {
      "epoch": 0.5936675461741425,
      "grad_norm": 0.9018494496257856,
      "learning_rate": 8.867790476513818e-06,
      "loss": 0.0665,
      "step": 450
    },
    {
      "epoch": 0.5949868073878628,
      "grad_norm": 0.5943979685711214,
      "learning_rate": 8.860482177670482e-06,
      "loss": 0.0461,
      "step": 451
    },
    {
      "epoch": 0.5963060686015831,
      "grad_norm": 0.4301614688278578,
      "learning_rate": 8.853153399656513e-06,
      "loss": 0.0391,
      "step": 452
    },
    {
      "epoch": 0.5976253298153035,
      "grad_norm": 0.6967315070272085,
      "learning_rate": 8.845804181349773e-06,
      "loss": 0.077,
      "step": 453
    },
    {
      "epoch": 0.5989445910290238,
      "grad_norm": 0.4953833355499116,
      "learning_rate": 8.838434561736556e-06,
      "loss": 0.0566,
      "step": 454
    },
    {
      "epoch": 0.600263852242744,
      "grad_norm": 0.5907548455305766,
      "learning_rate": 8.831044579911383e-06,
      "loss": 0.0655,
      "step": 455
    },
    {
      "epoch": 0.6015831134564644,
      "grad_norm": 0.6236652449819667,
      "learning_rate": 8.823634275076792e-06,
      "loss": 0.0554,
      "step": 456
    },
    {
      "epoch": 0.6029023746701847,
      "grad_norm": 0.4168923412952401,
      "learning_rate": 8.816203686543128e-06,
      "loss": 0.0337,
      "step": 457
    },
    {
      "epoch": 0.604221635883905,
      "grad_norm": 0.48071330924907624,
      "learning_rate": 8.808752853728341e-06,
      "loss": 0.0492,
      "step": 458
    },
    {
      "epoch": 0.6055408970976254,
      "grad_norm": 0.4842702889425807,
      "learning_rate": 8.801281816157776e-06,
      "loss": 0.0515,
      "step": 459
    },
    {
      "epoch": 0.6068601583113457,
      "grad_norm": 0.7877007315164113,
      "learning_rate": 8.793790613463956e-06,
      "loss": 0.072,
      "step": 460
    },
    {
      "epoch": 0.6081794195250659,
      "grad_norm": 0.4597573113919865,
      "learning_rate": 8.786279285386374e-06,
      "loss": 0.0507,
      "step": 461
    },
    {
      "epoch": 0.6094986807387863,
      "grad_norm": 0.7295743451125444,
      "learning_rate": 8.778747871771293e-06,
      "loss": 0.092,
      "step": 462
    },
    {
      "epoch": 0.6108179419525066,
      "grad_norm": 0.6380120848265783,
      "learning_rate": 8.771196412571516e-06,
      "loss": 0.0754,
      "step": 463
    },
    {
      "epoch": 0.6121372031662269,
      "grad_norm": 0.6151625612709314,
      "learning_rate": 8.763624947846195e-06,
      "loss": 0.0492,
      "step": 464
    },
    {
      "epoch": 0.6134564643799473,
      "grad_norm": 0.6104995830766858,
      "learning_rate": 8.756033517760601e-06,
      "loss": 0.0594,
      "step": 465
    },
    {
      "epoch": 0.6147757255936676,
      "grad_norm": 0.4662314682875708,
      "learning_rate": 8.748422162585915e-06,
      "loss": 0.0405,
      "step": 466
    },
    {
      "epoch": 0.6160949868073878,
      "grad_norm": 0.7037103770745874,
      "learning_rate": 8.740790922699024e-06,
      "loss": 0.0681,
      "step": 467
    },
    {
      "epoch": 0.6174142480211082,
      "grad_norm": 0.46215628548577664,
      "learning_rate": 8.733139838582299e-06,
      "loss": 0.0396,
      "step": 468
    },
    {
      "epoch": 0.6187335092348285,
      "grad_norm": 0.5102031881535187,
      "learning_rate": 8.725468950823378e-06,
      "loss": 0.0515,
      "step": 469
    },
    {
      "epoch": 0.6200527704485488,
      "grad_norm": 0.4651826535622034,
      "learning_rate": 8.717778300114952e-06,
      "loss": 0.0419,
      "step": 470
    },
    {
      "epoch": 0.6213720316622692,
      "grad_norm": 1.1325412075126133,
      "learning_rate": 8.710067927254555e-06,
      "loss": 0.0792,
      "step": 471
    },
    {
      "epoch": 0.6226912928759895,
      "grad_norm": 0.45867525224339245,
      "learning_rate": 8.702337873144343e-06,
      "loss": 0.0492,
      "step": 472
    },
    {
      "epoch": 0.6240105540897097,
      "grad_norm": 0.5504256616903217,
      "learning_rate": 8.694588178790876e-06,
      "loss": 0.0392,
      "step": 473
    },
    {
      "epoch": 0.6253298153034301,
      "grad_norm": 0.9430488090364006,
      "learning_rate": 8.686818885304907e-06,
      "loss": 0.0693,
      "step": 474
    },
    {
      "epoch": 0.6266490765171504,
      "grad_norm": 0.46564772689942074,
      "learning_rate": 8.679030033901148e-06,
      "loss": 0.0528,
      "step": 475
    },
    {
      "epoch": 0.6279683377308707,
      "grad_norm": 0.6909276166774045,
      "learning_rate": 8.671221665898074e-06,
      "loss": 0.0705,
      "step": 476
    },
    {
      "epoch": 0.6292875989445911,
      "grad_norm": 0.8418254861723139,
      "learning_rate": 8.663393822717686e-06,
      "loss": 0.0593,
      "step": 477
    },
    {
      "epoch": 0.6306068601583114,
      "grad_norm": 0.6295040305026375,
      "learning_rate": 8.655546545885294e-06,
      "loss": 0.0634,
      "step": 478
    },
    {
      "epoch": 0.6319261213720316,
      "grad_norm": 0.6645364650357778,
      "learning_rate": 8.64767987702931e-06,
      "loss": 0.0476,
      "step": 479
    },
    {
      "epoch": 0.633245382585752,
      "grad_norm": 0.6135679991723683,
      "learning_rate": 8.63979385788101e-06,
      "loss": 0.0454,
      "step": 480
    },
    {
      "epoch": 0.6345646437994723,
      "grad_norm": 0.812727393806555,
      "learning_rate": 8.631888530274321e-06,
      "loss": 0.0791,
      "step": 481
    },
    {
      "epoch": 0.6358839050131926,
      "grad_norm": 0.5166039238114825,
      "learning_rate": 8.6239639361456e-06,
      "loss": 0.0489,
      "step": 482
    },
    {
      "epoch": 0.637203166226913,
      "grad_norm": 0.6158141240963521,
      "learning_rate": 8.616020117533406e-06,
      "loss": 0.0428,
      "step": 483
    },
    {
      "epoch": 0.6385224274406333,
      "grad_norm": 0.8250113594001276,
      "learning_rate": 8.608057116578283e-06,
      "loss": 0.0712,
      "step": 484
    },
    {
      "epoch": 0.6398416886543535,
      "grad_norm": 0.7451289676544177,
      "learning_rate": 8.600074975522534e-06,
      "loss": 0.06,
      "step": 485
    },
    {
      "epoch": 0.6411609498680739,
      "grad_norm": 0.5326398128654674,
      "learning_rate": 8.592073736709996e-06,
      "loss": 0.0603,
      "step": 486
    },
    {
      "epoch": 0.6424802110817942,
      "grad_norm": 0.5021499758399166,
      "learning_rate": 8.584053442585816e-06,
      "loss": 0.0505,
      "step": 487
    },
    {
      "epoch": 0.6437994722955145,
      "grad_norm": 0.6708410508573724,
      "learning_rate": 8.576014135696227e-06,
      "loss": 0.0507,
      "step": 488
    },
    {
      "epoch": 0.6451187335092349,
      "grad_norm": 1.58512995612873,
      "learning_rate": 8.567955858688319e-06,
      "loss": 0.0901,
      "step": 489
    },
    {
      "epoch": 0.6464379947229552,
      "grad_norm": 0.9243327973050967,
      "learning_rate": 8.559878654309818e-06,
      "loss": 0.0812,
      "step": 490
    },
    {
      "epoch": 0.6477572559366754,
      "grad_norm": 0.575309292027881,
      "learning_rate": 8.551782565408857e-06,
      "loss": 0.0504,
      "step": 491
    },
    {
      "epoch": 0.6490765171503958,
      "grad_norm": 0.6994300970953274,
      "learning_rate": 8.543667634933743e-06,
      "loss": 0.0764,
      "step": 492
    },
    {
      "epoch": 0.6503957783641161,
      "grad_norm": 0.7106135291337119,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.0744,
      "step": 493
    },
    {
      "epoch": 0.6517150395778364,
      "grad_norm": 1.1670675909226174,
      "learning_rate": 8.52738142155383e-06,
      "loss": 0.0592,
      "step": 494
    },
    {
      "epoch": 0.6530343007915568,
      "grad_norm": 0.6123285898097686,
      "learning_rate": 8.519210225044491e-06,
      "loss": 0.0491,
      "step": 495
    },
    {
      "epoch": 0.6543535620052771,
      "grad_norm": 0.5445510412762957,
      "learning_rate": 8.511020359751467e-06,
      "loss": 0.0606,
      "step": 496
    },
    {
      "epoch": 0.6556728232189973,
      "grad_norm": 0.5133147307482108,
      "learning_rate": 8.502811869120537e-06,
      "loss": 0.0555,
      "step": 497
    },
    {
      "epoch": 0.6569920844327177,
      "grad_norm": 0.5722478197244066,
      "learning_rate": 8.49458479669628e-06,
      "loss": 0.057,
      "step": 498
    },
    {
      "epoch": 0.658311345646438,
      "grad_norm": 0.49913744387906306,
      "learning_rate": 8.486339186121852e-06,
      "loss": 0.0557,
      "step": 499
    },
    {
      "epoch": 0.6596306068601583,
      "grad_norm": 0.448299138636958,
      "learning_rate": 8.478075081138746e-06,
      "loss": 0.0501,
      "step": 500
    },
    {
      "epoch": 0.6609498680738787,
      "grad_norm": 0.45827849751503286,
      "learning_rate": 8.46979252558657e-06,
      "loss": 0.0444,
      "step": 501
    },
    {
      "epoch": 0.662269129287599,
      "grad_norm": 0.40298508876722344,
      "learning_rate": 8.461491563402807e-06,
      "loss": 0.0517,
      "step": 502
    },
    {
      "epoch": 0.6635883905013192,
      "grad_norm": 0.9725533609650583,
      "learning_rate": 8.453172238622582e-06,
      "loss": 0.0578,
      "step": 503
    },
    {
      "epoch": 0.6649076517150396,
      "grad_norm": 2.3295313624861302,
      "learning_rate": 8.444834595378434e-06,
      "loss": 0.0843,
      "step": 504
    },
    {
      "epoch": 0.6662269129287599,
      "grad_norm": 0.6248802177290632,
      "learning_rate": 8.436478677900073e-06,
      "loss": 0.0764,
      "step": 505
    },
    {
      "epoch": 0.6675461741424802,
      "grad_norm": 0.7879242217892278,
      "learning_rate": 8.428104530514156e-06,
      "loss": 0.0578,
      "step": 506
    },
    {
      "epoch": 0.6688654353562006,
      "grad_norm": 0.723177780469888,
      "learning_rate": 8.419712197644042e-06,
      "loss": 0.0565,
      "step": 507
    },
    {
      "epoch": 0.6701846965699209,
      "grad_norm": 0.7752616672345674,
      "learning_rate": 8.411301723809563e-06,
      "loss": 0.0611,
      "step": 508
    },
    {
      "epoch": 0.6715039577836411,
      "grad_norm": 0.5940661722658356,
      "learning_rate": 8.402873153626787e-06,
      "loss": 0.0468,
      "step": 509
    },
    {
      "epoch": 0.6728232189973615,
      "grad_norm": 0.5920695638139996,
      "learning_rate": 8.394426531807777e-06,
      "loss": 0.0534,
      "step": 510
    },
    {
      "epoch": 0.6741424802110818,
      "grad_norm": 0.7901406180432282,
      "learning_rate": 8.38596190316036e-06,
      "loss": 0.0744,
      "step": 511
    },
    {
      "epoch": 0.6754617414248021,
      "grad_norm": 0.8264412056071058,
      "learning_rate": 8.37747931258788e-06,
      "loss": 0.0664,
      "step": 512
    },
    {
      "epoch": 0.6767810026385225,
      "grad_norm": 0.5451316404401848,
      "learning_rate": 8.368978805088972e-06,
      "loss": 0.0428,
      "step": 513
    },
    {
      "epoch": 0.6781002638522428,
      "grad_norm": 0.6935044264705226,
      "learning_rate": 8.360460425757316e-06,
      "loss": 0.0567,
      "step": 514
    },
    {
      "epoch": 0.679419525065963,
      "grad_norm": 1.3586667420447256,
      "learning_rate": 8.351924219781393e-06,
      "loss": 0.0584,
      "step": 515
    },
    {
      "epoch": 0.6807387862796834,
      "grad_norm": 0.4889504771837496,
      "learning_rate": 8.34337023244426e-06,
      "loss": 0.0558,
      "step": 516
    },
    {
      "epoch": 0.6820580474934037,
      "grad_norm": 0.4663280409333404,
      "learning_rate": 8.3347985091233e-06,
      "loss": 0.0524,
      "step": 517
    },
    {
      "epoch": 0.683377308707124,
      "grad_norm": 0.6536313802749268,
      "learning_rate": 8.326209095289973e-06,
      "loss": 0.0708,
      "step": 518
    },
    {
      "epoch": 0.6846965699208444,
      "grad_norm": 0.5460048593573863,
      "learning_rate": 8.31760203650959e-06,
      "loss": 0.0574,
      "step": 519
    },
    {
      "epoch": 0.6860158311345647,
      "grad_norm": 0.5057746693426484,
      "learning_rate": 8.308977378441072e-06,
      "loss": 0.0535,
      "step": 520
    },
    {
      "epoch": 0.6873350923482849,
      "grad_norm": 0.5246741406807683,
      "learning_rate": 8.300335166836688e-06,
      "loss": 0.061,
      "step": 521
    },
    {
      "epoch": 0.6886543535620053,
      "grad_norm": 0.7714492708651828,
      "learning_rate": 8.291675447541834e-06,
      "loss": 0.0757,
      "step": 522
    },
    {
      "epoch": 0.6899736147757256,
      "grad_norm": 0.47313909010317895,
      "learning_rate": 8.282998266494781e-06,
      "loss": 0.0652,
      "step": 523
    },
    {
      "epoch": 0.6912928759894459,
      "grad_norm": 1.3924325907131665,
      "learning_rate": 8.274303669726427e-06,
      "loss": 0.0732,
      "step": 524
    },
    {
      "epoch": 0.6926121372031663,
      "grad_norm": 1.447236862207532,
      "learning_rate": 8.26559170336006e-06,
      "loss": 0.0932,
      "step": 525
    },
    {
      "epoch": 0.6939313984168866,
      "grad_norm": 0.7203129772030047,
      "learning_rate": 8.256862413611113e-06,
      "loss": 0.0671,
      "step": 526
    },
    {
      "epoch": 0.6952506596306068,
      "grad_norm": 0.5928412775645093,
      "learning_rate": 8.24811584678691e-06,
      "loss": 0.0667,
      "step": 527
    },
    {
      "epoch": 0.6965699208443272,
      "grad_norm": 0.43230530042462856,
      "learning_rate": 8.239352049286435e-06,
      "loss": 0.0398,
      "step": 528
    },
    {
      "epoch": 0.6978891820580475,
      "grad_norm": 0.5428941662684312,
      "learning_rate": 8.230571067600071e-06,
      "loss": 0.0499,
      "step": 529
    },
    {
      "epoch": 0.6992084432717678,
      "grad_norm": 0.6009844543596541,
      "learning_rate": 8.221772948309363e-06,
      "loss": 0.0486,
      "step": 530
    },
    {
      "epoch": 0.7005277044854882,
      "grad_norm": 0.6214551055901713,
      "learning_rate": 8.212957738086766e-06,
      "loss": 0.0586,
      "step": 531
    },
    {
      "epoch": 0.7018469656992085,
      "grad_norm": 0.45273005517181125,
      "learning_rate": 8.204125483695403e-06,
      "loss": 0.0482,
      "step": 532
    },
    {
      "epoch": 0.7031662269129287,
      "grad_norm": 0.5258680926988596,
      "learning_rate": 8.195276231988811e-06,
      "loss": 0.06,
      "step": 533
    },
    {
      "epoch": 0.7044854881266491,
      "grad_norm": 0.5853091007879238,
      "learning_rate": 8.186410029910694e-06,
      "loss": 0.0597,
      "step": 534
    },
    {
      "epoch": 0.7058047493403694,
      "grad_norm": 0.8812261465136906,
      "learning_rate": 8.177526924494675e-06,
      "loss": 0.064,
      "step": 535
    },
    {
      "epoch": 0.7071240105540897,
      "grad_norm": 0.4749522463294217,
      "learning_rate": 8.168626962864045e-06,
      "loss": 0.0499,
      "step": 536
    },
    {
      "epoch": 0.7084432717678101,
      "grad_norm": 0.4848789823718647,
      "learning_rate": 8.15971019223152e-06,
      "loss": 0.0474,
      "step": 537
    },
    {
      "epoch": 0.7097625329815304,
      "grad_norm": 0.623357361345339,
      "learning_rate": 8.15077665989898e-06,
      "loss": 0.065,
      "step": 538
    },
    {
      "epoch": 0.7110817941952506,
      "grad_norm": 0.6147503017541885,
      "learning_rate": 8.14182641325722e-06,
      "loss": 0.0645,
      "step": 539
    },
    {
      "epoch": 0.712401055408971,
      "grad_norm": 0.7159205067722418,
      "learning_rate": 8.132859499785708e-06,
      "loss": 0.0622,
      "step": 540
    },
    {
      "epoch": 0.7137203166226913,
      "grad_norm": 0.601001813393786,
      "learning_rate": 8.123875967052324e-06,
      "loss": 0.0667,
      "step": 541
    },
    {
      "epoch": 0.7150395778364116,
      "grad_norm": 0.661382907823683,
      "learning_rate": 8.114875862713107e-06,
      "loss": 0.0477,
      "step": 542
    },
    {
      "epoch": 0.716358839050132,
      "grad_norm": 0.6808774589369142,
      "learning_rate": 8.105859234512011e-06,
      "loss": 0.0728,
      "step": 543
    },
    {
      "epoch": 0.7176781002638523,
      "grad_norm": 0.5645330972316417,
      "learning_rate": 8.09682613028064e-06,
      "loss": 0.0442,
      "step": 544
    },
    {
      "epoch": 0.7189973614775725,
      "grad_norm": 0.7000610285805919,
      "learning_rate": 8.087776597938005e-06,
      "loss": 0.0607,
      "step": 545
    },
    {
      "epoch": 0.7203166226912929,
      "grad_norm": 0.9717949396233259,
      "learning_rate": 8.078710685490266e-06,
      "loss": 0.0925,
      "step": 546
    },
    {
      "epoch": 0.7216358839050132,
      "grad_norm": 0.5768127265884399,
      "learning_rate": 8.069628441030472e-06,
      "loss": 0.0528,
      "step": 547
    },
    {
      "epoch": 0.7229551451187335,
      "grad_norm": 0.7540897188613321,
      "learning_rate": 8.060529912738316e-06,
      "loss": 0.0549,
      "step": 548
    },
    {
      "epoch": 0.7242744063324539,
      "grad_norm": 0.7070564475373282,
      "learning_rate": 8.051415148879866e-06,
      "loss": 0.0607,
      "step": 549
    },
    {
      "epoch": 0.7255936675461742,
      "grad_norm": 0.4581610747944336,
      "learning_rate": 8.042284197807323e-06,
      "loss": 0.0424,
      "step": 550
    },
    {
      "epoch": 0.7269129287598944,
      "grad_norm": 0.8281322422201968,
      "learning_rate": 8.03313710795876e-06,
      "loss": 0.0554,
      "step": 551
    },
    {
      "epoch": 0.7282321899736148,
      "grad_norm": 0.5221549426233524,
      "learning_rate": 8.023973927857857e-06,
      "loss": 0.0574,
      "step": 552
    },
    {
      "epoch": 0.7295514511873351,
      "grad_norm": 1.2249746651625872,
      "learning_rate": 8.014794706113655e-06,
      "loss": 0.101,
      "step": 553
    },
    {
      "epoch": 0.7308707124010554,
      "grad_norm": 0.6878460489135141,
      "learning_rate": 8.005599491420288e-06,
      "loss": 0.056,
      "step": 554
    },
    {
      "epoch": 0.7321899736147758,
      "grad_norm": 0.5396147843239631,
      "learning_rate": 7.996388332556735e-06,
      "loss": 0.0479,
      "step": 555
    },
    {
      "epoch": 0.7335092348284961,
      "grad_norm": 0.6003872582317632,
      "learning_rate": 7.987161278386555e-06,
      "loss": 0.0523,
      "step": 556
    },
    {
      "epoch": 0.7348284960422163,
      "grad_norm": 0.5390200103337964,
      "learning_rate": 7.977918377857625e-06,
      "loss": 0.0439,
      "step": 557
    },
    {
      "epoch": 0.7361477572559367,
      "grad_norm": 0.8055499994103575,
      "learning_rate": 7.968659680001887e-06,
      "loss": 0.0544,
      "step": 558
    },
    {
      "epoch": 0.737467018469657,
      "grad_norm": 0.7605406050858123,
      "learning_rate": 7.959385233935087e-06,
      "loss": 0.0617,
      "step": 559
    },
    {
      "epoch": 0.7387862796833773,
      "grad_norm": 0.5172162310985919,
      "learning_rate": 7.950095088856509e-06,
      "loss": 0.0434,
      "step": 560
    },
    {
      "epoch": 0.7401055408970977,
      "grad_norm": 0.6613151481614415,
      "learning_rate": 7.940789294048716e-06,
      "loss": 0.0836,
      "step": 561
    },
    {
      "epoch": 0.741424802110818,
      "grad_norm": 0.6934766814145727,
      "learning_rate": 7.931467898877298e-06,
      "loss": 0.0511,
      "step": 562
    },
    {
      "epoch": 0.7427440633245382,
      "grad_norm": 0.6553016483708853,
      "learning_rate": 7.922130952790591e-06,
      "loss": 0.0666,
      "step": 563
    },
    {
      "epoch": 0.7440633245382586,
      "grad_norm": 0.6453614377276752,
      "learning_rate": 7.912778505319436e-06,
      "loss": 0.0639,
      "step": 564
    },
    {
      "epoch": 0.7453825857519789,
      "grad_norm": 0.6281402686074141,
      "learning_rate": 7.9034106060769e-06,
      "loss": 0.0565,
      "step": 565
    },
    {
      "epoch": 0.7467018469656992,
      "grad_norm": 0.48369335771379274,
      "learning_rate": 7.894027304758023e-06,
      "loss": 0.0489,
      "step": 566
    },
    {
      "epoch": 0.7480211081794196,
      "grad_norm": 0.4919829789879673,
      "learning_rate": 7.884628651139543e-06,
      "loss": 0.0538,
      "step": 567
    },
    {
      "epoch": 0.7493403693931399,
      "grad_norm": 0.6107428053033679,
      "learning_rate": 7.875214695079647e-06,
      "loss": 0.0543,
      "step": 568
    },
    {
      "epoch": 0.7506596306068601,
      "grad_norm": 0.784889189073831,
      "learning_rate": 7.865785486517696e-06,
      "loss": 0.0696,
      "step": 569
    },
    {
      "epoch": 0.7519788918205804,
      "grad_norm": 0.7238248168827699,
      "learning_rate": 7.856341075473963e-06,
      "loss": 0.0878,
      "step": 570
    },
    {
      "epoch": 0.7532981530343008,
      "grad_norm": 0.498276868425689,
      "learning_rate": 7.846881512049364e-06,
      "loss": 0.0405,
      "step": 571
    },
    {
      "epoch": 0.7546174142480211,
      "grad_norm": 0.6434293792784026,
      "learning_rate": 7.837406846425205e-06,
      "loss": 0.0565,
      "step": 572
    },
    {
      "epoch": 0.7559366754617414,
      "grad_norm": 1.013063418755939,
      "learning_rate": 7.827917128862895e-06,
      "loss": 0.0598,
      "step": 573
    },
    {
      "epoch": 0.7572559366754618,
      "grad_norm": 0.46003480135417324,
      "learning_rate": 7.818412409703695e-06,
      "loss": 0.0427,
      "step": 574
    },
    {
      "epoch": 0.758575197889182,
      "grad_norm": 0.40641102052743017,
      "learning_rate": 7.808892739368453e-06,
      "loss": 0.0501,
      "step": 575
    },
    {
      "epoch": 0.7598944591029023,
      "grad_norm": 0.6471268390587719,
      "learning_rate": 7.799358168357323e-06,
      "loss": 0.0673,
      "step": 576
    },
    {
      "epoch": 0.7612137203166227,
      "grad_norm": 1.1321389794336494,
      "learning_rate": 7.789808747249505e-06,
      "loss": 0.0536,
      "step": 577
    },
    {
      "epoch": 0.762532981530343,
      "grad_norm": 0.5012512664683457,
      "learning_rate": 7.78024452670298e-06,
      "loss": 0.0502,
      "step": 578
    },
    {
      "epoch": 0.7638522427440633,
      "grad_norm": 0.7629572078707867,
      "learning_rate": 7.770665557454235e-06,
      "loss": 0.0743,
      "step": 579
    },
    {
      "epoch": 0.7651715039577837,
      "grad_norm": 0.4546048265700308,
      "learning_rate": 7.761071890317994e-06,
      "loss": 0.0449,
      "step": 580
    },
    {
      "epoch": 0.7664907651715039,
      "grad_norm": 0.584573861216974,
      "learning_rate": 7.751463576186957e-06,
      "loss": 0.0649,
      "step": 581
    },
    {
      "epoch": 0.7678100263852242,
      "grad_norm": 0.5063892476419637,
      "learning_rate": 7.741840666031517e-06,
      "loss": 0.0449,
      "step": 582
    },
    {
      "epoch": 0.7691292875989446,
      "grad_norm": 0.6406284654329897,
      "learning_rate": 7.7322032108995e-06,
      "loss": 0.0585,
      "step": 583
    },
    {
      "epoch": 0.7704485488126649,
      "grad_norm": 0.8624454092052274,
      "learning_rate": 7.72255126191589e-06,
      "loss": 0.0727,
      "step": 584
    },
    {
      "epoch": 0.7717678100263852,
      "grad_norm": 0.4723653050525128,
      "learning_rate": 7.712884870282558e-06,
      "loss": 0.0571,
      "step": 585
    },
    {
      "epoch": 0.7730870712401056,
      "grad_norm": 0.492657143902818,
      "learning_rate": 7.703204087277989e-06,
      "loss": 0.0381,
      "step": 586
    },
    {
      "epoch": 0.7744063324538258,
      "grad_norm": 0.7072714311281186,
      "learning_rate": 7.693508964257015e-06,
      "loss": 0.0712,
      "step": 587
    },
    {
      "epoch": 0.7757255936675461,
      "grad_norm": 0.4894057590546519,
      "learning_rate": 7.683799552650534e-06,
      "loss": 0.0421,
      "step": 588
    },
    {
      "epoch": 0.7770448548812665,
      "grad_norm": 0.7198612599000973,
      "learning_rate": 7.674075903965254e-06,
      "loss": 0.0552,
      "step": 589
    },
    {
      "epoch": 0.7783641160949868,
      "grad_norm": 0.41818543083228865,
      "learning_rate": 7.66433806978339e-06,
      "loss": 0.0465,
      "step": 590
    },
    {
      "epoch": 0.7796833773087071,
      "grad_norm": 0.40455196583885183,
      "learning_rate": 7.654586101762426e-06,
      "loss": 0.0496,
      "step": 591
    },
    {
      "epoch": 0.7810026385224275,
      "grad_norm": 0.40990203769244127,
      "learning_rate": 7.644820051634813e-06,
      "loss": 0.0502,
      "step": 592
    },
    {
      "epoch": 0.7823218997361477,
      "grad_norm": 0.5855401429694791,
      "learning_rate": 7.63503997120771e-06,
      "loss": 0.0588,
      "step": 593
    },
    {
      "epoch": 0.783641160949868,
      "grad_norm": 0.7633148802383083,
      "learning_rate": 7.625245912362699e-06,
      "loss": 0.0786,
      "step": 594
    },
    {
      "epoch": 0.7849604221635884,
      "grad_norm": 0.4709504379995211,
      "learning_rate": 7.615437927055521e-06,
      "loss": 0.0446,
      "step": 595
    },
    {
      "epoch": 0.7862796833773087,
      "grad_norm": 0.6482167373589524,
      "learning_rate": 7.605616067315793e-06,
      "loss": 0.0703,
      "step": 596
    },
    {
      "epoch": 0.787598944591029,
      "grad_norm": 0.5930419099399733,
      "learning_rate": 7.595780385246729e-06,
      "loss": 0.051,
      "step": 597
    },
    {
      "epoch": 0.7889182058047494,
      "grad_norm": 0.5511986661419541,
      "learning_rate": 7.585930933024874e-06,
      "loss": 0.0525,
      "step": 598
    },
    {
      "epoch": 0.7902374670184696,
      "grad_norm": 0.6286727069894341,
      "learning_rate": 7.576067762899817e-06,
      "loss": 0.0785,
      "step": 599
    },
    {
      "epoch": 0.7915567282321899,
      "grad_norm": 0.5164407352417415,
      "learning_rate": 7.56619092719392e-06,
      "loss": 0.0567,
      "step": 600
    },
    {
      "epoch": 0.7928759894459103,
      "grad_norm": 0.4818775374789297,
      "learning_rate": 7.556300478302038e-06,
      "loss": 0.05,
      "step": 601
    },
    {
      "epoch": 0.7941952506596306,
      "grad_norm": 0.7496483074018772,
      "learning_rate": 7.546396468691241e-06,
      "loss": 0.0801,
      "step": 602
    },
    {
      "epoch": 0.7955145118733509,
      "grad_norm": 0.523511921974713,
      "learning_rate": 7.536478950900537e-06,
      "loss": 0.0638,
      "step": 603
    },
    {
      "epoch": 0.7968337730870713,
      "grad_norm": 0.5506566059940079,
      "learning_rate": 7.526547977540592e-06,
      "loss": 0.0673,
      "step": 604
    },
    {
      "epoch": 0.7981530343007915,
      "grad_norm": 0.6637744672920765,
      "learning_rate": 7.516603601293453e-06,
      "loss": 0.0684,
      "step": 605
    },
    {
      "epoch": 0.7994722955145118,
      "grad_norm": 0.45775146693356783,
      "learning_rate": 7.506645874912264e-06,
      "loss": 0.0703,
      "step": 606
    },
    {
      "epoch": 0.8007915567282322,
      "grad_norm": 0.5296473777169002,
      "learning_rate": 7.4966748512209884e-06,
      "loss": 0.0576,
      "step": 607
    },
    {
      "epoch": 0.8021108179419525,
      "grad_norm": 0.5365328810956881,
      "learning_rate": 7.486690583114137e-06,
      "loss": 0.0445,
      "step": 608
    },
    {
      "epoch": 0.8034300791556728,
      "grad_norm": 0.4620464689864355,
      "learning_rate": 7.47669312355647e-06,
      "loss": 0.0379,
      "step": 609
    },
    {
      "epoch": 0.8047493403693932,
      "grad_norm": 0.5379473346536254,
      "learning_rate": 7.466682525582732e-06,
      "loss": 0.0378,
      "step": 610
    },
    {
      "epoch": 0.8060686015831134,
      "grad_norm": 0.5683775687269517,
      "learning_rate": 7.456658842297364e-06,
      "loss": 0.0574,
      "step": 611
    },
    {
      "epoch": 0.8073878627968337,
      "grad_norm": 0.7608705203135304,
      "learning_rate": 7.446622126874219e-06,
      "loss": 0.05,
      "step": 612
    },
    {
      "epoch": 0.8087071240105541,
      "grad_norm": 0.6638681521654088,
      "learning_rate": 7.436572432556286e-06,
      "loss": 0.0462,
      "step": 613
    },
    {
      "epoch": 0.8100263852242744,
      "grad_norm": 0.4691995535122923,
      "learning_rate": 7.4265098126554065e-06,
      "loss": 0.0457,
      "step": 614
    },
    {
      "epoch": 0.8113456464379947,
      "grad_norm": 0.7250619676691925,
      "learning_rate": 7.416434320551984e-06,
      "loss": 0.0684,
      "step": 615
    },
    {
      "epoch": 0.8126649076517151,
      "grad_norm": 0.5602711634901041,
      "learning_rate": 7.406346009694713e-06,
      "loss": 0.0428,
      "step": 616
    },
    {
      "epoch": 0.8139841688654353,
      "grad_norm": 0.605453105353561,
      "learning_rate": 7.396244933600285e-06,
      "loss": 0.0573,
      "step": 617
    },
    {
      "epoch": 0.8153034300791556,
      "grad_norm": 0.6004698644481296,
      "learning_rate": 7.386131145853111e-06,
      "loss": 0.0473,
      "step": 618
    },
    {
      "epoch": 0.816622691292876,
      "grad_norm": 0.7365332296846882,
      "learning_rate": 7.376004700105034e-06,
      "loss": 0.0578,
      "step": 619
    },
    {
      "epoch": 0.8179419525065963,
      "grad_norm": 0.7070376183840003,
      "learning_rate": 7.365865650075046e-06,
      "loss": 0.0647,
      "step": 620
    },
    {
      "epoch": 0.8192612137203166,
      "grad_norm": 0.4015642712910436,
      "learning_rate": 7.355714049549001e-06,
      "loss": 0.0341,
      "step": 621
    },
    {
      "epoch": 0.820580474934037,
      "grad_norm": 0.6779327868527507,
      "learning_rate": 7.345549952379334e-06,
      "loss": 0.0499,
      "step": 622
    },
    {
      "epoch": 0.8218997361477572,
      "grad_norm": 0.8680340537992209,
      "learning_rate": 7.335373412484772e-06,
      "loss": 0.0687,
      "step": 623
    },
    {
      "epoch": 0.8232189973614775,
      "grad_norm": 0.637295783632708,
      "learning_rate": 7.325184483850043e-06,
      "loss": 0.0604,
      "step": 624
    },
    {
      "epoch": 0.8245382585751979,
      "grad_norm": 0.8229308507890835,
      "learning_rate": 7.314983220525604e-06,
      "loss": 0.0642,
      "step": 625
    },
    {
      "epoch": 0.8258575197889182,
      "grad_norm": 0.9148264295561875,
      "learning_rate": 7.304769676627339e-06,
      "loss": 0.0657,
      "step": 626
    },
    {
      "epoch": 0.8271767810026385,
      "grad_norm": 0.5892300500844424,
      "learning_rate": 7.294543906336279e-06,
      "loss": 0.0508,
      "step": 627
    },
    {
      "epoch": 0.8284960422163589,
      "grad_norm": 0.7696278821983158,
      "learning_rate": 7.284305963898315e-06,
      "loss": 0.0593,
      "step": 628
    },
    {
      "epoch": 0.8298153034300791,
      "grad_norm": 0.7817662132002492,
      "learning_rate": 7.274055903623911e-06,
      "loss": 0.0661,
      "step": 629
    },
    {
      "epoch": 0.8311345646437994,
      "grad_norm": 0.599073351844615,
      "learning_rate": 7.2637937798878085e-06,
      "loss": 0.0493,
      "step": 630
    },
    {
      "epoch": 0.8324538258575198,
      "grad_norm": 0.45284300597451216,
      "learning_rate": 7.25351964712875e-06,
      "loss": 0.0484,
      "step": 631
    },
    {
      "epoch": 0.8337730870712401,
      "grad_norm": 0.6016647313251364,
      "learning_rate": 7.243233559849179e-06,
      "loss": 0.0465,
      "step": 632
    },
    {
      "epoch": 0.8350923482849604,
      "grad_norm": 0.49389914293306764,
      "learning_rate": 7.232935572614957e-06,
      "loss": 0.0447,
      "step": 633
    },
    {
      "epoch": 0.8364116094986808,
      "grad_norm": 0.7349002176146826,
      "learning_rate": 7.222625740055072e-06,
      "loss": 0.0563,
      "step": 634
    },
    {
      "epoch": 0.837730870712401,
      "grad_norm": 0.7897682173746414,
      "learning_rate": 7.212304116861354e-06,
      "loss": 0.0766,
      "step": 635
    },
    {
      "epoch": 0.8390501319261213,
      "grad_norm": 0.5196108727337423,
      "learning_rate": 7.201970757788172e-06,
      "loss": 0.0649,
      "step": 636
    },
    {
      "epoch": 0.8403693931398417,
      "grad_norm": 0.4247043219150824,
      "learning_rate": 7.191625717652158e-06,
      "loss": 0.0362,
      "step": 637
    },
    {
      "epoch": 0.841688654353562,
      "grad_norm": 0.4002182375426419,
      "learning_rate": 7.18126905133191e-06,
      "loss": 0.0475,
      "step": 638
    },
    {
      "epoch": 0.8430079155672823,
      "grad_norm": 0.5029922343647233,
      "learning_rate": 7.170900813767694e-06,
      "loss": 0.0478,
      "step": 639
    },
    {
      "epoch": 0.8443271767810027,
      "grad_norm": 0.7565385955901749,
      "learning_rate": 7.160521059961169e-06,
      "loss": 0.0782,
      "step": 640
    },
    {
      "epoch": 0.8456464379947229,
      "grad_norm": 0.9185509777670232,
      "learning_rate": 7.150129844975079e-06,
      "loss": 0.0659,
      "step": 641
    },
    {
      "epoch": 0.8469656992084432,
      "grad_norm": 0.4891563297818167,
      "learning_rate": 7.1397272239329684e-06,
      "loss": 0.0532,
      "step": 642
    },
    {
      "epoch": 0.8482849604221636,
      "grad_norm": 0.4994994913838797,
      "learning_rate": 7.129313252018892e-06,
      "loss": 0.0594,
      "step": 643
    },
    {
      "epoch": 0.8496042216358839,
      "grad_norm": 0.5191192379556658,
      "learning_rate": 7.118887984477116e-06,
      "loss": 0.0529,
      "step": 644
    },
    {
      "epoch": 0.8509234828496042,
      "grad_norm": 1.133290514169336,
      "learning_rate": 7.108451476611828e-06,
      "loss": 0.0444,
      "step": 645
    },
    {
      "epoch": 0.8522427440633246,
      "grad_norm": 1.9081274613732992,
      "learning_rate": 7.098003783786844e-06,
      "loss": 0.0777,
      "step": 646
    },
    {
      "epoch": 0.8535620052770448,
      "grad_norm": 0.5197615187332945,
      "learning_rate": 7.087544961425317e-06,
      "loss": 0.0502,
      "step": 647
    },
    {
      "epoch": 0.8548812664907651,
      "grad_norm": 0.5557110777870159,
      "learning_rate": 7.0770750650094335e-06,
      "loss": 0.0514,
      "step": 648
    },
    {
      "epoch": 0.8562005277044855,
      "grad_norm": 0.576391920153884,
      "learning_rate": 7.0665941500801306e-06,
      "loss": 0.0779,
      "step": 649
    },
    {
      "epoch": 0.8575197889182058,
      "grad_norm": 2.55924940843789,
      "learning_rate": 7.056102272236799e-06,
      "loss": 0.0565,
      "step": 650
    },
    {
      "epoch": 0.8588390501319261,
      "grad_norm": 0.5559838540095863,
      "learning_rate": 7.045599487136981e-06,
      "loss": 0.0392,
      "step": 651
    },
    {
      "epoch": 0.8601583113456465,
      "grad_norm": 1.1829207239081179,
      "learning_rate": 7.035085850496079e-06,
      "loss": 0.0614,
      "step": 652
    },
    {
      "epoch": 0.8614775725593667,
      "grad_norm": 1.0447984538206716,
      "learning_rate": 7.024561418087068e-06,
      "loss": 0.0817,
      "step": 653
    },
    {
      "epoch": 0.862796833773087,
      "grad_norm": 0.7783307249135734,
      "learning_rate": 7.014026245740185e-06,
      "loss": 0.0591,
      "step": 654
    },
    {
      "epoch": 0.8641160949868074,
      "grad_norm": 0.6461374423225957,
      "learning_rate": 7.003480389342645e-06,
      "loss": 0.0494,
      "step": 655
    },
    {
      "epoch": 0.8654353562005277,
      "grad_norm": 0.9384339853723209,
      "learning_rate": 6.992923904838341e-06,
      "loss": 0.0721,
      "step": 656
    },
    {
      "epoch": 0.866754617414248,
      "grad_norm": 0.6660800226382337,
      "learning_rate": 6.982356848227537e-06,
      "loss": 0.052,
      "step": 657
    },
    {
      "epoch": 0.8680738786279684,
      "grad_norm": 0.619900101241196,
      "learning_rate": 6.971779275566593e-06,
      "loss": 0.0578,
      "step": 658
    },
    {
      "epoch": 0.8693931398416886,
      "grad_norm": 0.5715915626402758,
      "learning_rate": 6.9611912429676485e-06,
      "loss": 0.0543,
      "step": 659
    },
    {
      "epoch": 0.8707124010554089,
      "grad_norm": 0.7270686004959181,
      "learning_rate": 6.9505928065983275e-06,
      "loss": 0.057,
      "step": 660
    },
    {
      "epoch": 0.8720316622691293,
      "grad_norm": 0.5198432335246478,
      "learning_rate": 6.939984022681451e-06,
      "loss": 0.0351,
      "step": 661
    },
    {
      "epoch": 0.8733509234828496,
      "grad_norm": 0.8985719129453157,
      "learning_rate": 6.929364947494729e-06,
      "loss": 0.0547,
      "step": 662
    },
    {
      "epoch": 0.8746701846965699,
      "grad_norm": 0.5769158166108265,
      "learning_rate": 6.918735637370462e-06,
      "loss": 0.0427,
      "step": 663
    },
    {
      "epoch": 0.8759894459102903,
      "grad_norm": 0.6653103479472001,
      "learning_rate": 6.908096148695251e-06,
      "loss": 0.0599,
      "step": 664
    },
    {
      "epoch": 0.8773087071240105,
      "grad_norm": 0.8404062260080382,
      "learning_rate": 6.897446537909686e-06,
      "loss": 0.0725,
      "step": 665
    },
    {
      "epoch": 0.8786279683377308,
      "grad_norm": 0.4752059893802516,
      "learning_rate": 6.886786861508061e-06,
      "loss": 0.0545,
      "step": 666
    },
    {
      "epoch": 0.8799472295514512,
      "grad_norm": 0.4778488711685745,
      "learning_rate": 6.876117176038058e-06,
      "loss": 0.0567,
      "step": 667
    },
    {
      "epoch": 0.8812664907651715,
      "grad_norm": 0.46031192038266594,
      "learning_rate": 6.865437538100456e-06,
      "loss": 0.0407,
      "step": 668
    },
    {
      "epoch": 0.8825857519788918,
      "grad_norm": 0.7180279054018667,
      "learning_rate": 6.85474800434884e-06,
      "loss": 0.0712,
      "step": 669
    },
    {
      "epoch": 0.8839050131926122,
      "grad_norm": 0.45300566773960316,
      "learning_rate": 6.8440486314892775e-06,
      "loss": 0.0488,
      "step": 670
    },
    {
      "epoch": 0.8852242744063324,
      "grad_norm": 0.5156880972814101,
      "learning_rate": 6.833339476280038e-06,
      "loss": 0.0581,
      "step": 671
    },
    {
      "epoch": 0.8865435356200527,
      "grad_norm": 0.36978448089350285,
      "learning_rate": 6.822620595531286e-06,
      "loss": 0.0439,
      "step": 672
    },
    {
      "epoch": 0.8878627968337731,
      "grad_norm": 0.6409149604296679,
      "learning_rate": 6.811892046104772e-06,
      "loss": 0.0724,
      "step": 673
    },
    {
      "epoch": 0.8891820580474934,
      "grad_norm": 0.5164562855547227,
      "learning_rate": 6.801153884913541e-06,
      "loss": 0.0617,
      "step": 674
    },
    {
      "epoch": 0.8905013192612137,
      "grad_norm": 0.5137537287977944,
      "learning_rate": 6.790406168921631e-06,
      "loss": 0.0537,
      "step": 675
    },
    {
      "epoch": 0.8918205804749341,
      "grad_norm": 0.6248217252051893,
      "learning_rate": 6.779648955143754e-06,
      "loss": 0.0595,
      "step": 676
    },
    {
      "epoch": 0.8931398416886543,
      "grad_norm": 0.9117584538075659,
      "learning_rate": 6.768882300645019e-06,
      "loss": 0.0778,
      "step": 677
    },
    {
      "epoch": 0.8944591029023746,
      "grad_norm": 1.0045913675480198,
      "learning_rate": 6.758106262540611e-06,
      "loss": 0.0365,
      "step": 678
    },
    {
      "epoch": 0.895778364116095,
      "grad_norm": 0.8182875631557209,
      "learning_rate": 6.747320897995493e-06,
      "loss": 0.0597,
      "step": 679
    },
    {
      "epoch": 0.8970976253298153,
      "grad_norm": 0.40301293105178015,
      "learning_rate": 6.736526264224101e-06,
      "loss": 0.0377,
      "step": 680
    },
    {
      "epoch": 0.8984168865435356,
      "grad_norm": 0.4830059646576911,
      "learning_rate": 6.7257224184900505e-06,
      "loss": 0.0581,
      "step": 681
    },
    {
      "epoch": 0.899736147757256,
      "grad_norm": 0.7384347801633252,
      "learning_rate": 6.714909418105816e-06,
      "loss": 0.0502,
      "step": 682
    },
    {
      "epoch": 0.9010554089709762,
      "grad_norm": 0.44478418897913596,
      "learning_rate": 6.70408732043244e-06,
      "loss": 0.0425,
      "step": 683
    },
    {
      "epoch": 0.9023746701846965,
      "grad_norm": 0.47600625569629695,
      "learning_rate": 6.693256182879224e-06,
      "loss": 0.047,
      "step": 684
    },
    {
      "epoch": 0.9036939313984169,
      "grad_norm": 1.5977235088888984,
      "learning_rate": 6.682416062903425e-06,
      "loss": 0.0674,
      "step": 685
    },
    {
      "epoch": 0.9050131926121372,
      "grad_norm": 0.5372132111670395,
      "learning_rate": 6.671567018009948e-06,
      "loss": 0.0584,
      "step": 686
    },
    {
      "epoch": 0.9063324538258575,
      "grad_norm": 0.6442744048697985,
      "learning_rate": 6.660709105751046e-06,
      "loss": 0.0538,
      "step": 687
    },
    {
      "epoch": 0.9076517150395779,
      "grad_norm": 0.506555769885308,
      "learning_rate": 6.649842383726011e-06,
      "loss": 0.0525,
      "step": 688
    },
    {
      "epoch": 0.9089709762532981,
      "grad_norm": 0.7054885308014072,
      "learning_rate": 6.638966909580866e-06,
      "loss": 0.0517,
      "step": 689
    },
    {
      "epoch": 0.9102902374670184,
      "grad_norm": 0.7442326445547589,
      "learning_rate": 6.628082741008068e-06,
      "loss": 0.0611,
      "step": 690
    },
    {
      "epoch": 0.9116094986807388,
      "grad_norm": 0.5183656940281165,
      "learning_rate": 6.617189935746191e-06,
      "loss": 0.064,
      "step": 691
    },
    {
      "epoch": 0.9129287598944591,
      "grad_norm": 0.4464207504937014,
      "learning_rate": 6.606288551579629e-06,
      "loss": 0.0446,
      "step": 692
    },
    {
      "epoch": 0.9142480211081794,
      "grad_norm": 0.37526282747721584,
      "learning_rate": 6.595378646338285e-06,
      "loss": 0.0288,
      "step": 693
    },
    {
      "epoch": 0.9155672823218998,
      "grad_norm": 0.5777950907682003,
      "learning_rate": 6.584460277897262e-06,
      "loss": 0.0677,
      "step": 694
    },
    {
      "epoch": 0.91688654353562,
      "grad_norm": 0.6652617133702854,
      "learning_rate": 6.573533504176562e-06,
      "loss": 0.0715,
      "step": 695
    },
    {
      "epoch": 0.9182058047493403,
      "grad_norm": 0.4667904960438139,
      "learning_rate": 6.562598383140773e-06,
      "loss": 0.0834,
      "step": 696
    },
    {
      "epoch": 0.9195250659630607,
      "grad_norm": 0.8345298859743209,
      "learning_rate": 6.551654972798765e-06,
      "loss": 0.058,
      "step": 697
    },
    {
      "epoch": 0.920844327176781,
      "grad_norm": 0.44292680991761424,
      "learning_rate": 6.540703331203382e-06,
      "loss": 0.0509,
      "step": 698
    },
    {
      "epoch": 0.9221635883905013,
      "grad_norm": 0.6291584576374266,
      "learning_rate": 6.529743516451135e-06,
      "loss": 0.048,
      "step": 699
    },
    {
      "epoch": 0.9234828496042217,
      "grad_norm": 0.38726410745858225,
      "learning_rate": 6.518775586681887e-06,
      "loss": 0.0428,
      "step": 700
    },
    {
      "epoch": 0.924802110817942,
      "grad_norm": 0.6681003494953462,
      "learning_rate": 6.507799600078554e-06,
      "loss": 0.0642,
      "step": 701
    },
    {
      "epoch": 0.9261213720316622,
      "grad_norm": 0.3838767558388101,
      "learning_rate": 6.496815614866792e-06,
      "loss": 0.0406,
      "step": 702
    },
    {
      "epoch": 0.9274406332453826,
      "grad_norm": 0.9951221686609095,
      "learning_rate": 6.485823689314683e-06,
      "loss": 0.0705,
      "step": 703
    },
    {
      "epoch": 0.9287598944591029,
      "grad_norm": 0.5300121081133167,
      "learning_rate": 6.4748238817324395e-06,
      "loss": 0.0679,
      "step": 704
    },
    {
      "epoch": 0.9300791556728232,
      "grad_norm": 0.6579693283135651,
      "learning_rate": 6.46381625047208e-06,
      "loss": 0.0443,
      "step": 705
    },
    {
      "epoch": 0.9313984168865436,
      "grad_norm": 0.7154530047746038,
      "learning_rate": 6.452800853927128e-06,
      "loss": 0.0639,
      "step": 706
    },
    {
      "epoch": 0.9327176781002638,
      "grad_norm": 0.43183046299555017,
      "learning_rate": 6.441777750532302e-06,
      "loss": 0.0432,
      "step": 707
    },
    {
      "epoch": 0.9340369393139841,
      "grad_norm": 0.6482390999792111,
      "learning_rate": 6.430746998763204e-06,
      "loss": 0.0404,
      "step": 708
    },
    {
      "epoch": 0.9353562005277045,
      "grad_norm": 0.5954280306571919,
      "learning_rate": 6.419708657136008e-06,
      "loss": 0.0476,
      "step": 709
    },
    {
      "epoch": 0.9366754617414248,
      "grad_norm": 0.5670884695354337,
      "learning_rate": 6.408662784207149e-06,
      "loss": 0.0465,
      "step": 710
    },
    {
      "epoch": 0.9379947229551451,
      "grad_norm": 0.39195350309825466,
      "learning_rate": 6.397609438573018e-06,
      "loss": 0.0367,
      "step": 711
    },
    {
      "epoch": 0.9393139841688655,
      "grad_norm": 0.5135065451574998,
      "learning_rate": 6.386548678869644e-06,
      "loss": 0.0545,
      "step": 712
    },
    {
      "epoch": 0.9406332453825857,
      "grad_norm": 0.5440916522763131,
      "learning_rate": 6.375480563772391e-06,
      "loss": 0.0516,
      "step": 713
    },
    {
      "epoch": 0.941952506596306,
      "grad_norm": 0.6146072852339167,
      "learning_rate": 6.3644051519956366e-06,
      "loss": 0.0657,
      "step": 714
    },
    {
      "epoch": 0.9432717678100264,
      "grad_norm": 0.46822032894373955,
      "learning_rate": 6.353322502292468e-06,
      "loss": 0.0487,
      "step": 715
    },
    {
      "epoch": 0.9445910290237467,
      "grad_norm": 0.7703382520399475,
      "learning_rate": 6.342232673454371e-06,
      "loss": 0.0824,
      "step": 716
    },
    {
      "epoch": 0.945910290237467,
      "grad_norm": 0.5658022209850417,
      "learning_rate": 6.331135724310912e-06,
      "loss": 0.0423,
      "step": 717
    },
    {
      "epoch": 0.9472295514511874,
      "grad_norm": 0.6941744607255129,
      "learning_rate": 6.320031713729429e-06,
      "loss": 0.0369,
      "step": 718
    },
    {
      "epoch": 0.9485488126649076,
      "grad_norm": 0.6379949588819936,
      "learning_rate": 6.308920700614722e-06,
      "loss": 0.055,
      "step": 719
    },
    {
      "epoch": 0.9498680738786279,
      "grad_norm": 0.5566867478458106,
      "learning_rate": 6.2978027439087405e-06,
      "loss": 0.0635,
      "step": 720
    },
    {
      "epoch": 0.9511873350923483,
      "grad_norm": 0.49482508553292304,
      "learning_rate": 6.286677902590262e-06,
      "loss": 0.0485,
      "step": 721
    },
    {
      "epoch": 0.9525065963060686,
      "grad_norm": 0.42405552441199723,
      "learning_rate": 6.2755462356745885e-06,
      "loss": 0.0398,
      "step": 722
    },
    {
      "epoch": 0.9538258575197889,
      "grad_norm": 0.3921926452470498,
      "learning_rate": 6.2644078022132344e-06,
      "loss": 0.0449,
      "step": 723
    },
    {
      "epoch": 0.9551451187335093,
      "grad_norm": 0.622187385729326,
      "learning_rate": 6.2532626612936035e-06,
      "loss": 0.0492,
      "step": 724
    },
    {
      "epoch": 0.9564643799472295,
      "grad_norm": 0.8846149336240449,
      "learning_rate": 6.2421108720386835e-06,
      "loss": 0.0718,
      "step": 725
    },
    {
      "epoch": 0.9577836411609498,
      "grad_norm": 0.835978497608325,
      "learning_rate": 6.2309524936067344e-06,
      "loss": 0.0651,
      "step": 726
    },
    {
      "epoch": 0.9591029023746702,
      "grad_norm": 0.49758894330199727,
      "learning_rate": 6.219787585190964e-06,
      "loss": 0.0422,
      "step": 727
    },
    {
      "epoch": 0.9604221635883905,
      "grad_norm": 0.5331036943980498,
      "learning_rate": 6.208616206019225e-06,
      "loss": 0.0579,
      "step": 728
    },
    {
      "epoch": 0.9617414248021108,
      "grad_norm": 0.6148288660973584,
      "learning_rate": 6.197438415353694e-06,
      "loss": 0.063,
      "step": 729
    },
    {
      "epoch": 0.9630606860158312,
      "grad_norm": 0.49695243951671314,
      "learning_rate": 6.1862542724905605e-06,
      "loss": 0.0487,
      "step": 730
    },
    {
      "epoch": 0.9643799472295514,
      "grad_norm": 0.4105127820148204,
      "learning_rate": 6.175063836759712e-06,
      "loss": 0.0488,
      "step": 731
    },
    {
      "epoch": 0.9656992084432717,
      "grad_norm": 0.47446528356917433,
      "learning_rate": 6.163867167524419e-06,
      "loss": 0.047,
      "step": 732
    },
    {
      "epoch": 0.9670184696569921,
      "grad_norm": 0.4703664127153639,
      "learning_rate": 6.1526643241810145e-06,
      "loss": 0.0419,
      "step": 733
    },
    {
      "epoch": 0.9683377308707124,
      "grad_norm": 0.39746241068010874,
      "learning_rate": 6.1414553661585905e-06,
      "loss": 0.0422,
      "step": 734
    },
    {
      "epoch": 0.9696569920844327,
      "grad_norm": 0.5170990993750981,
      "learning_rate": 6.130240352918675e-06,
      "loss": 0.0661,
      "step": 735
    },
    {
      "epoch": 0.9709762532981531,
      "grad_norm": 0.5152004279143637,
      "learning_rate": 6.119019343954914e-06,
      "loss": 0.0374,
      "step": 736
    },
    {
      "epoch": 0.9722955145118733,
      "grad_norm": 0.522388520081193,
      "learning_rate": 6.107792398792763e-06,
      "loss": 0.053,
      "step": 737
    },
    {
      "epoch": 0.9736147757255936,
      "grad_norm": 0.7152588462194387,
      "learning_rate": 6.096559576989166e-06,
      "loss": 0.0692,
      "step": 738
    },
    {
      "epoch": 0.974934036939314,
      "grad_norm": 0.4545424319736431,
      "learning_rate": 6.085320938132245e-06,
      "loss": 0.0469,
      "step": 739
    },
    {
      "epoch": 0.9762532981530343,
      "grad_norm": 0.7390386089115659,
      "learning_rate": 6.074076541840978e-06,
      "loss": 0.0574,
      "step": 740
    },
    {
      "epoch": 0.9775725593667546,
      "grad_norm": 0.5797819489692182,
      "learning_rate": 6.062826447764883e-06,
      "loss": 0.0551,
      "step": 741
    },
    {
      "epoch": 0.978891820580475,
      "grad_norm": 0.9043388866815173,
      "learning_rate": 6.05157071558371e-06,
      "loss": 0.0487,
      "step": 742
    },
    {
      "epoch": 0.9802110817941952,
      "grad_norm": 0.5300095964363111,
      "learning_rate": 6.0403094050071106e-06,
      "loss": 0.0461,
      "step": 743
    },
    {
      "epoch": 0.9815303430079155,
      "grad_norm": 1.2135080457223286,
      "learning_rate": 6.029042575774334e-06,
      "loss": 0.0744,
      "step": 744
    },
    {
      "epoch": 0.9828496042216359,
      "grad_norm": 0.3508635868880483,
      "learning_rate": 6.017770287653905e-06,
      "loss": 0.0414,
      "step": 745
    },
    {
      "epoch": 0.9841688654353562,
      "grad_norm": 0.46813312237331894,
      "learning_rate": 6.006492600443301e-06,
      "loss": 0.0479,
      "step": 746
    },
    {
      "epoch": 0.9854881266490765,
      "grad_norm": 0.4606890493355314,
      "learning_rate": 5.995209573968648e-06,
      "loss": 0.0369,
      "step": 747
    },
    {
      "epoch": 0.9868073878627969,
      "grad_norm": 0.7963291719507631,
      "learning_rate": 5.9839212680843925e-06,
      "loss": 0.0853,
      "step": 748
    },
    {
      "epoch": 0.9881266490765171,
      "grad_norm": 0.5155724288962596,
      "learning_rate": 5.9726277426729866e-06,
      "loss": 0.0695,
      "step": 749
    },
    {
      "epoch": 0.9894459102902374,
      "grad_norm": 0.4630288406158184,
      "learning_rate": 5.961329057644571e-06,
      "loss": 0.0384,
      "step": 750
    },
    {
      "epoch": 0.9907651715039578,
      "grad_norm": 1.051448267922444,
      "learning_rate": 5.95002527293666e-06,
      "loss": 0.0651,
      "step": 751
    },
    {
      "epoch": 0.9920844327176781,
      "grad_norm": 0.6193715825035618,
      "learning_rate": 5.938716448513819e-06,
      "loss": 0.0567,
      "step": 752
    },
    {
      "epoch": 0.9934036939313984,
      "grad_norm": 0.8381031857754667,
      "learning_rate": 5.927402644367345e-06,
      "loss": 0.0618,
      "step": 753
    },
    {
      "epoch": 0.9947229551451188,
      "grad_norm": 0.3803899696750758,
      "learning_rate": 5.916083920514959e-06,
      "loss": 0.0348,
      "step": 754
    },
    {
      "epoch": 0.996042216358839,
      "grad_norm": 0.6495643260632261,
      "learning_rate": 5.904760337000473e-06,
      "loss": 0.0718,
      "step": 755
    },
    {
      "epoch": 0.9973614775725593,
      "grad_norm": 1.1087437172957955,
      "learning_rate": 5.893431953893483e-06,
      "loss": 0.0466,
      "step": 756
    },
    {
      "epoch": 0.9986807387862797,
      "grad_norm": 1.202764067963593,
      "learning_rate": 5.882098831289044e-06,
      "loss": 0.039,
      "step": 757
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.46604548846150806,
      "learning_rate": 5.8707610293073524e-06,
      "loss": 0.0458,
      "step": 758
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.05592323839664459,
      "eval_runtime": 128.0063,
      "eval_samples_per_second": 39.881,
      "eval_steps_per_second": 1.25,
      "step": 758
    },
    {
      "epoch": 1.0013192612137203,
      "grad_norm": 0.6989593085889051,
      "learning_rate": 5.85941860809343e-06,
      "loss": 0.0458,
      "step": 759
    },
    {
      "epoch": 1.0026385224274406,
      "grad_norm": 0.5927038885689029,
      "learning_rate": 5.848071627816804e-06,
      "loss": 0.0538,
      "step": 760
    },
    {
      "epoch": 1.003957783641161,
      "grad_norm": 0.8857298937072677,
      "learning_rate": 5.836720148671182e-06,
      "loss": 0.0467,
      "step": 761
    },
    {
      "epoch": 1.0052770448548813,
      "grad_norm": 0.4109405171330407,
      "learning_rate": 5.82536423087414e-06,
      "loss": 0.038,
      "step": 762
    },
    {
      "epoch": 1.0065963060686016,
      "grad_norm": 1.5021621311748752,
      "learning_rate": 5.814003934666802e-06,
      "loss": 0.0574,
      "step": 763
    },
    {
      "epoch": 1.007915567282322,
      "grad_norm": 0.4718122478262424,
      "learning_rate": 5.8026393203135145e-06,
      "loss": 0.0473,
      "step": 764
    },
    {
      "epoch": 1.0092348284960422,
      "grad_norm": 0.6482837984831183,
      "learning_rate": 5.791270448101533e-06,
      "loss": 0.055,
      "step": 765
    },
    {
      "epoch": 1.0105540897097625,
      "grad_norm": 0.47324139957617567,
      "learning_rate": 5.779897378340705e-06,
      "loss": 0.0366,
      "step": 766
    },
    {
      "epoch": 1.0118733509234827,
      "grad_norm": 0.5751432075833953,
      "learning_rate": 5.768520171363135e-06,
      "loss": 0.0374,
      "step": 767
    },
    {
      "epoch": 1.0131926121372032,
      "grad_norm": 0.7353563730795333,
      "learning_rate": 5.757138887522884e-06,
      "loss": 0.045,
      "step": 768
    },
    {
      "epoch": 1.0145118733509235,
      "grad_norm": 0.6061560135738652,
      "learning_rate": 5.745753587195636e-06,
      "loss": 0.0595,
      "step": 769
    },
    {
      "epoch": 1.0158311345646438,
      "grad_norm": 0.49181644883396125,
      "learning_rate": 5.734364330778381e-06,
      "loss": 0.0415,
      "step": 770
    },
    {
      "epoch": 1.017150395778364,
      "grad_norm": 0.6430420819909749,
      "learning_rate": 5.722971178689096e-06,
      "loss": 0.0722,
      "step": 771
    },
    {
      "epoch": 1.0184696569920844,
      "grad_norm": 0.4839549081518064,
      "learning_rate": 5.711574191366427e-06,
      "loss": 0.0372,
      "step": 772
    },
    {
      "epoch": 1.0197889182058049,
      "grad_norm": 0.5188479898199087,
      "learning_rate": 5.70017342926936e-06,
      "loss": 0.0479,
      "step": 773
    },
    {
      "epoch": 1.0211081794195251,
      "grad_norm": 0.7917447001213317,
      "learning_rate": 5.68876895287691e-06,
      "loss": 0.0463,
      "step": 774
    },
    {
      "epoch": 1.0224274406332454,
      "grad_norm": 0.46624077553959226,
      "learning_rate": 5.677360822687794e-06,
      "loss": 0.046,
      "step": 775
    },
    {
      "epoch": 1.0237467018469657,
      "grad_norm": 0.4953051288461476,
      "learning_rate": 5.66594909922011e-06,
      "loss": 0.0367,
      "step": 776
    },
    {
      "epoch": 1.025065963060686,
      "grad_norm": 0.6059827706644242,
      "learning_rate": 5.654533843011019e-06,
      "loss": 0.0404,
      "step": 777
    },
    {
      "epoch": 1.0263852242744063,
      "grad_norm": 0.6064456753320037,
      "learning_rate": 5.6431151146164255e-06,
      "loss": 0.0479,
      "step": 778
    },
    {
      "epoch": 1.0277044854881265,
      "grad_norm": 0.775592672100191,
      "learning_rate": 5.631692974610647e-06,
      "loss": 0.0619,
      "step": 779
    },
    {
      "epoch": 1.029023746701847,
      "grad_norm": 0.4732107305616271,
      "learning_rate": 5.6202674835861045e-06,
      "loss": 0.0394,
      "step": 780
    },
    {
      "epoch": 1.0303430079155673,
      "grad_norm": 0.5693888680251252,
      "learning_rate": 5.6088387021529966e-06,
      "loss": 0.0526,
      "step": 781
    },
    {
      "epoch": 1.0316622691292876,
      "grad_norm": 0.6931708016123965,
      "learning_rate": 5.597406690938969e-06,
      "loss": 0.0513,
      "step": 782
    },
    {
      "epoch": 1.0329815303430079,
      "grad_norm": 0.8142189449285727,
      "learning_rate": 5.585971510588808e-06,
      "loss": 0.0424,
      "step": 783
    },
    {
      "epoch": 1.0343007915567282,
      "grad_norm": 0.6652724688329306,
      "learning_rate": 5.574533221764109e-06,
      "loss": 0.0367,
      "step": 784
    },
    {
      "epoch": 1.0356200527704487,
      "grad_norm": 0.3300674518769234,
      "learning_rate": 5.563091885142959e-06,
      "loss": 0.0272,
      "step": 785
    },
    {
      "epoch": 1.036939313984169,
      "grad_norm": 0.5380054416165114,
      "learning_rate": 5.551647561419611e-06,
      "loss": 0.0343,
      "step": 786
    },
    {
      "epoch": 1.0382585751978892,
      "grad_norm": 0.7056758338546055,
      "learning_rate": 5.540200311304164e-06,
      "loss": 0.0276,
      "step": 787
    },
    {
      "epoch": 1.0395778364116095,
      "grad_norm": 0.8092651006949911,
      "learning_rate": 5.528750195522244e-06,
      "loss": 0.0377,
      "step": 788
    },
    {
      "epoch": 1.0408970976253298,
      "grad_norm": 1.0814167405738828,
      "learning_rate": 5.517297274814674e-06,
      "loss": 0.0722,
      "step": 789
    },
    {
      "epoch": 1.04221635883905,
      "grad_norm": 0.6447685602617403,
      "learning_rate": 5.505841609937162e-06,
      "loss": 0.0522,
      "step": 790
    },
    {
      "epoch": 1.0435356200527703,
      "grad_norm": 0.45972221138369695,
      "learning_rate": 5.494383261659968e-06,
      "loss": 0.0306,
      "step": 791
    },
    {
      "epoch": 1.0448548812664908,
      "grad_norm": 0.5502963065155776,
      "learning_rate": 5.4829222907675895e-06,
      "loss": 0.0341,
      "step": 792
    },
    {
      "epoch": 1.0461741424802111,
      "grad_norm": 0.5923787629429828,
      "learning_rate": 5.47145875805844e-06,
      "loss": 0.0483,
      "step": 793
    },
    {
      "epoch": 1.0474934036939314,
      "grad_norm": 0.40453312954209114,
      "learning_rate": 5.459992724344516e-06,
      "loss": 0.0358,
      "step": 794
    },
    {
      "epoch": 1.0488126649076517,
      "grad_norm": 0.4634998773083594,
      "learning_rate": 5.448524250451086e-06,
      "loss": 0.042,
      "step": 795
    },
    {
      "epoch": 1.050131926121372,
      "grad_norm": 0.4911088924835721,
      "learning_rate": 5.437053397216364e-06,
      "loss": 0.0536,
      "step": 796
    },
    {
      "epoch": 1.0514511873350922,
      "grad_norm": 1.1796657206533812,
      "learning_rate": 5.425580225491182e-06,
      "loss": 0.0415,
      "step": 797
    },
    {
      "epoch": 1.0527704485488127,
      "grad_norm": 0.510337434330926,
      "learning_rate": 5.4141047961386724e-06,
      "loss": 0.0512,
      "step": 798
    },
    {
      "epoch": 1.054089709762533,
      "grad_norm": 0.40102710131189545,
      "learning_rate": 5.40262717003395e-06,
      "loss": 0.043,
      "step": 799
    },
    {
      "epoch": 1.0554089709762533,
      "grad_norm": 0.4964134440989535,
      "learning_rate": 5.3911474080637705e-06,
      "loss": 0.0408,
      "step": 800
    },
    {
      "epoch": 1.0567282321899736,
      "grad_norm": 0.7046860769411418,
      "learning_rate": 5.379665571126232e-06,
      "loss": 0.0322,
      "step": 801
    },
    {
      "epoch": 1.0580474934036939,
      "grad_norm": 0.732062985823511,
      "learning_rate": 5.368181720130434e-06,
      "loss": 0.0567,
      "step": 802
    },
    {
      "epoch": 1.0593667546174141,
      "grad_norm": 0.5027796864782664,
      "learning_rate": 5.356695915996162e-06,
      "loss": 0.0351,
      "step": 803
    },
    {
      "epoch": 1.0606860158311346,
      "grad_norm": 0.4996625186955253,
      "learning_rate": 5.345208219653562e-06,
      "loss": 0.0439,
      "step": 804
    },
    {
      "epoch": 1.062005277044855,
      "grad_norm": 0.9152620815670903,
      "learning_rate": 5.333718692042817e-06,
      "loss": 0.0689,
      "step": 805
    },
    {
      "epoch": 1.0633245382585752,
      "grad_norm": 0.411853927570493,
      "learning_rate": 5.322227394113826e-06,
      "loss": 0.0344,
      "step": 806
    },
    {
      "epoch": 1.0646437994722955,
      "grad_norm": 0.47740672031416576,
      "learning_rate": 5.310734386825877e-06,
      "loss": 0.0352,
      "step": 807
    },
    {
      "epoch": 1.0659630606860158,
      "grad_norm": 0.5250816082342158,
      "learning_rate": 5.299239731147332e-06,
      "loss": 0.0558,
      "step": 808
    },
    {
      "epoch": 1.0672823218997363,
      "grad_norm": 0.42169584826349693,
      "learning_rate": 5.287743488055288e-06,
      "loss": 0.042,
      "step": 809
    },
    {
      "epoch": 1.0686015831134565,
      "grad_norm": 0.4662553845238838,
      "learning_rate": 5.2762457185352685e-06,
      "loss": 0.0518,
      "step": 810
    },
    {
      "epoch": 1.0699208443271768,
      "grad_norm": 0.4205032546352608,
      "learning_rate": 5.264746483580897e-06,
      "loss": 0.0307,
      "step": 811
    },
    {
      "epoch": 1.071240105540897,
      "grad_norm": 0.3760397290861168,
      "learning_rate": 5.253245844193564e-06,
      "loss": 0.0332,
      "step": 812
    },
    {
      "epoch": 1.0725593667546174,
      "grad_norm": 0.42731752861802264,
      "learning_rate": 5.241743861382116e-06,
      "loss": 0.0366,
      "step": 813
    },
    {
      "epoch": 1.0738786279683377,
      "grad_norm": 0.6398543464753047,
      "learning_rate": 5.2302405961625225e-06,
      "loss": 0.0611,
      "step": 814
    },
    {
      "epoch": 1.075197889182058,
      "grad_norm": 0.5784969161066025,
      "learning_rate": 5.2187361095575596e-06,
      "loss": 0.0588,
      "step": 815
    },
    {
      "epoch": 1.0765171503957784,
      "grad_norm": 0.39490983342015973,
      "learning_rate": 5.2072304625964785e-06,
      "loss": 0.0296,
      "step": 816
    },
    {
      "epoch": 1.0778364116094987,
      "grad_norm": 0.57451932340248,
      "learning_rate": 5.195723716314688e-06,
      "loss": 0.0531,
      "step": 817
    },
    {
      "epoch": 1.079155672823219,
      "grad_norm": 0.6864743480441236,
      "learning_rate": 5.1842159317534304e-06,
      "loss": 0.0342,
      "step": 818
    },
    {
      "epoch": 1.0804749340369393,
      "grad_norm": 0.4862515416335429,
      "learning_rate": 5.172707169959451e-06,
      "loss": 0.0463,
      "step": 819
    },
    {
      "epoch": 1.0817941952506596,
      "grad_norm": 0.6315245551709583,
      "learning_rate": 5.161197491984684e-06,
      "loss": 0.0537,
      "step": 820
    },
    {
      "epoch": 1.08311345646438,
      "grad_norm": 0.6106456993177825,
      "learning_rate": 5.149686958885923e-06,
      "loss": 0.0614,
      "step": 821
    },
    {
      "epoch": 1.0844327176781003,
      "grad_norm": 0.49893031079039063,
      "learning_rate": 5.138175631724495e-06,
      "loss": 0.0491,
      "step": 822
    },
    {
      "epoch": 1.0857519788918206,
      "grad_norm": 0.31147981251138374,
      "learning_rate": 5.12666357156594e-06,
      "loss": 0.0278,
      "step": 823
    },
    {
      "epoch": 1.087071240105541,
      "grad_norm": 0.5875872281434777,
      "learning_rate": 5.11515083947969e-06,
      "loss": 0.0437,
      "step": 824
    },
    {
      "epoch": 1.0883905013192612,
      "grad_norm": 0.5730905962293109,
      "learning_rate": 5.103637496538738e-06,
      "loss": 0.0476,
      "step": 825
    },
    {
      "epoch": 1.0897097625329815,
      "grad_norm": 0.8895822545480768,
      "learning_rate": 5.092123603819318e-06,
      "loss": 0.0434,
      "step": 826
    },
    {
      "epoch": 1.0910290237467017,
      "grad_norm": 0.5226509819931012,
      "learning_rate": 5.080609222400582e-06,
      "loss": 0.0453,
      "step": 827
    },
    {
      "epoch": 1.0923482849604222,
      "grad_norm": 0.49301189424014497,
      "learning_rate": 5.069094413364272e-06,
      "loss": 0.0395,
      "step": 828
    },
    {
      "epoch": 1.0936675461741425,
      "grad_norm": 0.4358556106625212,
      "learning_rate": 5.0575792377944e-06,
      "loss": 0.0345,
      "step": 829
    },
    {
      "epoch": 1.0949868073878628,
      "grad_norm": 0.5273548010655754,
      "learning_rate": 5.046063756776926e-06,
      "loss": 0.0385,
      "step": 830
    },
    {
      "epoch": 1.096306068601583,
      "grad_norm": 1.3462457041300329,
      "learning_rate": 5.03454803139942e-06,
      "loss": 0.0402,
      "step": 831
    },
    {
      "epoch": 1.0976253298153034,
      "grad_norm": 0.4370812057351214,
      "learning_rate": 5.0230321227507595e-06,
      "loss": 0.0325,
      "step": 832
    },
    {
      "epoch": 1.0989445910290236,
      "grad_norm": 0.6267600963038585,
      "learning_rate": 5.0115160919207894e-06,
      "loss": 0.0459,
      "step": 833
    },
    {
      "epoch": 1.1002638522427441,
      "grad_norm": 0.6313390475018821,
      "learning_rate": 5e-06,
      "loss": 0.0571,
      "step": 834
    },
    {
      "epoch": 1.1015831134564644,
      "grad_norm": 0.5417310869246437,
      "learning_rate": 4.988483908079212e-06,
      "loss": 0.0515,
      "step": 835
    },
    {
      "epoch": 1.1029023746701847,
      "grad_norm": 0.3807618659150837,
      "learning_rate": 4.976967877249242e-06,
      "loss": 0.0345,
      "step": 836
    },
    {
      "epoch": 1.104221635883905,
      "grad_norm": 0.421468296572213,
      "learning_rate": 4.9654519686005815e-06,
      "loss": 0.0367,
      "step": 837
    },
    {
      "epoch": 1.1055408970976253,
      "grad_norm": 0.4542996221952275,
      "learning_rate": 4.953936243223077e-06,
      "loss": 0.0365,
      "step": 838
    },
    {
      "epoch": 1.1068601583113455,
      "grad_norm": 0.46168172671671587,
      "learning_rate": 4.942420762205601e-06,
      "loss": 0.0488,
      "step": 839
    },
    {
      "epoch": 1.108179419525066,
      "grad_norm": 0.5673446414814094,
      "learning_rate": 4.93090558663573e-06,
      "loss": 0.0425,
      "step": 840
    },
    {
      "epoch": 1.1094986807387863,
      "grad_norm": 0.4602088265244389,
      "learning_rate": 4.9193907775994196e-06,
      "loss": 0.0378,
      "step": 841
    },
    {
      "epoch": 1.1108179419525066,
      "grad_norm": 0.3983517164777114,
      "learning_rate": 4.907876396180684e-06,
      "loss": 0.0348,
      "step": 842
    },
    {
      "epoch": 1.1121372031662269,
      "grad_norm": 0.38537918505531044,
      "learning_rate": 4.896362503461264e-06,
      "loss": 0.0379,
      "step": 843
    },
    {
      "epoch": 1.1134564643799472,
      "grad_norm": 0.4910602117897347,
      "learning_rate": 4.884849160520311e-06,
      "loss": 0.039,
      "step": 844
    },
    {
      "epoch": 1.1147757255936674,
      "grad_norm": 0.8410599371803616,
      "learning_rate": 4.873336428434062e-06,
      "loss": 0.0795,
      "step": 845
    },
    {
      "epoch": 1.116094986807388,
      "grad_norm": 0.4238465190982125,
      "learning_rate": 4.861824368275508e-06,
      "loss": 0.039,
      "step": 846
    },
    {
      "epoch": 1.1174142480211082,
      "grad_norm": 0.8651752378440294,
      "learning_rate": 4.850313041114078e-06,
      "loss": 0.0403,
      "step": 847
    },
    {
      "epoch": 1.1187335092348285,
      "grad_norm": 0.417802081435934,
      "learning_rate": 4.838802508015316e-06,
      "loss": 0.0406,
      "step": 848
    },
    {
      "epoch": 1.1200527704485488,
      "grad_norm": 0.3506635201550878,
      "learning_rate": 4.82729283004055e-06,
      "loss": 0.0319,
      "step": 849
    },
    {
      "epoch": 1.121372031662269,
      "grad_norm": 0.7348279579830447,
      "learning_rate": 4.815784068246571e-06,
      "loss": 0.0482,
      "step": 850
    },
    {
      "epoch": 1.1226912928759893,
      "grad_norm": 0.4950364378777992,
      "learning_rate": 4.8042762836853135e-06,
      "loss": 0.0366,
      "step": 851
    },
    {
      "epoch": 1.1240105540897098,
      "grad_norm": 0.382207830811648,
      "learning_rate": 4.792769537403523e-06,
      "loss": 0.0306,
      "step": 852
    },
    {
      "epoch": 1.1253298153034301,
      "grad_norm": 0.40855083469378334,
      "learning_rate": 4.781263890442442e-06,
      "loss": 0.0368,
      "step": 853
    },
    {
      "epoch": 1.1266490765171504,
      "grad_norm": 0.5999735367431606,
      "learning_rate": 4.769759403837479e-06,
      "loss": 0.0455,
      "step": 854
    },
    {
      "epoch": 1.1279683377308707,
      "grad_norm": 0.714553966367233,
      "learning_rate": 4.758256138617886e-06,
      "loss": 0.0433,
      "step": 855
    },
    {
      "epoch": 1.129287598944591,
      "grad_norm": 0.5194206554238696,
      "learning_rate": 4.746754155806437e-06,
      "loss": 0.0351,
      "step": 856
    },
    {
      "epoch": 1.1306068601583115,
      "grad_norm": 0.7276753108318581,
      "learning_rate": 4.735253516419104e-06,
      "loss": 0.0624,
      "step": 857
    },
    {
      "epoch": 1.1319261213720317,
      "grad_norm": 0.580276316267312,
      "learning_rate": 4.723754281464732e-06,
      "loss": 0.0465,
      "step": 858
    },
    {
      "epoch": 1.133245382585752,
      "grad_norm": 0.5024855831871866,
      "learning_rate": 4.712256511944714e-06,
      "loss": 0.0349,
      "step": 859
    },
    {
      "epoch": 1.1345646437994723,
      "grad_norm": 0.5832512512257071,
      "learning_rate": 4.700760268852669e-06,
      "loss": 0.0344,
      "step": 860
    },
    {
      "epoch": 1.1358839050131926,
      "grad_norm": 0.37995497107724613,
      "learning_rate": 4.6892656131741235e-06,
      "loss": 0.0322,
      "step": 861
    },
    {
      "epoch": 1.1372031662269129,
      "grad_norm": 0.5083535498753513,
      "learning_rate": 4.677772605886175e-06,
      "loss": 0.0341,
      "step": 862
    },
    {
      "epoch": 1.1385224274406331,
      "grad_norm": 0.4017695107966174,
      "learning_rate": 4.666281307957184e-06,
      "loss": 0.0386,
      "step": 863
    },
    {
      "epoch": 1.1398416886543536,
      "grad_norm": 0.46305677767852116,
      "learning_rate": 4.65479178034644e-06,
      "loss": 0.0378,
      "step": 864
    },
    {
      "epoch": 1.141160949868074,
      "grad_norm": 0.5531056834185942,
      "learning_rate": 4.643304084003839e-06,
      "loss": 0.0398,
      "step": 865
    },
    {
      "epoch": 1.1424802110817942,
      "grad_norm": 0.6317168378781234,
      "learning_rate": 4.631818279869567e-06,
      "loss": 0.0448,
      "step": 866
    },
    {
      "epoch": 1.1437994722955145,
      "grad_norm": 0.44611502163005134,
      "learning_rate": 4.62033442887377e-06,
      "loss": 0.0474,
      "step": 867
    },
    {
      "epoch": 1.1451187335092348,
      "grad_norm": 0.4298271394707465,
      "learning_rate": 4.608852591936231e-06,
      "loss": 0.0366,
      "step": 868
    },
    {
      "epoch": 1.1464379947229553,
      "grad_norm": 0.5739408870671835,
      "learning_rate": 4.597372829966053e-06,
      "loss": 0.0536,
      "step": 869
    },
    {
      "epoch": 1.1477572559366755,
      "grad_norm": 0.6389398509440999,
      "learning_rate": 4.585895203861328e-06,
      "loss": 0.04,
      "step": 870
    },
    {
      "epoch": 1.1490765171503958,
      "grad_norm": 0.5450066757975351,
      "learning_rate": 4.57441977450882e-06,
      "loss": 0.0453,
      "step": 871
    },
    {
      "epoch": 1.150395778364116,
      "grad_norm": 0.5364173970448743,
      "learning_rate": 4.562946602783637e-06,
      "loss": 0.0639,
      "step": 872
    },
    {
      "epoch": 1.1517150395778364,
      "grad_norm": 0.8995373745104792,
      "learning_rate": 4.551475749548915e-06,
      "loss": 0.073,
      "step": 873
    },
    {
      "epoch": 1.1530343007915567,
      "grad_norm": 0.5515888196554394,
      "learning_rate": 4.540007275655485e-06,
      "loss": 0.0311,
      "step": 874
    },
    {
      "epoch": 1.154353562005277,
      "grad_norm": 0.4806942231289383,
      "learning_rate": 4.528541241941562e-06,
      "loss": 0.0383,
      "step": 875
    },
    {
      "epoch": 1.1556728232189974,
      "grad_norm": 0.45703528312267117,
      "learning_rate": 4.517077709232411e-06,
      "loss": 0.031,
      "step": 876
    },
    {
      "epoch": 1.1569920844327177,
      "grad_norm": 0.5649548132455126,
      "learning_rate": 4.505616738340034e-06,
      "loss": 0.0588,
      "step": 877
    },
    {
      "epoch": 1.158311345646438,
      "grad_norm": 0.42689901016539084,
      "learning_rate": 4.49415839006284e-06,
      "loss": 0.0391,
      "step": 878
    },
    {
      "epoch": 1.1596306068601583,
      "grad_norm": 0.47023892707216963,
      "learning_rate": 4.482702725185328e-06,
      "loss": 0.0351,
      "step": 879
    },
    {
      "epoch": 1.1609498680738786,
      "grad_norm": 0.5880176309157975,
      "learning_rate": 4.471249804477758e-06,
      "loss": 0.0495,
      "step": 880
    },
    {
      "epoch": 1.162269129287599,
      "grad_norm": 0.5468215371838773,
      "learning_rate": 4.4597996886958365e-06,
      "loss": 0.0416,
      "step": 881
    },
    {
      "epoch": 1.1635883905013193,
      "grad_norm": 0.36052344580049805,
      "learning_rate": 4.448352438580391e-06,
      "loss": 0.031,
      "step": 882
    },
    {
      "epoch": 1.1649076517150396,
      "grad_norm": 0.48404972523614165,
      "learning_rate": 4.436908114857043e-06,
      "loss": 0.0464,
      "step": 883
    },
    {
      "epoch": 1.16622691292876,
      "grad_norm": 0.595188768341347,
      "learning_rate": 4.4254667782358925e-06,
      "loss": 0.0511,
      "step": 884
    },
    {
      "epoch": 1.1675461741424802,
      "grad_norm": 0.4826949321918809,
      "learning_rate": 4.4140284894111954e-06,
      "loss": 0.0645,
      "step": 885
    },
    {
      "epoch": 1.1688654353562005,
      "grad_norm": 0.4942481166728681,
      "learning_rate": 4.402593309061034e-06,
      "loss": 0.0505,
      "step": 886
    },
    {
      "epoch": 1.1701846965699207,
      "grad_norm": 0.94776252788579,
      "learning_rate": 4.391161297847007e-06,
      "loss": 0.0467,
      "step": 887
    },
    {
      "epoch": 1.1715039577836412,
      "grad_norm": 0.5913035358203962,
      "learning_rate": 4.379732516413897e-06,
      "loss": 0.0319,
      "step": 888
    },
    {
      "epoch": 1.1728232189973615,
      "grad_norm": 0.5561776348517845,
      "learning_rate": 4.368307025389355e-06,
      "loss": 0.0473,
      "step": 889
    },
    {
      "epoch": 1.1741424802110818,
      "grad_norm": 0.4089933220910145,
      "learning_rate": 4.356884885383578e-06,
      "loss": 0.0396,
      "step": 890
    },
    {
      "epoch": 1.175461741424802,
      "grad_norm": 0.3693219469627808,
      "learning_rate": 4.345466156988984e-06,
      "loss": 0.0415,
      "step": 891
    },
    {
      "epoch": 1.1767810026385224,
      "grad_norm": 0.34435079223721005,
      "learning_rate": 4.334050900779893e-06,
      "loss": 0.0301,
      "step": 892
    },
    {
      "epoch": 1.1781002638522429,
      "grad_norm": 0.4369942166434897,
      "learning_rate": 4.322639177312209e-06,
      "loss": 0.034,
      "step": 893
    },
    {
      "epoch": 1.1794195250659631,
      "grad_norm": 0.44910021752481283,
      "learning_rate": 4.3112310471230925e-06,
      "loss": 0.0525,
      "step": 894
    },
    {
      "epoch": 1.1807387862796834,
      "grad_norm": 0.4047144759474002,
      "learning_rate": 4.299826570730642e-06,
      "loss": 0.0293,
      "step": 895
    },
    {
      "epoch": 1.1820580474934037,
      "grad_norm": 0.5291844637455119,
      "learning_rate": 4.2884258086335755e-06,
      "loss": 0.0454,
      "step": 896
    },
    {
      "epoch": 1.183377308707124,
      "grad_norm": 0.5033898941289409,
      "learning_rate": 4.277028821310907e-06,
      "loss": 0.0435,
      "step": 897
    },
    {
      "epoch": 1.1846965699208443,
      "grad_norm": 0.5416991303244074,
      "learning_rate": 4.265635669221622e-06,
      "loss": 0.0455,
      "step": 898
    },
    {
      "epoch": 1.1860158311345645,
      "grad_norm": 0.48088831064375376,
      "learning_rate": 4.254246412804366e-06,
      "loss": 0.0462,
      "step": 899
    },
    {
      "epoch": 1.187335092348285,
      "grad_norm": 0.5345434830020392,
      "learning_rate": 4.2428611124771184e-06,
      "loss": 0.051,
      "step": 900
    },
    {
      "epoch": 1.1886543535620053,
      "grad_norm": 0.4392194033994731,
      "learning_rate": 4.231479828636867e-06,
      "loss": 0.0377,
      "step": 901
    },
    {
      "epoch": 1.1899736147757256,
      "grad_norm": 0.4508926786816226,
      "learning_rate": 4.220102621659298e-06,
      "loss": 0.0356,
      "step": 902
    },
    {
      "epoch": 1.1912928759894459,
      "grad_norm": 0.4711436334816274,
      "learning_rate": 4.2087295518984675e-06,
      "loss": 0.0294,
      "step": 903
    },
    {
      "epoch": 1.1926121372031662,
      "grad_norm": 0.46115423803718414,
      "learning_rate": 4.197360679686489e-06,
      "loss": 0.0325,
      "step": 904
    },
    {
      "epoch": 1.1939313984168864,
      "grad_norm": 0.4330528618484114,
      "learning_rate": 4.185996065333201e-06,
      "loss": 0.0371,
      "step": 905
    },
    {
      "epoch": 1.195250659630607,
      "grad_norm": 0.5135651330330013,
      "learning_rate": 4.174635769125862e-06,
      "loss": 0.0435,
      "step": 906
    },
    {
      "epoch": 1.1965699208443272,
      "grad_norm": 0.43414421175381523,
      "learning_rate": 4.163279851328821e-06,
      "loss": 0.035,
      "step": 907
    },
    {
      "epoch": 1.1978891820580475,
      "grad_norm": 0.5253170032158321,
      "learning_rate": 4.151928372183198e-06,
      "loss": 0.0618,
      "step": 908
    },
    {
      "epoch": 1.1992084432717678,
      "grad_norm": 0.5792566627120969,
      "learning_rate": 4.1405813919065715e-06,
      "loss": 0.0477,
      "step": 909
    },
    {
      "epoch": 1.200527704485488,
      "grad_norm": 0.4846686952615566,
      "learning_rate": 4.129238970692651e-06,
      "loss": 0.0352,
      "step": 910
    },
    {
      "epoch": 1.2018469656992083,
      "grad_norm": 0.4193436838142133,
      "learning_rate": 4.11790116871096e-06,
      "loss": 0.0409,
      "step": 911
    },
    {
      "epoch": 1.2031662269129288,
      "grad_norm": 0.5357265529510714,
      "learning_rate": 4.10656804610652e-06,
      "loss": 0.0476,
      "step": 912
    },
    {
      "epoch": 1.2044854881266491,
      "grad_norm": 0.47412399206332867,
      "learning_rate": 4.09523966299953e-06,
      "loss": 0.0437,
      "step": 913
    },
    {
      "epoch": 1.2058047493403694,
      "grad_norm": 0.40662175721213795,
      "learning_rate": 4.083916079485044e-06,
      "loss": 0.034,
      "step": 914
    },
    {
      "epoch": 1.2071240105540897,
      "grad_norm": 0.3550538451157904,
      "learning_rate": 4.072597355632656e-06,
      "loss": 0.0376,
      "step": 915
    },
    {
      "epoch": 1.20844327176781,
      "grad_norm": 0.36691408572488504,
      "learning_rate": 4.061283551486185e-06,
      "loss": 0.0309,
      "step": 916
    },
    {
      "epoch": 1.2097625329815302,
      "grad_norm": 0.7508982125167554,
      "learning_rate": 4.0499747270633425e-06,
      "loss": 0.0527,
      "step": 917
    },
    {
      "epoch": 1.2110817941952507,
      "grad_norm": 0.4981336921560989,
      "learning_rate": 4.038670942355431e-06,
      "loss": 0.0569,
      "step": 918
    },
    {
      "epoch": 1.212401055408971,
      "grad_norm": 0.31515678753079435,
      "learning_rate": 4.027372257327017e-06,
      "loss": 0.0331,
      "step": 919
    },
    {
      "epoch": 1.2137203166226913,
      "grad_norm": 0.3508178905013858,
      "learning_rate": 4.016078731915608e-06,
      "loss": 0.0246,
      "step": 920
    },
    {
      "epoch": 1.2150395778364116,
      "grad_norm": 0.45178518763707604,
      "learning_rate": 4.004790426031353e-06,
      "loss": 0.0547,
      "step": 921
    },
    {
      "epoch": 1.2163588390501319,
      "grad_norm": 0.8034034003023315,
      "learning_rate": 3.993507399556699e-06,
      "loss": 0.0275,
      "step": 922
    },
    {
      "epoch": 1.2176781002638521,
      "grad_norm": 0.4725326112795675,
      "learning_rate": 3.982229712346096e-06,
      "loss": 0.0348,
      "step": 923
    },
    {
      "epoch": 1.2189973614775726,
      "grad_norm": 0.35517168294538254,
      "learning_rate": 3.970957424225666e-06,
      "loss": 0.0264,
      "step": 924
    },
    {
      "epoch": 1.220316622691293,
      "grad_norm": 0.3695274760387495,
      "learning_rate": 3.959690594992889e-06,
      "loss": 0.0289,
      "step": 925
    },
    {
      "epoch": 1.2216358839050132,
      "grad_norm": 0.42133612321865277,
      "learning_rate": 3.9484292844162905e-06,
      "loss": 0.0369,
      "step": 926
    },
    {
      "epoch": 1.2229551451187335,
      "grad_norm": 0.47170932535065363,
      "learning_rate": 3.937173552235117e-06,
      "loss": 0.0345,
      "step": 927
    },
    {
      "epoch": 1.2242744063324538,
      "grad_norm": 0.46520666681896905,
      "learning_rate": 3.925923458159023e-06,
      "loss": 0.0445,
      "step": 928
    },
    {
      "epoch": 1.225593667546174,
      "grad_norm": 0.4219364739293325,
      "learning_rate": 3.914679061867754e-06,
      "loss": 0.0341,
      "step": 929
    },
    {
      "epoch": 1.2269129287598945,
      "grad_norm": 0.49134686593171684,
      "learning_rate": 3.903440423010835e-06,
      "loss": 0.0426,
      "step": 930
    },
    {
      "epoch": 1.2282321899736148,
      "grad_norm": 0.35418526461050054,
      "learning_rate": 3.892207601207238e-06,
      "loss": 0.0328,
      "step": 931
    },
    {
      "epoch": 1.229551451187335,
      "grad_norm": 0.4155466911167312,
      "learning_rate": 3.880980656045087e-06,
      "loss": 0.0337,
      "step": 932
    },
    {
      "epoch": 1.2308707124010554,
      "grad_norm": 0.5147604758526607,
      "learning_rate": 3.869759647081326e-06,
      "loss": 0.0454,
      "step": 933
    },
    {
      "epoch": 1.2321899736147757,
      "grad_norm": 0.5061623638024214,
      "learning_rate": 3.858544633841409e-06,
      "loss": 0.0523,
      "step": 934
    },
    {
      "epoch": 1.233509234828496,
      "grad_norm": 0.43230450710501256,
      "learning_rate": 3.847335675818985e-06,
      "loss": 0.0427,
      "step": 935
    },
    {
      "epoch": 1.2348284960422165,
      "grad_norm": 0.3674699699030769,
      "learning_rate": 3.836132832475583e-06,
      "loss": 0.0332,
      "step": 936
    },
    {
      "epoch": 1.2361477572559367,
      "grad_norm": 0.41966842743992777,
      "learning_rate": 3.8249361632402884e-06,
      "loss": 0.0348,
      "step": 937
    },
    {
      "epoch": 1.237467018469657,
      "grad_norm": 0.4675730882835975,
      "learning_rate": 3.813745727509439e-06,
      "loss": 0.0263,
      "step": 938
    },
    {
      "epoch": 1.2387862796833773,
      "grad_norm": 0.5078899880579145,
      "learning_rate": 3.802561584646307e-06,
      "loss": 0.0455,
      "step": 939
    },
    {
      "epoch": 1.2401055408970976,
      "grad_norm": 0.42307371705498914,
      "learning_rate": 3.7913837939807763e-06,
      "loss": 0.0311,
      "step": 940
    },
    {
      "epoch": 1.2414248021108178,
      "grad_norm": 0.4776324071645813,
      "learning_rate": 3.7802124148090365e-06,
      "loss": 0.0387,
      "step": 941
    },
    {
      "epoch": 1.2427440633245384,
      "grad_norm": 0.506615988672597,
      "learning_rate": 3.769047506393267e-06,
      "loss": 0.0432,
      "step": 942
    },
    {
      "epoch": 1.2440633245382586,
      "grad_norm": 0.42547177727650803,
      "learning_rate": 3.757889127961316e-06,
      "loss": 0.0353,
      "step": 943
    },
    {
      "epoch": 1.245382585751979,
      "grad_norm": 0.5796557163775428,
      "learning_rate": 3.7467373387063973e-06,
      "loss": 0.0631,
      "step": 944
    },
    {
      "epoch": 1.2467018469656992,
      "grad_norm": 0.505430231894429,
      "learning_rate": 3.7355921977867672e-06,
      "loss": 0.0444,
      "step": 945
    },
    {
      "epoch": 1.2480211081794195,
      "grad_norm": 0.44127983017914596,
      "learning_rate": 3.7244537643254115e-06,
      "loss": 0.0348,
      "step": 946
    },
    {
      "epoch": 1.2493403693931397,
      "grad_norm": 0.705598008981505,
      "learning_rate": 3.7133220974097383e-06,
      "loss": 0.0596,
      "step": 947
    },
    {
      "epoch": 1.2506596306068603,
      "grad_norm": 0.5361833719044061,
      "learning_rate": 3.70219725609126e-06,
      "loss": 0.0541,
      "step": 948
    },
    {
      "epoch": 1.2519788918205805,
      "grad_norm": 0.5815052949402663,
      "learning_rate": 3.6910792993852773e-06,
      "loss": 0.0459,
      "step": 949
    },
    {
      "epoch": 1.2532981530343008,
      "grad_norm": 0.35605491319913013,
      "learning_rate": 3.679968286270571e-06,
      "loss": 0.0322,
      "step": 950
    },
    {
      "epoch": 1.254617414248021,
      "grad_norm": 0.3331663956238606,
      "learning_rate": 3.6688642756890895e-06,
      "loss": 0.0289,
      "step": 951
    },
    {
      "epoch": 1.2559366754617414,
      "grad_norm": 0.44158160096532845,
      "learning_rate": 3.6577673265456296e-06,
      "loss": 0.044,
      "step": 952
    },
    {
      "epoch": 1.2572559366754619,
      "grad_norm": 0.4746536128360755,
      "learning_rate": 3.646677497707532e-06,
      "loss": 0.0379,
      "step": 953
    },
    {
      "epoch": 1.258575197889182,
      "grad_norm": 0.4350590512773799,
      "learning_rate": 3.6355948480043647e-06,
      "loss": 0.0438,
      "step": 954
    },
    {
      "epoch": 1.2598944591029024,
      "grad_norm": 0.6315170149439323,
      "learning_rate": 3.62451943622761e-06,
      "loss": 0.0752,
      "step": 955
    },
    {
      "epoch": 1.2612137203166227,
      "grad_norm": 0.6284106690387802,
      "learning_rate": 3.6134513211303555e-06,
      "loss": 0.0422,
      "step": 956
    },
    {
      "epoch": 1.262532981530343,
      "grad_norm": 0.3910609991343049,
      "learning_rate": 3.6023905614269834e-06,
      "loss": 0.0361,
      "step": 957
    },
    {
      "epoch": 1.2638522427440633,
      "grad_norm": 0.410139517265721,
      "learning_rate": 3.5913372157928515e-06,
      "loss": 0.0434,
      "step": 958
    },
    {
      "epoch": 1.2651715039577835,
      "grad_norm": 0.42724744417074056,
      "learning_rate": 3.580291342863993e-06,
      "loss": 0.0345,
      "step": 959
    },
    {
      "epoch": 1.266490765171504,
      "grad_norm": 0.42009080909638846,
      "learning_rate": 3.569253001236795e-06,
      "loss": 0.0475,
      "step": 960
    },
    {
      "epoch": 1.2678100263852243,
      "grad_norm": 0.49743514051851584,
      "learning_rate": 3.5582222494676984e-06,
      "loss": 0.0315,
      "step": 961
    },
    {
      "epoch": 1.2691292875989446,
      "grad_norm": 0.5477142452802849,
      "learning_rate": 3.5471991460728725e-06,
      "loss": 0.0386,
      "step": 962
    },
    {
      "epoch": 1.270448548812665,
      "grad_norm": 0.41591914851810763,
      "learning_rate": 3.5361837495279217e-06,
      "loss": 0.0357,
      "step": 963
    },
    {
      "epoch": 1.2717678100263852,
      "grad_norm": 0.4996536327861138,
      "learning_rate": 3.5251761182675626e-06,
      "loss": 0.0358,
      "step": 964
    },
    {
      "epoch": 1.2730870712401057,
      "grad_norm": 0.4519451879403226,
      "learning_rate": 3.514176310685318e-06,
      "loss": 0.0427,
      "step": 965
    },
    {
      "epoch": 1.2744063324538257,
      "grad_norm": 0.5130131641395598,
      "learning_rate": 3.5031843851332105e-06,
      "loss": 0.0453,
      "step": 966
    },
    {
      "epoch": 1.2757255936675462,
      "grad_norm": 0.40396338697034684,
      "learning_rate": 3.492200399921447e-06,
      "loss": 0.0279,
      "step": 967
    },
    {
      "epoch": 1.2770448548812665,
      "grad_norm": 0.4384440979919693,
      "learning_rate": 3.481224413318114e-06,
      "loss": 0.0379,
      "step": 968
    },
    {
      "epoch": 1.2783641160949868,
      "grad_norm": 0.5529379502906039,
      "learning_rate": 3.470256483548866e-06,
      "loss": 0.0512,
      "step": 969
    },
    {
      "epoch": 1.279683377308707,
      "grad_norm": 0.48363343424476124,
      "learning_rate": 3.459296668796619e-06,
      "loss": 0.0371,
      "step": 970
    },
    {
      "epoch": 1.2810026385224274,
      "grad_norm": 0.6110356799840317,
      "learning_rate": 3.4483450272012364e-06,
      "loss": 0.0514,
      "step": 971
    },
    {
      "epoch": 1.2823218997361479,
      "grad_norm": 0.3797144655454073,
      "learning_rate": 3.4374016168592296e-06,
      "loss": 0.0323,
      "step": 972
    },
    {
      "epoch": 1.2836411609498681,
      "grad_norm": 0.386622854183218,
      "learning_rate": 3.426466495823441e-06,
      "loss": 0.0325,
      "step": 973
    },
    {
      "epoch": 1.2849604221635884,
      "grad_norm": 0.549381263129161,
      "learning_rate": 3.4155397221027396e-06,
      "loss": 0.0343,
      "step": 974
    },
    {
      "epoch": 1.2862796833773087,
      "grad_norm": 0.3809569973876317,
      "learning_rate": 3.4046213536617164e-06,
      "loss": 0.0246,
      "step": 975
    },
    {
      "epoch": 1.287598944591029,
      "grad_norm": 0.3785065948314082,
      "learning_rate": 3.393711448420372e-06,
      "loss": 0.0333,
      "step": 976
    },
    {
      "epoch": 1.2889182058047495,
      "grad_norm": 0.47345085429095884,
      "learning_rate": 3.3828100642538097e-06,
      "loss": 0.0329,
      "step": 977
    },
    {
      "epoch": 1.2902374670184695,
      "grad_norm": 0.46504857911868974,
      "learning_rate": 3.371917258991933e-06,
      "loss": 0.0468,
      "step": 978
    },
    {
      "epoch": 1.29155672823219,
      "grad_norm": 0.43387007700506586,
      "learning_rate": 3.3610330904191353e-06,
      "loss": 0.0389,
      "step": 979
    },
    {
      "epoch": 1.2928759894459103,
      "grad_norm": 0.43289132694716775,
      "learning_rate": 3.3501576162739903e-06,
      "loss": 0.0357,
      "step": 980
    },
    {
      "epoch": 1.2941952506596306,
      "grad_norm": 0.5338596807204832,
      "learning_rate": 3.339290894248954e-06,
      "loss": 0.0448,
      "step": 981
    },
    {
      "epoch": 1.2955145118733509,
      "grad_norm": 0.44703974769176946,
      "learning_rate": 3.328432981990053e-06,
      "loss": 0.045,
      "step": 982
    },
    {
      "epoch": 1.2968337730870712,
      "grad_norm": 0.4312648230012682,
      "learning_rate": 3.317583937096577e-06,
      "loss": 0.0375,
      "step": 983
    },
    {
      "epoch": 1.2981530343007917,
      "grad_norm": 0.39804375835899897,
      "learning_rate": 3.306743817120777e-06,
      "loss": 0.0305,
      "step": 984
    },
    {
      "epoch": 1.299472295514512,
      "grad_norm": 0.40547275062936783,
      "learning_rate": 3.2959126795675616e-06,
      "loss": 0.0365,
      "step": 985
    },
    {
      "epoch": 1.3007915567282322,
      "grad_norm": 0.44179488739988027,
      "learning_rate": 3.2850905818941853e-06,
      "loss": 0.0491,
      "step": 986
    },
    {
      "epoch": 1.3021108179419525,
      "grad_norm": 0.4599921621944571,
      "learning_rate": 3.2742775815099503e-06,
      "loss": 0.039,
      "step": 987
    },
    {
      "epoch": 1.3034300791556728,
      "grad_norm": 0.47203341421145645,
      "learning_rate": 3.2634737357758994e-06,
      "loss": 0.0338,
      "step": 988
    },
    {
      "epoch": 1.3047493403693933,
      "grad_norm": 0.35696717651954174,
      "learning_rate": 3.252679102004509e-06,
      "loss": 0.0405,
      "step": 989
    },
    {
      "epoch": 1.3060686015831133,
      "grad_norm": 0.41069975024427896,
      "learning_rate": 3.2418937374593895e-06,
      "loss": 0.0354,
      "step": 990
    },
    {
      "epoch": 1.3073878627968338,
      "grad_norm": 0.48133408200291883,
      "learning_rate": 3.231117699354982e-06,
      "loss": 0.0427,
      "step": 991
    },
    {
      "epoch": 1.3087071240105541,
      "grad_norm": 0.4323486452790916,
      "learning_rate": 3.220351044856247e-06,
      "loss": 0.0238,
      "step": 992
    },
    {
      "epoch": 1.3100263852242744,
      "grad_norm": 0.4824686861935317,
      "learning_rate": 3.209593831078371e-06,
      "loss": 0.0346,
      "step": 993
    },
    {
      "epoch": 1.3113456464379947,
      "grad_norm": 0.5357251409622038,
      "learning_rate": 3.198846115086459e-06,
      "loss": 0.045,
      "step": 994
    },
    {
      "epoch": 1.312664907651715,
      "grad_norm": 0.4895627677343242,
      "learning_rate": 3.188107953895229e-06,
      "loss": 0.0442,
      "step": 995
    },
    {
      "epoch": 1.3139841688654355,
      "grad_norm": 0.565268430457988,
      "learning_rate": 3.177379404468715e-06,
      "loss": 0.0454,
      "step": 996
    },
    {
      "epoch": 1.3153034300791557,
      "grad_norm": 0.4785911360758799,
      "learning_rate": 3.1666605237199625e-06,
      "loss": 0.0352,
      "step": 997
    },
    {
      "epoch": 1.316622691292876,
      "grad_norm": 0.32890904146738625,
      "learning_rate": 3.1559513685107233e-06,
      "loss": 0.0197,
      "step": 998
    },
    {
      "epoch": 1.3179419525065963,
      "grad_norm": 0.43538266074480436,
      "learning_rate": 3.145251995651162e-06,
      "loss": 0.0375,
      "step": 999
    },
    {
      "epoch": 1.3192612137203166,
      "grad_norm": 0.3726106605998871,
      "learning_rate": 3.1345624618995444e-06,
      "loss": 0.0333,
      "step": 1000
    },
    {
      "epoch": 1.320580474934037,
      "grad_norm": 1.0522326452050843,
      "learning_rate": 3.1238828239619447e-06,
      "loss": 0.0418,
      "step": 1001
    },
    {
      "epoch": 1.3218997361477571,
      "grad_norm": 0.46264860868018753,
      "learning_rate": 3.11321313849194e-06,
      "loss": 0.0376,
      "step": 1002
    },
    {
      "epoch": 1.3232189973614776,
      "grad_norm": 0.5999884424789149,
      "learning_rate": 3.102553462090314e-06,
      "loss": 0.0469,
      "step": 1003
    },
    {
      "epoch": 1.324538258575198,
      "grad_norm": 0.44953860397274076,
      "learning_rate": 3.0919038513047507e-06,
      "loss": 0.024,
      "step": 1004
    },
    {
      "epoch": 1.3258575197889182,
      "grad_norm": 0.4671921335921803,
      "learning_rate": 3.081264362629539e-06,
      "loss": 0.0343,
      "step": 1005
    },
    {
      "epoch": 1.3271767810026385,
      "grad_norm": 0.36928026939799047,
      "learning_rate": 3.070635052505273e-06,
      "loss": 0.0367,
      "step": 1006
    },
    {
      "epoch": 1.3284960422163588,
      "grad_norm": 0.46448096365916186,
      "learning_rate": 3.0600159773185506e-06,
      "loss": 0.0314,
      "step": 1007
    },
    {
      "epoch": 1.3298153034300793,
      "grad_norm": 0.5066608290740877,
      "learning_rate": 3.0494071934016737e-06,
      "loss": 0.0444,
      "step": 1008
    },
    {
      "epoch": 1.3311345646437995,
      "grad_norm": 0.40340427606125134,
      "learning_rate": 3.038808757032353e-06,
      "loss": 0.0387,
      "step": 1009
    },
    {
      "epoch": 1.3324538258575198,
      "grad_norm": 0.39352281811558804,
      "learning_rate": 3.0282207244334084e-06,
      "loss": 0.0332,
      "step": 1010
    },
    {
      "epoch": 1.33377308707124,
      "grad_norm": 0.38733414560498847,
      "learning_rate": 3.0176431517724636e-06,
      "loss": 0.0275,
      "step": 1011
    },
    {
      "epoch": 1.3350923482849604,
      "grad_norm": 0.4256685091983431,
      "learning_rate": 3.007076095161662e-06,
      "loss": 0.0473,
      "step": 1012
    },
    {
      "epoch": 1.3364116094986809,
      "grad_norm": 0.4165491156723393,
      "learning_rate": 2.996519610657356e-06,
      "loss": 0.0412,
      "step": 1013
    },
    {
      "epoch": 1.337730870712401,
      "grad_norm": 0.41360475602040825,
      "learning_rate": 2.9859737542598157e-06,
      "loss": 0.0435,
      "step": 1014
    },
    {
      "epoch": 1.3390501319261214,
      "grad_norm": 0.37936895912809304,
      "learning_rate": 2.975438581912933e-06,
      "loss": 0.0352,
      "step": 1015
    },
    {
      "epoch": 1.3403693931398417,
      "grad_norm": 0.32943810250125843,
      "learning_rate": 2.9649141495039225e-06,
      "loss": 0.03,
      "step": 1016
    },
    {
      "epoch": 1.341688654353562,
      "grad_norm": 0.49629621970844595,
      "learning_rate": 2.9544005128630217e-06,
      "loss": 0.0382,
      "step": 1017
    },
    {
      "epoch": 1.3430079155672823,
      "grad_norm": 0.6887166327460013,
      "learning_rate": 2.943897727763202e-06,
      "loss": 0.0708,
      "step": 1018
    },
    {
      "epoch": 1.3443271767810026,
      "grad_norm": 0.4093448467540715,
      "learning_rate": 2.9334058499198702e-06,
      "loss": 0.0335,
      "step": 1019
    },
    {
      "epoch": 1.345646437994723,
      "grad_norm": 0.4297548004407467,
      "learning_rate": 2.9229249349905686e-06,
      "loss": 0.0392,
      "step": 1020
    },
    {
      "epoch": 1.3469656992084433,
      "grad_norm": 0.39780686585256225,
      "learning_rate": 2.912455038574686e-06,
      "loss": 0.036,
      "step": 1021
    },
    {
      "epoch": 1.3482849604221636,
      "grad_norm": 0.3844234871086875,
      "learning_rate": 2.9019962162131564e-06,
      "loss": 0.0261,
      "step": 1022
    },
    {
      "epoch": 1.349604221635884,
      "grad_norm": 0.6872195624516678,
      "learning_rate": 2.891548523388173e-06,
      "loss": 0.0573,
      "step": 1023
    },
    {
      "epoch": 1.3509234828496042,
      "grad_norm": 0.462684090932404,
      "learning_rate": 2.8811120155228843e-06,
      "loss": 0.0514,
      "step": 1024
    },
    {
      "epoch": 1.3522427440633247,
      "grad_norm": 0.32721023602849464,
      "learning_rate": 2.870686747981108e-06,
      "loss": 0.0283,
      "step": 1025
    },
    {
      "epoch": 1.3535620052770447,
      "grad_norm": 0.506546676516316,
      "learning_rate": 2.8602727760670336e-06,
      "loss": 0.0412,
      "step": 1026
    },
    {
      "epoch": 1.3548812664907652,
      "grad_norm": 0.501844856487769,
      "learning_rate": 2.8498701550249234e-06,
      "loss": 0.0396,
      "step": 1027
    },
    {
      "epoch": 1.3562005277044855,
      "grad_norm": 0.4854469757169651,
      "learning_rate": 2.839478940038833e-06,
      "loss": 0.0359,
      "step": 1028
    },
    {
      "epoch": 1.3575197889182058,
      "grad_norm": 0.5197094676195364,
      "learning_rate": 2.8290991862323068e-06,
      "loss": 0.0436,
      "step": 1029
    },
    {
      "epoch": 1.358839050131926,
      "grad_norm": 0.45575980675496397,
      "learning_rate": 2.8187309486680924e-06,
      "loss": 0.0418,
      "step": 1030
    },
    {
      "epoch": 1.3601583113456464,
      "grad_norm": 0.5335924104959506,
      "learning_rate": 2.8083742823478423e-06,
      "loss": 0.0423,
      "step": 1031
    },
    {
      "epoch": 1.3614775725593669,
      "grad_norm": 0.6099518497682432,
      "learning_rate": 2.7980292422118282e-06,
      "loss": 0.0553,
      "step": 1032
    },
    {
      "epoch": 1.3627968337730871,
      "grad_norm": 0.45857130513461763,
      "learning_rate": 2.7876958831386487e-06,
      "loss": 0.0412,
      "step": 1033
    },
    {
      "epoch": 1.3641160949868074,
      "grad_norm": 0.4102160077304873,
      "learning_rate": 2.777374259944929e-06,
      "loss": 0.0308,
      "step": 1034
    },
    {
      "epoch": 1.3654353562005277,
      "grad_norm": 0.45878667026434056,
      "learning_rate": 2.7670644273850446e-06,
      "loss": 0.047,
      "step": 1035
    },
    {
      "epoch": 1.366754617414248,
      "grad_norm": 0.44773233254363415,
      "learning_rate": 2.7567664401508225e-06,
      "loss": 0.0472,
      "step": 1036
    },
    {
      "epoch": 1.3680738786279685,
      "grad_norm": 0.42822024536667164,
      "learning_rate": 2.7464803528712506e-06,
      "loss": 0.037,
      "step": 1037
    },
    {
      "epoch": 1.3693931398416885,
      "grad_norm": 0.6915956645143235,
      "learning_rate": 2.736206220112192e-06,
      "loss": 0.0295,
      "step": 1038
    },
    {
      "epoch": 1.370712401055409,
      "grad_norm": 0.8305607893389385,
      "learning_rate": 2.725944096376092e-06,
      "loss": 0.0391,
      "step": 1039
    },
    {
      "epoch": 1.3720316622691293,
      "grad_norm": 0.37663284086172133,
      "learning_rate": 2.7156940361016864e-06,
      "loss": 0.0298,
      "step": 1040
    },
    {
      "epoch": 1.3733509234828496,
      "grad_norm": 0.538910224000725,
      "learning_rate": 2.705456093663723e-06,
      "loss": 0.0444,
      "step": 1041
    },
    {
      "epoch": 1.3746701846965699,
      "grad_norm": 0.5319313138990732,
      "learning_rate": 2.6952303233726628e-06,
      "loss": 0.0436,
      "step": 1042
    },
    {
      "epoch": 1.3759894459102902,
      "grad_norm": 0.32666054786344273,
      "learning_rate": 2.6850167794743966e-06,
      "loss": 0.0307,
      "step": 1043
    },
    {
      "epoch": 1.3773087071240107,
      "grad_norm": 0.5617149051548302,
      "learning_rate": 2.6748155161499568e-06,
      "loss": 0.0506,
      "step": 1044
    },
    {
      "epoch": 1.378627968337731,
      "grad_norm": 0.471177550303539,
      "learning_rate": 2.664626587515231e-06,
      "loss": 0.0348,
      "step": 1045
    },
    {
      "epoch": 1.3799472295514512,
      "grad_norm": 0.4611663413099852,
      "learning_rate": 2.6544500476206675e-06,
      "loss": 0.0379,
      "step": 1046
    },
    {
      "epoch": 1.3812664907651715,
      "grad_norm": 0.7059929228923759,
      "learning_rate": 2.6442859504510005e-06,
      "loss": 0.0401,
      "step": 1047
    },
    {
      "epoch": 1.3825857519788918,
      "grad_norm": 0.4531225005491031,
      "learning_rate": 2.634134349924956e-06,
      "loss": 0.0396,
      "step": 1048
    },
    {
      "epoch": 1.383905013192612,
      "grad_norm": 0.41911814107181067,
      "learning_rate": 2.6239952998949676e-06,
      "loss": 0.0409,
      "step": 1049
    },
    {
      "epoch": 1.3852242744063323,
      "grad_norm": 0.36612524416461345,
      "learning_rate": 2.6138688541468903e-06,
      "loss": 0.0296,
      "step": 1050
    },
    {
      "epoch": 1.3865435356200528,
      "grad_norm": 0.3780056660999297,
      "learning_rate": 2.603755066399718e-06,
      "loss": 0.037,
      "step": 1051
    },
    {
      "epoch": 1.3878627968337731,
      "grad_norm": 0.46963275669637067,
      "learning_rate": 2.5936539903052893e-06,
      "loss": 0.05,
      "step": 1052
    },
    {
      "epoch": 1.3891820580474934,
      "grad_norm": 0.409153783085971,
      "learning_rate": 2.583565679448018e-06,
      "loss": 0.0247,
      "step": 1053
    },
    {
      "epoch": 1.3905013192612137,
      "grad_norm": 0.42054547753287036,
      "learning_rate": 2.573490187344596e-06,
      "loss": 0.0361,
      "step": 1054
    },
    {
      "epoch": 1.391820580474934,
      "grad_norm": 0.5255045835106068,
      "learning_rate": 2.563427567443715e-06,
      "loss": 0.0469,
      "step": 1055
    },
    {
      "epoch": 1.3931398416886545,
      "grad_norm": 0.9825521752923335,
      "learning_rate": 2.5533778731257824e-06,
      "loss": 0.0554,
      "step": 1056
    },
    {
      "epoch": 1.3944591029023747,
      "grad_norm": 0.38448801305592534,
      "learning_rate": 2.5433411577026396e-06,
      "loss": 0.0386,
      "step": 1057
    },
    {
      "epoch": 1.395778364116095,
      "grad_norm": 0.4886538622120843,
      "learning_rate": 2.5333174744172705e-06,
      "loss": 0.0451,
      "step": 1058
    },
    {
      "epoch": 1.3970976253298153,
      "grad_norm": 0.5256881739187035,
      "learning_rate": 2.523306876443532e-06,
      "loss": 0.053,
      "step": 1059
    },
    {
      "epoch": 1.3984168865435356,
      "grad_norm": 0.5884455219037508,
      "learning_rate": 2.513309416885865e-06,
      "loss": 0.0518,
      "step": 1060
    },
    {
      "epoch": 1.3997361477572559,
      "grad_norm": 0.45490704561246353,
      "learning_rate": 2.5033251487790124e-06,
      "loss": 0.0336,
      "step": 1061
    },
    {
      "epoch": 1.4010554089709761,
      "grad_norm": 0.44697786677075335,
      "learning_rate": 2.493354125087738e-06,
      "loss": 0.0424,
      "step": 1062
    },
    {
      "epoch": 1.4023746701846966,
      "grad_norm": 0.5197934850801327,
      "learning_rate": 2.48339639870655e-06,
      "loss": 0.026,
      "step": 1063
    },
    {
      "epoch": 1.403693931398417,
      "grad_norm": 0.36564648892730267,
      "learning_rate": 2.4734520224594094e-06,
      "loss": 0.0318,
      "step": 1064
    },
    {
      "epoch": 1.4050131926121372,
      "grad_norm": 0.5920632674980368,
      "learning_rate": 2.4635210490994648e-06,
      "loss": 0.0643,
      "step": 1065
    },
    {
      "epoch": 1.4063324538258575,
      "grad_norm": 0.5951808956519485,
      "learning_rate": 2.4536035313087603e-06,
      "loss": 0.0642,
      "step": 1066
    },
    {
      "epoch": 1.4076517150395778,
      "grad_norm": 0.5155676663484681,
      "learning_rate": 2.4436995216979635e-06,
      "loss": 0.036,
      "step": 1067
    },
    {
      "epoch": 1.4089709762532983,
      "grad_norm": 0.39084449761148354,
      "learning_rate": 2.4338090728060808e-06,
      "loss": 0.0399,
      "step": 1068
    },
    {
      "epoch": 1.4102902374670185,
      "grad_norm": 0.41720046914502273,
      "learning_rate": 2.4239322371001855e-06,
      "loss": 0.0434,
      "step": 1069
    },
    {
      "epoch": 1.4116094986807388,
      "grad_norm": 0.49043289317821803,
      "learning_rate": 2.414069066975128e-06,
      "loss": 0.0495,
      "step": 1070
    },
    {
      "epoch": 1.412928759894459,
      "grad_norm": 0.43929751116025734,
      "learning_rate": 2.404219614753273e-06,
      "loss": 0.0355,
      "step": 1071
    },
    {
      "epoch": 1.4142480211081794,
      "grad_norm": 0.8439643057831976,
      "learning_rate": 2.3943839326842096e-06,
      "loss": 0.0544,
      "step": 1072
    },
    {
      "epoch": 1.4155672823218997,
      "grad_norm": 0.46091464446991814,
      "learning_rate": 2.3845620729444803e-06,
      "loss": 0.0422,
      "step": 1073
    },
    {
      "epoch": 1.41688654353562,
      "grad_norm": 0.5467159719235117,
      "learning_rate": 2.3747540876373026e-06,
      "loss": 0.0515,
      "step": 1074
    },
    {
      "epoch": 1.4182058047493404,
      "grad_norm": 0.5121155616555213,
      "learning_rate": 2.364960028792292e-06,
      "loss": 0.0382,
      "step": 1075
    },
    {
      "epoch": 1.4195250659630607,
      "grad_norm": 0.5440053881184854,
      "learning_rate": 2.3551799483651894e-06,
      "loss": 0.0422,
      "step": 1076
    },
    {
      "epoch": 1.420844327176781,
      "grad_norm": 0.4357587566255143,
      "learning_rate": 2.345413898237576e-06,
      "loss": 0.043,
      "step": 1077
    },
    {
      "epoch": 1.4221635883905013,
      "grad_norm": 0.49773296255549826,
      "learning_rate": 2.335661930216611e-06,
      "loss": 0.0433,
      "step": 1078
    },
    {
      "epoch": 1.4234828496042216,
      "grad_norm": 0.5789417671457069,
      "learning_rate": 2.325924096034749e-06,
      "loss": 0.0495,
      "step": 1079
    },
    {
      "epoch": 1.424802110817942,
      "grad_norm": 0.5346328901027718,
      "learning_rate": 2.316200447349466e-06,
      "loss": 0.0422,
      "step": 1080
    },
    {
      "epoch": 1.4261213720316623,
      "grad_norm": 0.41947416672577986,
      "learning_rate": 2.306491035742987e-06,
      "loss": 0.0376,
      "step": 1081
    },
    {
      "epoch": 1.4274406332453826,
      "grad_norm": 0.5242085536717893,
      "learning_rate": 2.296795912722014e-06,
      "loss": 0.0459,
      "step": 1082
    },
    {
      "epoch": 1.428759894459103,
      "grad_norm": 0.5715780963086183,
      "learning_rate": 2.2871151297174448e-06,
      "loss": 0.0625,
      "step": 1083
    },
    {
      "epoch": 1.4300791556728232,
      "grad_norm": 0.6352727794275724,
      "learning_rate": 2.2774487380841116e-06,
      "loss": 0.0523,
      "step": 1084
    },
    {
      "epoch": 1.4313984168865435,
      "grad_norm": 0.3460910768833545,
      "learning_rate": 2.267796789100501e-06,
      "loss": 0.0333,
      "step": 1085
    },
    {
      "epoch": 1.4327176781002637,
      "grad_norm": 0.3787713556887917,
      "learning_rate": 2.258159333968484e-06,
      "loss": 0.0337,
      "step": 1086
    },
    {
      "epoch": 1.4340369393139842,
      "grad_norm": 0.4349330927269478,
      "learning_rate": 2.2485364238130435e-06,
      "loss": 0.0374,
      "step": 1087
    },
    {
      "epoch": 1.4353562005277045,
      "grad_norm": 0.5745490778847844,
      "learning_rate": 2.2389281096820077e-06,
      "loss": 0.051,
      "step": 1088
    },
    {
      "epoch": 1.4366754617414248,
      "grad_norm": 0.4307299555917101,
      "learning_rate": 2.2293344425457677e-06,
      "loss": 0.0304,
      "step": 1089
    },
    {
      "epoch": 1.437994722955145,
      "grad_norm": 0.38163317920987777,
      "learning_rate": 2.21975547329702e-06,
      "loss": 0.0398,
      "step": 1090
    },
    {
      "epoch": 1.4393139841688654,
      "grad_norm": 0.5784514783227251,
      "learning_rate": 2.210191252750495e-06,
      "loss": 0.0441,
      "step": 1091
    },
    {
      "epoch": 1.4406332453825859,
      "grad_norm": 0.3504151982802342,
      "learning_rate": 2.2006418316426773e-06,
      "loss": 0.0279,
      "step": 1092
    },
    {
      "epoch": 1.4419525065963061,
      "grad_norm": 0.37141871869746296,
      "learning_rate": 2.191107260631548e-06,
      "loss": 0.0301,
      "step": 1093
    },
    {
      "epoch": 1.4432717678100264,
      "grad_norm": 0.462488599649108,
      "learning_rate": 2.1815875902963058e-06,
      "loss": 0.0518,
      "step": 1094
    },
    {
      "epoch": 1.4445910290237467,
      "grad_norm": 0.4680050954217483,
      "learning_rate": 2.1720828711371073e-06,
      "loss": 0.0376,
      "step": 1095
    },
    {
      "epoch": 1.445910290237467,
      "grad_norm": 0.3459584036515211,
      "learning_rate": 2.1625931535747964e-06,
      "loss": 0.0446,
      "step": 1096
    },
    {
      "epoch": 1.4472295514511873,
      "grad_norm": 0.4247643610002131,
      "learning_rate": 2.1531184879506353e-06,
      "loss": 0.0343,
      "step": 1097
    },
    {
      "epoch": 1.4485488126649075,
      "grad_norm": 0.44795236208765893,
      "learning_rate": 2.1436589245260375e-06,
      "loss": 0.0383,
      "step": 1098
    },
    {
      "epoch": 1.449868073878628,
      "grad_norm": 0.3569544473987064,
      "learning_rate": 2.1342145134823055e-06,
      "loss": 0.0279,
      "step": 1099
    },
    {
      "epoch": 1.4511873350923483,
      "grad_norm": 0.48039130871149777,
      "learning_rate": 2.1247853049203543e-06,
      "loss": 0.0431,
      "step": 1100
    },
    {
      "epoch": 1.4525065963060686,
      "grad_norm": 0.4454599786746931,
      "learning_rate": 2.1153713488604584e-06,
      "loss": 0.0388,
      "step": 1101
    },
    {
      "epoch": 1.4538258575197889,
      "grad_norm": 0.4149647708392341,
      "learning_rate": 2.1059726952419782e-06,
      "loss": 0.0285,
      "step": 1102
    },
    {
      "epoch": 1.4551451187335092,
      "grad_norm": 1.07940821497593,
      "learning_rate": 2.096589393923099e-06,
      "loss": 0.0405,
      "step": 1103
    },
    {
      "epoch": 1.4564643799472297,
      "grad_norm": 0.41333463006294796,
      "learning_rate": 2.087221494680563e-06,
      "loss": 0.0335,
      "step": 1104
    },
    {
      "epoch": 1.45778364116095,
      "grad_norm": 0.47934857213428844,
      "learning_rate": 2.077869047209409e-06,
      "loss": 0.0395,
      "step": 1105
    },
    {
      "epoch": 1.4591029023746702,
      "grad_norm": 0.4301358428454764,
      "learning_rate": 2.068532101122704e-06,
      "loss": 0.0367,
      "step": 1106
    },
    {
      "epoch": 1.4604221635883905,
      "grad_norm": 0.7383799743036201,
      "learning_rate": 2.059210705951284e-06,
      "loss": 0.0358,
      "step": 1107
    },
    {
      "epoch": 1.4617414248021108,
      "grad_norm": 0.5076306495749731,
      "learning_rate": 2.0499049111434922e-06,
      "loss": 0.0395,
      "step": 1108
    },
    {
      "epoch": 1.463060686015831,
      "grad_norm": 0.6142156593697868,
      "learning_rate": 2.040614766064913e-06,
      "loss": 0.0603,
      "step": 1109
    },
    {
      "epoch": 1.4643799472295513,
      "grad_norm": 0.540003245309238,
      "learning_rate": 2.0313403199981125e-06,
      "loss": 0.0418,
      "step": 1110
    },
    {
      "epoch": 1.4656992084432718,
      "grad_norm": 0.48032030319386504,
      "learning_rate": 2.0220816221423766e-06,
      "loss": 0.0491,
      "step": 1111
    },
    {
      "epoch": 1.4670184696569921,
      "grad_norm": 0.7196975923390573,
      "learning_rate": 2.012838721613447e-06,
      "loss": 0.0409,
      "step": 1112
    },
    {
      "epoch": 1.4683377308707124,
      "grad_norm": 0.49600655875059707,
      "learning_rate": 2.0036116674432653e-06,
      "loss": 0.0445,
      "step": 1113
    },
    {
      "epoch": 1.4696569920844327,
      "grad_norm": 0.4826715832187464,
      "learning_rate": 1.9944005085797124e-06,
      "loss": 0.0447,
      "step": 1114
    },
    {
      "epoch": 1.470976253298153,
      "grad_norm": 0.4555273017972598,
      "learning_rate": 1.985205293886346e-06,
      "loss": 0.0319,
      "step": 1115
    },
    {
      "epoch": 1.4722955145118735,
      "grad_norm": 0.3522225723283794,
      "learning_rate": 1.9760260721421426e-06,
      "loss": 0.0304,
      "step": 1116
    },
    {
      "epoch": 1.4736147757255937,
      "grad_norm": 1.1574544927654264,
      "learning_rate": 1.9668628920412414e-06,
      "loss": 0.04,
      "step": 1117
    },
    {
      "epoch": 1.474934036939314,
      "grad_norm": 0.630790094221889,
      "learning_rate": 1.9577158021926774e-06,
      "loss": 0.0511,
      "step": 1118
    },
    {
      "epoch": 1.4762532981530343,
      "grad_norm": 0.6010138707763073,
      "learning_rate": 1.9485848511201356e-06,
      "loss": 0.0496,
      "step": 1119
    },
    {
      "epoch": 1.4775725593667546,
      "grad_norm": 0.46159248622529453,
      "learning_rate": 1.9394700872616856e-06,
      "loss": 0.0438,
      "step": 1120
    },
    {
      "epoch": 1.4788918205804749,
      "grad_norm": 0.33882700008461164,
      "learning_rate": 1.9303715589695276e-06,
      "loss": 0.0257,
      "step": 1121
    },
    {
      "epoch": 1.4802110817941951,
      "grad_norm": 0.6716654235718109,
      "learning_rate": 1.921289314509734e-06,
      "loss": 0.0597,
      "step": 1122
    },
    {
      "epoch": 1.4815303430079156,
      "grad_norm": 0.4586407143060212,
      "learning_rate": 1.912223402061996e-06,
      "loss": 0.0482,
      "step": 1123
    },
    {
      "epoch": 1.482849604221636,
      "grad_norm": 0.4701092667659717,
      "learning_rate": 1.9031738697193618e-06,
      "loss": 0.0381,
      "step": 1124
    },
    {
      "epoch": 1.4841688654353562,
      "grad_norm": 0.7401903083147542,
      "learning_rate": 1.894140765487991e-06,
      "loss": 0.0394,
      "step": 1125
    },
    {
      "epoch": 1.4854881266490765,
      "grad_norm": 0.8448139423441899,
      "learning_rate": 1.8851241372868938e-06,
      "loss": 0.0497,
      "step": 1126
    },
    {
      "epoch": 1.4868073878627968,
      "grad_norm": 0.4632288927244968,
      "learning_rate": 1.876124032947677e-06,
      "loss": 0.0386,
      "step": 1127
    },
    {
      "epoch": 1.4881266490765173,
      "grad_norm": 0.45762129332012513,
      "learning_rate": 1.8671405002142918e-06,
      "loss": 0.0319,
      "step": 1128
    },
    {
      "epoch": 1.4894459102902375,
      "grad_norm": 0.557904708285742,
      "learning_rate": 1.8581735867427814e-06,
      "loss": 0.0422,
      "step": 1129
    },
    {
      "epoch": 1.4907651715039578,
      "grad_norm": 0.46525204351630345,
      "learning_rate": 1.8492233401010218e-06,
      "loss": 0.0362,
      "step": 1130
    },
    {
      "epoch": 1.492084432717678,
      "grad_norm": 0.4519024069507292,
      "learning_rate": 1.8402898077684806e-06,
      "loss": 0.0352,
      "step": 1131
    },
    {
      "epoch": 1.4934036939313984,
      "grad_norm": 0.4104534322908132,
      "learning_rate": 1.831373037135955e-06,
      "loss": 0.0366,
      "step": 1132
    },
    {
      "epoch": 1.4947229551451187,
      "grad_norm": 0.5588912712416868,
      "learning_rate": 1.8224730755053271e-06,
      "loss": 0.046,
      "step": 1133
    },
    {
      "epoch": 1.496042216358839,
      "grad_norm": 0.40865740314473187,
      "learning_rate": 1.813589970089308e-06,
      "loss": 0.0403,
      "step": 1134
    },
    {
      "epoch": 1.4973614775725594,
      "grad_norm": 0.48716418995795524,
      "learning_rate": 1.8047237680111896e-06,
      "loss": 0.0333,
      "step": 1135
    },
    {
      "epoch": 1.4986807387862797,
      "grad_norm": 0.4569400934197829,
      "learning_rate": 1.7958745163045987e-06,
      "loss": 0.0418,
      "step": 1136
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.37936714163481816,
      "learning_rate": 1.7870422619132354e-06,
      "loss": 0.0379,
      "step": 1137
    },
    {
      "epoch": 1.5013192612137203,
      "grad_norm": 0.4334062619986392,
      "learning_rate": 1.778227051690639e-06,
      "loss": 0.0385,
      "step": 1138
    },
    {
      "epoch": 1.5026385224274406,
      "grad_norm": 0.38575696956056177,
      "learning_rate": 1.769428932399931e-06,
      "loss": 0.0352,
      "step": 1139
    },
    {
      "epoch": 1.503957783641161,
      "grad_norm": 0.8199548463395913,
      "learning_rate": 1.760647950713566e-06,
      "loss": 0.0486,
      "step": 1140
    },
    {
      "epoch": 1.5052770448548811,
      "grad_norm": 0.708932848202034,
      "learning_rate": 1.7518841532130903e-06,
      "loss": 0.035,
      "step": 1141
    },
    {
      "epoch": 1.5065963060686016,
      "grad_norm": 0.7190192350306006,
      "learning_rate": 1.74313758638889e-06,
      "loss": 0.0547,
      "step": 1142
    },
    {
      "epoch": 1.507915567282322,
      "grad_norm": 0.5802755650204077,
      "learning_rate": 1.7344082966399417e-06,
      "loss": 0.0408,
      "step": 1143
    },
    {
      "epoch": 1.5092348284960422,
      "grad_norm": 0.5244149473179786,
      "learning_rate": 1.7256963302735752e-06,
      "loss": 0.0374,
      "step": 1144
    },
    {
      "epoch": 1.5105540897097627,
      "grad_norm": 0.5014348404012459,
      "learning_rate": 1.7170017335052207e-06,
      "loss": 0.0405,
      "step": 1145
    },
    {
      "epoch": 1.5118733509234827,
      "grad_norm": 0.37500709644274266,
      "learning_rate": 1.7083245524581666e-06,
      "loss": 0.032,
      "step": 1146
    },
    {
      "epoch": 1.5131926121372032,
      "grad_norm": 0.49025529294811243,
      "learning_rate": 1.6996648331633126e-06,
      "loss": 0.0524,
      "step": 1147
    },
    {
      "epoch": 1.5145118733509235,
      "grad_norm": 0.6564367602612013,
      "learning_rate": 1.6910226215589303e-06,
      "loss": 0.059,
      "step": 1148
    },
    {
      "epoch": 1.5158311345646438,
      "grad_norm": 0.4820037232796035,
      "learning_rate": 1.6823979634904104e-06,
      "loss": 0.0292,
      "step": 1149
    },
    {
      "epoch": 1.517150395778364,
      "grad_norm": 0.46343906367583637,
      "learning_rate": 1.6737909047100292e-06,
      "loss": 0.0489,
      "step": 1150
    },
    {
      "epoch": 1.5184696569920844,
      "grad_norm": 2.0598255401544385,
      "learning_rate": 1.6652014908767016e-06,
      "loss": 0.051,
      "step": 1151
    },
    {
      "epoch": 1.5197889182058049,
      "grad_norm": 0.8097910851993284,
      "learning_rate": 1.6566297675557392e-06,
      "loss": 0.0583,
      "step": 1152
    },
    {
      "epoch": 1.521108179419525,
      "grad_norm": 0.48356131039133166,
      "learning_rate": 1.648075780218607e-06,
      "loss": 0.0466,
      "step": 1153
    },
    {
      "epoch": 1.5224274406332454,
      "grad_norm": 0.4873772171600058,
      "learning_rate": 1.6395395742426873e-06,
      "loss": 0.0461,
      "step": 1154
    },
    {
      "epoch": 1.5237467018469657,
      "grad_norm": 0.3879805072219005,
      "learning_rate": 1.63102119491103e-06,
      "loss": 0.0425,
      "step": 1155
    },
    {
      "epoch": 1.525065963060686,
      "grad_norm": 0.862053634040402,
      "learning_rate": 1.6225206874121219e-06,
      "loss": 0.0419,
      "step": 1156
    },
    {
      "epoch": 1.5263852242744065,
      "grad_norm": 1.0214659212719512,
      "learning_rate": 1.6140380968396418e-06,
      "loss": 0.0617,
      "step": 1157
    },
    {
      "epoch": 1.5277044854881265,
      "grad_norm": 0.45324201742786346,
      "learning_rate": 1.6055734681922225e-06,
      "loss": 0.0335,
      "step": 1158
    },
    {
      "epoch": 1.529023746701847,
      "grad_norm": 0.5472418519896244,
      "learning_rate": 1.5971268463732126e-06,
      "loss": 0.0438,
      "step": 1159
    },
    {
      "epoch": 1.5303430079155673,
      "grad_norm": 0.48629761142198785,
      "learning_rate": 1.588698276190438e-06,
      "loss": 0.0403,
      "step": 1160
    },
    {
      "epoch": 1.5316622691292876,
      "grad_norm": 0.39348202971703883,
      "learning_rate": 1.5802878023559598e-06,
      "loss": 0.0271,
      "step": 1161
    },
    {
      "epoch": 1.5329815303430079,
      "grad_norm": 0.40738852272834497,
      "learning_rate": 1.5718954694858457e-06,
      "loss": 0.0314,
      "step": 1162
    },
    {
      "epoch": 1.5343007915567282,
      "grad_norm": 0.7127989672912001,
      "learning_rate": 1.5635213220999279e-06,
      "loss": 0.0376,
      "step": 1163
    },
    {
      "epoch": 1.5356200527704487,
      "grad_norm": 0.5634050350048752,
      "learning_rate": 1.555165404621567e-06,
      "loss": 0.0385,
      "step": 1164
    },
    {
      "epoch": 1.5369393139841687,
      "grad_norm": 0.39618676604739544,
      "learning_rate": 1.546827761377418e-06,
      "loss": 0.0349,
      "step": 1165
    },
    {
      "epoch": 1.5382585751978892,
      "grad_norm": 0.5127069648860455,
      "learning_rate": 1.5385084365971947e-06,
      "loss": 0.0328,
      "step": 1166
    },
    {
      "epoch": 1.5395778364116095,
      "grad_norm": 0.48381020413982256,
      "learning_rate": 1.530207474413431e-06,
      "loss": 0.0216,
      "step": 1167
    },
    {
      "epoch": 1.5408970976253298,
      "grad_norm": 0.5403500805062056,
      "learning_rate": 1.5219249188612556e-06,
      "loss": 0.0423,
      "step": 1168
    },
    {
      "epoch": 1.5422163588390503,
      "grad_norm": 0.7163076289045839,
      "learning_rate": 1.51366081387815e-06,
      "loss": 0.0728,
      "step": 1169
    },
    {
      "epoch": 1.5435356200527703,
      "grad_norm": 0.4615364027150893,
      "learning_rate": 1.5054152033037206e-06,
      "loss": 0.0391,
      "step": 1170
    },
    {
      "epoch": 1.5448548812664908,
      "grad_norm": 0.4966279274667812,
      "learning_rate": 1.4971881308794633e-06,
      "loss": 0.0286,
      "step": 1171
    },
    {
      "epoch": 1.5461741424802111,
      "grad_norm": 0.6445822769969329,
      "learning_rate": 1.488979640248534e-06,
      "loss": 0.0383,
      "step": 1172
    },
    {
      "epoch": 1.5474934036939314,
      "grad_norm": 0.5511797875272039,
      "learning_rate": 1.4807897749555112e-06,
      "loss": 0.0319,
      "step": 1173
    },
    {
      "epoch": 1.5488126649076517,
      "grad_norm": 0.4028896054759746,
      "learning_rate": 1.4726185784461726e-06,
      "loss": 0.0376,
      "step": 1174
    },
    {
      "epoch": 1.550131926121372,
      "grad_norm": 0.4373971805377109,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.0346,
      "step": 1175
    },
    {
      "epoch": 1.5514511873350925,
      "grad_norm": 0.3813066629424425,
      "learning_rate": 1.4563323650662586e-06,
      "loss": 0.0313,
      "step": 1176
    },
    {
      "epoch": 1.5527704485488125,
      "grad_norm": 0.4811993173573029,
      "learning_rate": 1.4482174345911448e-06,
      "loss": 0.0485,
      "step": 1177
    },
    {
      "epoch": 1.554089709762533,
      "grad_norm": 0.46517591321514745,
      "learning_rate": 1.440121345690182e-06,
      "loss": 0.0347,
      "step": 1178
    },
    {
      "epoch": 1.5554089709762533,
      "grad_norm": 0.4512297860445867,
      "learning_rate": 1.432044141311683e-06,
      "loss": 0.0477,
      "step": 1179
    },
    {
      "epoch": 1.5567282321899736,
      "grad_norm": 0.5362718876076132,
      "learning_rate": 1.4239858643037753e-06,
      "loss": 0.0342,
      "step": 1180
    },
    {
      "epoch": 1.558047493403694,
      "grad_norm": 0.6080795983560243,
      "learning_rate": 1.415946557414185e-06,
      "loss": 0.046,
      "step": 1181
    },
    {
      "epoch": 1.5593667546174141,
      "grad_norm": 0.5075180443267415,
      "learning_rate": 1.4079262632900048e-06,
      "loss": 0.0405,
      "step": 1182
    },
    {
      "epoch": 1.5606860158311346,
      "grad_norm": 0.6975250836456051,
      "learning_rate": 1.399925024477466e-06,
      "loss": 0.0371,
      "step": 1183
    },
    {
      "epoch": 1.562005277044855,
      "grad_norm": 0.48944834666994125,
      "learning_rate": 1.3919428834217163e-06,
      "loss": 0.0435,
      "step": 1184
    },
    {
      "epoch": 1.5633245382585752,
      "grad_norm": 0.35255734243255743,
      "learning_rate": 1.3839798824665952e-06,
      "loss": 0.0415,
      "step": 1185
    },
    {
      "epoch": 1.5646437994722955,
      "grad_norm": 0.404850276527209,
      "learning_rate": 1.3760360638544012e-06,
      "loss": 0.0361,
      "step": 1186
    },
    {
      "epoch": 1.5659630606860158,
      "grad_norm": 0.6597734162030678,
      "learning_rate": 1.3681114697256792e-06,
      "loss": 0.0454,
      "step": 1187
    },
    {
      "epoch": 1.5672823218997363,
      "grad_norm": 0.5766222487962566,
      "learning_rate": 1.3602061421189899e-06,
      "loss": 0.0416,
      "step": 1188
    },
    {
      "epoch": 1.5686015831134563,
      "grad_norm": 0.4595353875796206,
      "learning_rate": 1.3523201229706901e-06,
      "loss": 0.0327,
      "step": 1189
    },
    {
      "epoch": 1.5699208443271768,
      "grad_norm": 0.33343538700583003,
      "learning_rate": 1.3444534541147058e-06,
      "loss": 0.0217,
      "step": 1190
    },
    {
      "epoch": 1.571240105540897,
      "grad_norm": 0.9071606502873505,
      "learning_rate": 1.3366061772823175e-06,
      "loss": 0.0551,
      "step": 1191
    },
    {
      "epoch": 1.5725593667546174,
      "grad_norm": 0.3435762760150652,
      "learning_rate": 1.3287783341019278e-06,
      "loss": 0.0236,
      "step": 1192
    },
    {
      "epoch": 1.5738786279683379,
      "grad_norm": 0.6241238414396604,
      "learning_rate": 1.3209699660988528e-06,
      "loss": 0.0438,
      "step": 1193
    },
    {
      "epoch": 1.575197889182058,
      "grad_norm": 0.40166273377163964,
      "learning_rate": 1.3131811146950946e-06,
      "loss": 0.041,
      "step": 1194
    },
    {
      "epoch": 1.5765171503957784,
      "grad_norm": 0.46253300012043386,
      "learning_rate": 1.3054118212091231e-06,
      "loss": 0.0302,
      "step": 1195
    },
    {
      "epoch": 1.5778364116094987,
      "grad_norm": 0.40714809268629054,
      "learning_rate": 1.2976621268556571e-06,
      "loss": 0.0407,
      "step": 1196
    },
    {
      "epoch": 1.579155672823219,
      "grad_norm": 0.32452512025205105,
      "learning_rate": 1.2899320727454472e-06,
      "loss": 0.0312,
      "step": 1197
    },
    {
      "epoch": 1.5804749340369393,
      "grad_norm": 0.4057644473007606,
      "learning_rate": 1.2822216998850506e-06,
      "loss": 0.0351,
      "step": 1198
    },
    {
      "epoch": 1.5817941952506596,
      "grad_norm": 0.5294484976583397,
      "learning_rate": 1.274531049176625e-06,
      "loss": 0.0397,
      "step": 1199
    },
    {
      "epoch": 1.58311345646438,
      "grad_norm": 0.5249033845880472,
      "learning_rate": 1.2668601614177017e-06,
      "loss": 0.0511,
      "step": 1200
    },
    {
      "epoch": 1.5844327176781001,
      "grad_norm": 0.5286195227893682,
      "learning_rate": 1.2592090773009757e-06,
      "loss": 0.0362,
      "step": 1201
    },
    {
      "epoch": 1.5857519788918206,
      "grad_norm": 0.43587737534445187,
      "learning_rate": 1.2515778374140858e-06,
      "loss": 0.0345,
      "step": 1202
    },
    {
      "epoch": 1.587071240105541,
      "grad_norm": 0.6504691814010135,
      "learning_rate": 1.2439664822394027e-06,
      "loss": 0.0468,
      "step": 1203
    },
    {
      "epoch": 1.5883905013192612,
      "grad_norm": 1.6349057673428626,
      "learning_rate": 1.2363750521538064e-06,
      "loss": 0.027,
      "step": 1204
    },
    {
      "epoch": 1.5897097625329817,
      "grad_norm": 0.5144142135467555,
      "learning_rate": 1.2288035874284847e-06,
      "loss": 0.0522,
      "step": 1205
    },
    {
      "epoch": 1.5910290237467017,
      "grad_norm": 0.7165703997721268,
      "learning_rate": 1.2212521282287093e-06,
      "loss": 0.0432,
      "step": 1206
    },
    {
      "epoch": 1.5923482849604222,
      "grad_norm": 0.5764139280217195,
      "learning_rate": 1.2137207146136265e-06,
      "loss": 0.0441,
      "step": 1207
    },
    {
      "epoch": 1.5936675461741425,
      "grad_norm": 0.38788539694879753,
      "learning_rate": 1.2062093865360458e-06,
      "loss": 0.0317,
      "step": 1208
    },
    {
      "epoch": 1.5949868073878628,
      "grad_norm": 0.4313898311629535,
      "learning_rate": 1.1987181838422252e-06,
      "loss": 0.0395,
      "step": 1209
    },
    {
      "epoch": 1.596306068601583,
      "grad_norm": 0.4785561711686053,
      "learning_rate": 1.1912471462716596e-06,
      "loss": 0.0349,
      "step": 1210
    },
    {
      "epoch": 1.5976253298153034,
      "grad_norm": 0.33015077804663906,
      "learning_rate": 1.1837963134568748e-06,
      "loss": 0.0239,
      "step": 1211
    },
    {
      "epoch": 1.5989445910290239,
      "grad_norm": 0.39922911811811035,
      "learning_rate": 1.1763657249232107e-06,
      "loss": 0.0298,
      "step": 1212
    },
    {
      "epoch": 1.600263852242744,
      "grad_norm": 0.45738013269687233,
      "learning_rate": 1.1689554200886183e-06,
      "loss": 0.0447,
      "step": 1213
    },
    {
      "epoch": 1.6015831134564644,
      "grad_norm": 0.5046432642457959,
      "learning_rate": 1.1615654382634444e-06,
      "loss": 0.0394,
      "step": 1214
    },
    {
      "epoch": 1.6029023746701847,
      "grad_norm": 0.49722274611796874,
      "learning_rate": 1.1541958186502288e-06,
      "loss": 0.0397,
      "step": 1215
    },
    {
      "epoch": 1.604221635883905,
      "grad_norm": 0.4384313895762189,
      "learning_rate": 1.146846600343488e-06,
      "loss": 0.0354,
      "step": 1216
    },
    {
      "epoch": 1.6055408970976255,
      "grad_norm": 0.3312059624947744,
      "learning_rate": 1.1395178223295188e-06,
      "loss": 0.0278,
      "step": 1217
    },
    {
      "epoch": 1.6068601583113455,
      "grad_norm": 0.4382612520669255,
      "learning_rate": 1.132209523486184e-06,
      "loss": 0.0368,
      "step": 1218
    },
    {
      "epoch": 1.608179419525066,
      "grad_norm": 0.5343958276159149,
      "learning_rate": 1.1249217425827063e-06,
      "loss": 0.0474,
      "step": 1219
    },
    {
      "epoch": 1.6094986807387863,
      "grad_norm": 0.49691100959061263,
      "learning_rate": 1.1176545182794674e-06,
      "loss": 0.0483,
      "step": 1220
    },
    {
      "epoch": 1.6108179419525066,
      "grad_norm": 0.4163737449098859,
      "learning_rate": 1.1104078891277981e-06,
      "loss": 0.0463,
      "step": 1221
    },
    {
      "epoch": 1.6121372031662269,
      "grad_norm": 0.48518725073291535,
      "learning_rate": 1.1031818935697763e-06,
      "loss": 0.0322,
      "step": 1222
    },
    {
      "epoch": 1.6134564643799472,
      "grad_norm": 0.7204476449881705,
      "learning_rate": 1.0959765699380204e-06,
      "loss": 0.0421,
      "step": 1223
    },
    {
      "epoch": 1.6147757255936677,
      "grad_norm": 0.5063474602550613,
      "learning_rate": 1.0887919564554893e-06,
      "loss": 0.0272,
      "step": 1224
    },
    {
      "epoch": 1.6160949868073877,
      "grad_norm": 0.4076166371752053,
      "learning_rate": 1.08162809123528e-06,
      "loss": 0.0309,
      "step": 1225
    },
    {
      "epoch": 1.6174142480211082,
      "grad_norm": 0.3268585941272018,
      "learning_rate": 1.0744850122804218e-06,
      "loss": 0.0319,
      "step": 1226
    },
    {
      "epoch": 1.6187335092348285,
      "grad_norm": 0.5850297589948477,
      "learning_rate": 1.0673627574836753e-06,
      "loss": 0.0382,
      "step": 1227
    },
    {
      "epoch": 1.6200527704485488,
      "grad_norm": 0.3998474874889947,
      "learning_rate": 1.0602613646273374e-06,
      "loss": 0.0319,
      "step": 1228
    },
    {
      "epoch": 1.6213720316622693,
      "grad_norm": 0.5798879351459741,
      "learning_rate": 1.0531808713830288e-06,
      "loss": 0.0667,
      "step": 1229
    },
    {
      "epoch": 1.6226912928759893,
      "grad_norm": 0.3914344031230687,
      "learning_rate": 1.046121315311508e-06,
      "loss": 0.0362,
      "step": 1230
    },
    {
      "epoch": 1.6240105540897098,
      "grad_norm": 0.49431640263561566,
      "learning_rate": 1.0390827338624622e-06,
      "loss": 0.0483,
      "step": 1231
    },
    {
      "epoch": 1.6253298153034301,
      "grad_norm": 0.4524775925868054,
      "learning_rate": 1.0320651643743128e-06,
      "loss": 0.0398,
      "step": 1232
    },
    {
      "epoch": 1.6266490765171504,
      "grad_norm": 0.5335799480206387,
      "learning_rate": 1.0250686440740177e-06,
      "loss": 0.0353,
      "step": 1233
    },
    {
      "epoch": 1.6279683377308707,
      "grad_norm": 0.46174733513990734,
      "learning_rate": 1.0180932100768714e-06,
      "loss": 0.0446,
      "step": 1234
    },
    {
      "epoch": 1.629287598944591,
      "grad_norm": 0.6857356411894094,
      "learning_rate": 1.0111388993863069e-06,
      "loss": 0.0362,
      "step": 1235
    },
    {
      "epoch": 1.6306068601583115,
      "grad_norm": 0.4245217775278478,
      "learning_rate": 1.0042057488937067e-06,
      "loss": 0.0355,
      "step": 1236
    },
    {
      "epoch": 1.6319261213720315,
      "grad_norm": 0.4322128765034416,
      "learning_rate": 9.972937953781985e-07,
      "loss": 0.0269,
      "step": 1237
    },
    {
      "epoch": 1.633245382585752,
      "grad_norm": 0.3973851330120696,
      "learning_rate": 9.904030755064659e-07,
      "loss": 0.0351,
      "step": 1238
    },
    {
      "epoch": 1.6345646437994723,
      "grad_norm": 0.46357405653390477,
      "learning_rate": 9.835336258325507e-07,
      "loss": 0.0329,
      "step": 1239
    },
    {
      "epoch": 1.6358839050131926,
      "grad_norm": 0.3994338842065798,
      "learning_rate": 9.76685482797662e-07,
      "loss": 0.0399,
      "step": 1240
    },
    {
      "epoch": 1.637203166226913,
      "grad_norm": 0.6076537453224168,
      "learning_rate": 9.69858682729976e-07,
      "loss": 0.0425,
      "step": 1241
    },
    {
      "epoch": 1.6385224274406331,
      "grad_norm": 0.37706954173942336,
      "learning_rate": 9.630532618444532e-07,
      "loss": 0.0284,
      "step": 1242
    },
    {
      "epoch": 1.6398416886543536,
      "grad_norm": 0.43604541408035125,
      "learning_rate": 9.562692562426408e-07,
      "loss": 0.0362,
      "step": 1243
    },
    {
      "epoch": 1.641160949868074,
      "grad_norm": 1.8554215408377712,
      "learning_rate": 9.495067019124793e-07,
      "loss": 0.027,
      "step": 1244
    },
    {
      "epoch": 1.6424802110817942,
      "grad_norm": 0.493805099763965,
      "learning_rate": 9.427656347281155e-07,
      "loss": 0.0356,
      "step": 1245
    },
    {
      "epoch": 1.6437994722955145,
      "grad_norm": 0.52867853062248,
      "learning_rate": 9.360460904497132e-07,
      "loss": 0.0452,
      "step": 1246
    },
    {
      "epoch": 1.6451187335092348,
      "grad_norm": 0.45808427819318753,
      "learning_rate": 9.29348104723255e-07,
      "loss": 0.0499,
      "step": 1247
    },
    {
      "epoch": 1.6464379947229553,
      "grad_norm": 0.4593564332792896,
      "learning_rate": 9.226717130803636e-07,
      "loss": 0.0427,
      "step": 1248
    },
    {
      "epoch": 1.6477572559366753,
      "grad_norm": 0.39012445742493634,
      "learning_rate": 9.160169509381083e-07,
      "loss": 0.041,
      "step": 1249
    },
    {
      "epoch": 1.6490765171503958,
      "grad_norm": 0.3868641320308475,
      "learning_rate": 9.093838535988181e-07,
      "loss": 0.035,
      "step": 1250
    },
    {
      "epoch": 1.650395778364116,
      "grad_norm": 0.33099630937478597,
      "learning_rate": 9.027724562498929e-07,
      "loss": 0.0285,
      "step": 1251
    },
    {
      "epoch": 1.6517150395778364,
      "grad_norm": 0.7114657305985929,
      "learning_rate": 8.961827939636198e-07,
      "loss": 0.0455,
      "step": 1252
    },
    {
      "epoch": 1.6530343007915569,
      "grad_norm": 0.34348560286066804,
      "learning_rate": 8.896149016969812e-07,
      "loss": 0.0323,
      "step": 1253
    },
    {
      "epoch": 1.654353562005277,
      "grad_norm": 0.4286068142548502,
      "learning_rate": 8.830688142914783e-07,
      "loss": 0.0278,
      "step": 1254
    },
    {
      "epoch": 1.6556728232189974,
      "grad_norm": 0.30546300488945166,
      "learning_rate": 8.765445664729383e-07,
      "loss": 0.0272,
      "step": 1255
    },
    {
      "epoch": 1.6569920844327177,
      "grad_norm": 0.31479895341967534,
      "learning_rate": 8.700421928513353e-07,
      "loss": 0.0219,
      "step": 1256
    },
    {
      "epoch": 1.658311345646438,
      "grad_norm": 0.4793459625577021,
      "learning_rate": 8.635617279206027e-07,
      "loss": 0.0566,
      "step": 1257
    },
    {
      "epoch": 1.6596306068601583,
      "grad_norm": 0.44954609692767467,
      "learning_rate": 8.571032060584555e-07,
      "loss": 0.0362,
      "step": 1258
    },
    {
      "epoch": 1.6609498680738786,
      "grad_norm": 0.4625040531332171,
      "learning_rate": 8.506666615261988e-07,
      "loss": 0.0327,
      "step": 1259
    },
    {
      "epoch": 1.662269129287599,
      "grad_norm": 0.44320479928975837,
      "learning_rate": 8.442521284685573e-07,
      "loss": 0.0383,
      "step": 1260
    },
    {
      "epoch": 1.6635883905013191,
      "grad_norm": 0.44263485851252193,
      "learning_rate": 8.378596409134854e-07,
      "loss": 0.0449,
      "step": 1261
    },
    {
      "epoch": 1.6649076517150396,
      "grad_norm": 0.588159347845755,
      "learning_rate": 8.314892327719937e-07,
      "loss": 0.044,
      "step": 1262
    },
    {
      "epoch": 1.66622691292876,
      "grad_norm": 0.35939392465672665,
      "learning_rate": 8.251409378379638e-07,
      "loss": 0.0264,
      "step": 1263
    },
    {
      "epoch": 1.6675461741424802,
      "grad_norm": 0.5671191712624316,
      "learning_rate": 8.188147897879667e-07,
      "loss": 0.0517,
      "step": 1264
    },
    {
      "epoch": 1.6688654353562007,
      "grad_norm": 0.358886847852907,
      "learning_rate": 8.125108221810935e-07,
      "loss": 0.0278,
      "step": 1265
    },
    {
      "epoch": 1.6701846965699207,
      "grad_norm": 0.42822626836193234,
      "learning_rate": 8.062290684587698e-07,
      "loss": 0.0416,
      "step": 1266
    },
    {
      "epoch": 1.6715039577836412,
      "grad_norm": 0.471691138664079,
      "learning_rate": 7.999695619445807e-07,
      "loss": 0.0384,
      "step": 1267
    },
    {
      "epoch": 1.6728232189973615,
      "grad_norm": 0.4307799868335856,
      "learning_rate": 7.937323358440935e-07,
      "loss": 0.0449,
      "step": 1268
    },
    {
      "epoch": 1.6741424802110818,
      "grad_norm": 0.4152109039926412,
      "learning_rate": 7.875174232446842e-07,
      "loss": 0.0265,
      "step": 1269
    },
    {
      "epoch": 1.675461741424802,
      "grad_norm": 0.43720557942166705,
      "learning_rate": 7.813248571153542e-07,
      "loss": 0.0321,
      "step": 1270
    },
    {
      "epoch": 1.6767810026385224,
      "grad_norm": 0.41610472560617734,
      "learning_rate": 7.75154670306566e-07,
      "loss": 0.0306,
      "step": 1271
    },
    {
      "epoch": 1.6781002638522429,
      "grad_norm": 0.4877701148340018,
      "learning_rate": 7.690068955500623e-07,
      "loss": 0.0343,
      "step": 1272
    },
    {
      "epoch": 1.679419525065963,
      "grad_norm": 0.604477161363688,
      "learning_rate": 7.628815654586935e-07,
      "loss": 0.0384,
      "step": 1273
    },
    {
      "epoch": 1.6807387862796834,
      "grad_norm": 0.9327966953078201,
      "learning_rate": 7.567787125262449e-07,
      "loss": 0.044,
      "step": 1274
    },
    {
      "epoch": 1.6820580474934037,
      "grad_norm": 0.49776099980226146,
      "learning_rate": 7.506983691272663e-07,
      "loss": 0.0368,
      "step": 1275
    },
    {
      "epoch": 1.683377308707124,
      "grad_norm": 0.488388274257174,
      "learning_rate": 7.446405675168938e-07,
      "loss": 0.0596,
      "step": 1276
    },
    {
      "epoch": 1.6846965699208445,
      "grad_norm": 0.35375559159393377,
      "learning_rate": 7.386053398306886e-07,
      "loss": 0.0307,
      "step": 1277
    },
    {
      "epoch": 1.6860158311345645,
      "grad_norm": 0.35309069441544605,
      "learning_rate": 7.325927180844589e-07,
      "loss": 0.0293,
      "step": 1278
    },
    {
      "epoch": 1.687335092348285,
      "grad_norm": 0.4344067842283927,
      "learning_rate": 7.266027341740917e-07,
      "loss": 0.0392,
      "step": 1279
    },
    {
      "epoch": 1.6886543535620053,
      "grad_norm": 0.506956664574459,
      "learning_rate": 7.206354198753862e-07,
      "loss": 0.0368,
      "step": 1280
    },
    {
      "epoch": 1.6899736147757256,
      "grad_norm": 1.4901594079060698,
      "learning_rate": 7.146908068438818e-07,
      "loss": 0.0559,
      "step": 1281
    },
    {
      "epoch": 1.6912928759894459,
      "grad_norm": 0.5192684059622095,
      "learning_rate": 7.087689266146935e-07,
      "loss": 0.0339,
      "step": 1282
    },
    {
      "epoch": 1.6926121372031662,
      "grad_norm": 0.631957801720068,
      "learning_rate": 7.028698106023396e-07,
      "loss": 0.0395,
      "step": 1283
    },
    {
      "epoch": 1.6939313984168867,
      "grad_norm": 0.5294491638321744,
      "learning_rate": 6.969934901005809e-07,
      "loss": 0.0288,
      "step": 1284
    },
    {
      "epoch": 1.6952506596306067,
      "grad_norm": 0.38669922322311945,
      "learning_rate": 6.911399962822518e-07,
      "loss": 0.0358,
      "step": 1285
    },
    {
      "epoch": 1.6965699208443272,
      "grad_norm": 0.434423689466899,
      "learning_rate": 6.853093601990946e-07,
      "loss": 0.0365,
      "step": 1286
    },
    {
      "epoch": 1.6978891820580475,
      "grad_norm": 0.3442245847955821,
      "learning_rate": 6.795016127815957e-07,
      "loss": 0.0315,
      "step": 1287
    },
    {
      "epoch": 1.6992084432717678,
      "grad_norm": 0.42198219872418546,
      "learning_rate": 6.737167848388227e-07,
      "loss": 0.0388,
      "step": 1288
    },
    {
      "epoch": 1.7005277044854883,
      "grad_norm": 0.5703992735819093,
      "learning_rate": 6.679549070582547e-07,
      "loss": 0.0506,
      "step": 1289
    },
    {
      "epoch": 1.7018469656992083,
      "grad_norm": 0.7031877797192093,
      "learning_rate": 6.622160100056296e-07,
      "loss": 0.0319,
      "step": 1290
    },
    {
      "epoch": 1.7031662269129288,
      "grad_norm": 0.48385734835300775,
      "learning_rate": 6.565001241247743e-07,
      "loss": 0.0418,
      "step": 1291
    },
    {
      "epoch": 1.7044854881266491,
      "grad_norm": 0.44649883806780155,
      "learning_rate": 6.508072797374454e-07,
      "loss": 0.0441,
      "step": 1292
    },
    {
      "epoch": 1.7058047493403694,
      "grad_norm": 0.3371838135534027,
      "learning_rate": 6.451375070431687e-07,
      "loss": 0.0293,
      "step": 1293
    },
    {
      "epoch": 1.7071240105540897,
      "grad_norm": 0.524737836501876,
      "learning_rate": 6.394908361190804e-07,
      "loss": 0.0301,
      "step": 1294
    },
    {
      "epoch": 1.70844327176781,
      "grad_norm": 0.42754210803421466,
      "learning_rate": 6.338672969197623e-07,
      "loss": 0.0296,
      "step": 1295
    },
    {
      "epoch": 1.7097625329815305,
      "grad_norm": 0.402266180002818,
      "learning_rate": 6.282669192770896e-07,
      "loss": 0.0372,
      "step": 1296
    },
    {
      "epoch": 1.7110817941952505,
      "grad_norm": 0.49238128867338316,
      "learning_rate": 6.226897329000687e-07,
      "loss": 0.0422,
      "step": 1297
    },
    {
      "epoch": 1.712401055408971,
      "grad_norm": 0.4021292733622415,
      "learning_rate": 6.171357673746798e-07,
      "loss": 0.0312,
      "step": 1298
    },
    {
      "epoch": 1.7137203166226913,
      "grad_norm": 0.4248158425984774,
      "learning_rate": 6.116050521637218e-07,
      "loss": 0.04,
      "step": 1299
    },
    {
      "epoch": 1.7150395778364116,
      "grad_norm": 0.7473101327400671,
      "learning_rate": 6.060976166066546e-07,
      "loss": 0.0468,
      "step": 1300
    },
    {
      "epoch": 1.716358839050132,
      "grad_norm": 0.35824922702307355,
      "learning_rate": 6.006134899194421e-07,
      "loss": 0.0312,
      "step": 1301
    },
    {
      "epoch": 1.7176781002638521,
      "grad_norm": 0.5925344230542687,
      "learning_rate": 5.951527011944008e-07,
      "loss": 0.0427,
      "step": 1302
    },
    {
      "epoch": 1.7189973614775726,
      "grad_norm": 0.405481695676839,
      "learning_rate": 5.897152794000421e-07,
      "loss": 0.0339,
      "step": 1303
    },
    {
      "epoch": 1.720316622691293,
      "grad_norm": 0.4482716029928217,
      "learning_rate": 5.843012533809211e-07,
      "loss": 0.038,
      "step": 1304
    },
    {
      "epoch": 1.7216358839050132,
      "grad_norm": 0.3595441867962531,
      "learning_rate": 5.789106518574816e-07,
      "loss": 0.0265,
      "step": 1305
    },
    {
      "epoch": 1.7229551451187335,
      "grad_norm": 0.599406764927074,
      "learning_rate": 5.735435034259057e-07,
      "loss": 0.0493,
      "step": 1306
    },
    {
      "epoch": 1.7242744063324538,
      "grad_norm": 0.40663759792765514,
      "learning_rate": 5.681998365579594e-07,
      "loss": 0.0241,
      "step": 1307
    },
    {
      "epoch": 1.7255936675461743,
      "grad_norm": 0.35374428779106726,
      "learning_rate": 5.628796796008435e-07,
      "loss": 0.0343,
      "step": 1308
    },
    {
      "epoch": 1.7269129287598943,
      "grad_norm": 0.36859477304072386,
      "learning_rate": 5.575830607770443e-07,
      "loss": 0.0354,
      "step": 1309
    },
    {
      "epoch": 1.7282321899736148,
      "grad_norm": 0.4613643600461862,
      "learning_rate": 5.52310008184182e-07,
      "loss": 0.0304,
      "step": 1310
    },
    {
      "epoch": 1.729551451187335,
      "grad_norm": 0.592134295913327,
      "learning_rate": 5.470605497948611e-07,
      "loss": 0.0336,
      "step": 1311
    },
    {
      "epoch": 1.7308707124010554,
      "grad_norm": 0.4694572788775886,
      "learning_rate": 5.418347134565249e-07,
      "loss": 0.0379,
      "step": 1312
    },
    {
      "epoch": 1.732189973614776,
      "grad_norm": 0.418013815042632,
      "learning_rate": 5.36632526891303e-07,
      "loss": 0.032,
      "step": 1313
    },
    {
      "epoch": 1.733509234828496,
      "grad_norm": 0.33921206405287646,
      "learning_rate": 5.314540176958699e-07,
      "loss": 0.0354,
      "step": 1314
    },
    {
      "epoch": 1.7348284960422165,
      "grad_norm": 0.45017885763771565,
      "learning_rate": 5.262992133412947e-07,
      "loss": 0.0443,
      "step": 1315
    },
    {
      "epoch": 1.7361477572559367,
      "grad_norm": 0.326162781727029,
      "learning_rate": 5.211681411728969e-07,
      "loss": 0.0296,
      "step": 1316
    },
    {
      "epoch": 1.737467018469657,
      "grad_norm": 0.4250567661186686,
      "learning_rate": 5.16060828410101e-07,
      "loss": 0.0445,
      "step": 1317
    },
    {
      "epoch": 1.7387862796833773,
      "grad_norm": 0.4018000498866085,
      "learning_rate": 5.109773021462921e-07,
      "loss": 0.0458,
      "step": 1318
    },
    {
      "epoch": 1.7401055408970976,
      "grad_norm": 0.7027370048578987,
      "learning_rate": 5.059175893486712e-07,
      "loss": 0.0411,
      "step": 1319
    },
    {
      "epoch": 1.741424802110818,
      "grad_norm": 0.4162139981216462,
      "learning_rate": 5.008817168581137e-07,
      "loss": 0.0402,
      "step": 1320
    },
    {
      "epoch": 1.7427440633245381,
      "grad_norm": 0.32860074949075946,
      "learning_rate": 4.958697113890271e-07,
      "loss": 0.0358,
      "step": 1321
    },
    {
      "epoch": 1.7440633245382586,
      "grad_norm": 0.42036600275964464,
      "learning_rate": 4.908815995292082e-07,
      "loss": 0.0344,
      "step": 1322
    },
    {
      "epoch": 1.745382585751979,
      "grad_norm": 0.52059293516186,
      "learning_rate": 4.859174077397022e-07,
      "loss": 0.0562,
      "step": 1323
    },
    {
      "epoch": 1.7467018469656992,
      "grad_norm": 0.5745641702799402,
      "learning_rate": 4.809771623546627e-07,
      "loss": 0.0492,
      "step": 1324
    },
    {
      "epoch": 1.7480211081794197,
      "grad_norm": 0.35422514466473387,
      "learning_rate": 4.76060889581213e-07,
      "loss": 0.0339,
      "step": 1325
    },
    {
      "epoch": 1.7493403693931397,
      "grad_norm": 0.3547143014757154,
      "learning_rate": 4.711686154993028e-07,
      "loss": 0.0407,
      "step": 1326
    },
    {
      "epoch": 1.7506596306068603,
      "grad_norm": 0.4717976706982805,
      "learning_rate": 4.6630036606157616e-07,
      "loss": 0.0282,
      "step": 1327
    },
    {
      "epoch": 1.7519788918205803,
      "grad_norm": 0.40394038783642244,
      "learning_rate": 4.614561670932288e-07,
      "loss": 0.0333,
      "step": 1328
    },
    {
      "epoch": 1.7532981530343008,
      "grad_norm": 0.5160848481103655,
      "learning_rate": 4.5663604429187547e-07,
      "loss": 0.0496,
      "step": 1329
    },
    {
      "epoch": 1.754617414248021,
      "grad_norm": 0.4232435269539701,
      "learning_rate": 4.5184002322740784e-07,
      "loss": 0.0339,
      "step": 1330
    },
    {
      "epoch": 1.7559366754617414,
      "grad_norm": 0.40699788742931914,
      "learning_rate": 4.470681293418655e-07,
      "loss": 0.0382,
      "step": 1331
    },
    {
      "epoch": 1.7572559366754619,
      "grad_norm": 0.4906969195565496,
      "learning_rate": 4.423203879492943e-07,
      "loss": 0.0389,
      "step": 1332
    },
    {
      "epoch": 1.758575197889182,
      "grad_norm": 0.4131458996348403,
      "learning_rate": 4.375968242356171e-07,
      "loss": 0.0396,
      "step": 1333
    },
    {
      "epoch": 1.7598944591029024,
      "grad_norm": 0.43933442998989664,
      "learning_rate": 4.3289746325849924e-07,
      "loss": 0.0419,
      "step": 1334
    },
    {
      "epoch": 1.7612137203166227,
      "grad_norm": 0.8251931213477443,
      "learning_rate": 4.282223299472138e-07,
      "loss": 0.0504,
      "step": 1335
    },
    {
      "epoch": 1.762532981530343,
      "grad_norm": 1.2581963511177185,
      "learning_rate": 4.2357144910251003e-07,
      "loss": 0.0426,
      "step": 1336
    },
    {
      "epoch": 1.7638522427440633,
      "grad_norm": 0.4329016484451927,
      "learning_rate": 4.189448453964845e-07,
      "loss": 0.0346,
      "step": 1337
    },
    {
      "epoch": 1.7651715039577835,
      "grad_norm": 0.3701763750178543,
      "learning_rate": 4.1434254337244404e-07,
      "loss": 0.0354,
      "step": 1338
    },
    {
      "epoch": 1.766490765171504,
      "grad_norm": 0.41529219024756975,
      "learning_rate": 4.0976456744478254e-07,
      "loss": 0.0299,
      "step": 1339
    },
    {
      "epoch": 1.767810026385224,
      "grad_norm": 0.4742675186457856,
      "learning_rate": 4.05210941898847e-07,
      "loss": 0.0513,
      "step": 1340
    },
    {
      "epoch": 1.7691292875989446,
      "grad_norm": 0.4247152497761631,
      "learning_rate": 4.006816908908101e-07,
      "loss": 0.0331,
      "step": 1341
    },
    {
      "epoch": 1.770448548812665,
      "grad_norm": 0.42305349223370864,
      "learning_rate": 3.9617683844754284e-07,
      "loss": 0.0361,
      "step": 1342
    },
    {
      "epoch": 1.7717678100263852,
      "grad_norm": 0.6014630922271789,
      "learning_rate": 3.916964084664848e-07,
      "loss": 0.0336,
      "step": 1343
    },
    {
      "epoch": 1.7730870712401057,
      "grad_norm": 0.3405494826572094,
      "learning_rate": 3.8724042471551925e-07,
      "loss": 0.0311,
      "step": 1344
    },
    {
      "epoch": 1.7744063324538257,
      "grad_norm": 0.37473108350403517,
      "learning_rate": 3.8280891083284646e-07,
      "loss": 0.0367,
      "step": 1345
    },
    {
      "epoch": 1.7757255936675462,
      "grad_norm": 0.611239757456385,
      "learning_rate": 3.784018903268588e-07,
      "loss": 0.0411,
      "step": 1346
    },
    {
      "epoch": 1.7770448548812665,
      "grad_norm": 0.5062233019350862,
      "learning_rate": 3.7401938657601555e-07,
      "loss": 0.0551,
      "step": 1347
    },
    {
      "epoch": 1.7783641160949868,
      "grad_norm": 0.3335548806824567,
      "learning_rate": 3.6966142282871873e-07,
      "loss": 0.0286,
      "step": 1348
    },
    {
      "epoch": 1.779683377308707,
      "grad_norm": 0.47839614180461826,
      "learning_rate": 3.653280222031913e-07,
      "loss": 0.0486,
      "step": 1349
    },
    {
      "epoch": 1.7810026385224274,
      "grad_norm": 0.4047331132243693,
      "learning_rate": 3.610192076873498e-07,
      "loss": 0.034,
      "step": 1350
    },
    {
      "epoch": 1.7823218997361479,
      "grad_norm": 0.6147779690764298,
      "learning_rate": 3.567350021386895e-07,
      "loss": 0.0557,
      "step": 1351
    },
    {
      "epoch": 1.783641160949868,
      "grad_norm": 0.4414521830326967,
      "learning_rate": 3.524754282841575e-07,
      "loss": 0.0301,
      "step": 1352
    },
    {
      "epoch": 1.7849604221635884,
      "grad_norm": 0.5631775902206335,
      "learning_rate": 3.482405087200352e-07,
      "loss": 0.0417,
      "step": 1353
    },
    {
      "epoch": 1.7862796833773087,
      "grad_norm": 0.35558876215797697,
      "learning_rate": 3.440302659118172e-07,
      "loss": 0.0235,
      "step": 1354
    },
    {
      "epoch": 1.787598944591029,
      "grad_norm": 0.4582685902858596,
      "learning_rate": 3.39844722194092e-07,
      "loss": 0.0416,
      "step": 1355
    },
    {
      "epoch": 1.7889182058047495,
      "grad_norm": 0.4048407621656869,
      "learning_rate": 3.356838997704226e-07,
      "loss": 0.0308,
      "step": 1356
    },
    {
      "epoch": 1.7902374670184695,
      "grad_norm": 0.4955145015945528,
      "learning_rate": 3.315478207132322e-07,
      "loss": 0.0519,
      "step": 1357
    },
    {
      "epoch": 1.79155672823219,
      "grad_norm": 0.3983405329563796,
      "learning_rate": 3.274365069636831e-07,
      "loss": 0.0348,
      "step": 1358
    },
    {
      "epoch": 1.7928759894459103,
      "grad_norm": 0.4454190763224718,
      "learning_rate": 3.233499803315637e-07,
      "loss": 0.0463,
      "step": 1359
    },
    {
      "epoch": 1.7941952506596306,
      "grad_norm": 0.46036440046593147,
      "learning_rate": 3.1928826249516984e-07,
      "loss": 0.0402,
      "step": 1360
    },
    {
      "epoch": 1.7955145118733509,
      "grad_norm": 0.4343802230808252,
      "learning_rate": 3.1525137500119207e-07,
      "loss": 0.0308,
      "step": 1361
    },
    {
      "epoch": 1.7968337730870712,
      "grad_norm": 0.39526583994521997,
      "learning_rate": 3.112393392645985e-07,
      "loss": 0.0258,
      "step": 1362
    },
    {
      "epoch": 1.7981530343007917,
      "grad_norm": 0.3522368913909313,
      "learning_rate": 3.072521765685249e-07,
      "loss": 0.0226,
      "step": 1363
    },
    {
      "epoch": 1.7994722955145117,
      "grad_norm": 0.5018803279485713,
      "learning_rate": 3.0328990806415935e-07,
      "loss": 0.0412,
      "step": 1364
    },
    {
      "epoch": 1.8007915567282322,
      "grad_norm": 0.6026623993978871,
      "learning_rate": 2.993525547706316e-07,
      "loss": 0.0416,
      "step": 1365
    },
    {
      "epoch": 1.8021108179419525,
      "grad_norm": 0.40115190329359646,
      "learning_rate": 2.9544013757489944e-07,
      "loss": 0.0308,
      "step": 1366
    },
    {
      "epoch": 1.8034300791556728,
      "grad_norm": 0.4042690712138384,
      "learning_rate": 2.915526772316402e-07,
      "loss": 0.0322,
      "step": 1367
    },
    {
      "epoch": 1.8047493403693933,
      "grad_norm": 0.5332516621967311,
      "learning_rate": 2.876901943631372e-07,
      "loss": 0.043,
      "step": 1368
    },
    {
      "epoch": 1.8060686015831133,
      "grad_norm": 0.5332780197622528,
      "learning_rate": 2.8385270945917584e-07,
      "loss": 0.039,
      "step": 1369
    },
    {
      "epoch": 1.8073878627968338,
      "grad_norm": 0.8182212019252886,
      "learning_rate": 2.8004024287692944e-07,
      "loss": 0.0386,
      "step": 1370
    },
    {
      "epoch": 1.8087071240105541,
      "grad_norm": 0.34092005245045315,
      "learning_rate": 2.76252814840855e-07,
      "loss": 0.0236,
      "step": 1371
    },
    {
      "epoch": 1.8100263852242744,
      "grad_norm": 0.5312862891963817,
      "learning_rate": 2.724904454425836e-07,
      "loss": 0.0394,
      "step": 1372
    },
    {
      "epoch": 1.8113456464379947,
      "grad_norm": 0.4783108589864548,
      "learning_rate": 2.6875315464081566e-07,
      "loss": 0.0425,
      "step": 1373
    },
    {
      "epoch": 1.812664907651715,
      "grad_norm": 0.49846941359788005,
      "learning_rate": 2.650409622612138e-07,
      "loss": 0.0358,
      "step": 1374
    },
    {
      "epoch": 1.8139841688654355,
      "grad_norm": 0.578478088885749,
      "learning_rate": 2.613538879962957e-07,
      "loss": 0.0616,
      "step": 1375
    },
    {
      "epoch": 1.8153034300791555,
      "grad_norm": 0.5716709856066361,
      "learning_rate": 2.5769195140533556e-07,
      "loss": 0.0412,
      "step": 1376
    },
    {
      "epoch": 1.816622691292876,
      "grad_norm": 0.365312638025696,
      "learning_rate": 2.540551719142548e-07,
      "loss": 0.0263,
      "step": 1377
    },
    {
      "epoch": 1.8179419525065963,
      "grad_norm": 0.4215565913003183,
      "learning_rate": 2.5044356881552045e-07,
      "loss": 0.0394,
      "step": 1378
    },
    {
      "epoch": 1.8192612137203166,
      "grad_norm": 0.4281630403371726,
      "learning_rate": 2.4685716126804485e-07,
      "loss": 0.0344,
      "step": 1379
    },
    {
      "epoch": 1.820580474934037,
      "grad_norm": 0.45713069089302194,
      "learning_rate": 2.4329596829708145e-07,
      "loss": 0.0387,
      "step": 1380
    },
    {
      "epoch": 1.8218997361477571,
      "grad_norm": 0.571386570983936,
      "learning_rate": 2.397600087941243e-07,
      "loss": 0.0486,
      "step": 1381
    },
    {
      "epoch": 1.8232189973614776,
      "grad_norm": 0.475957344717684,
      "learning_rate": 2.362493015168088e-07,
      "loss": 0.0537,
      "step": 1382
    },
    {
      "epoch": 1.824538258575198,
      "grad_norm": 0.41922931922262313,
      "learning_rate": 2.327638650888131e-07,
      "loss": 0.0289,
      "step": 1383
    },
    {
      "epoch": 1.8258575197889182,
      "grad_norm": 0.41085453215615697,
      "learning_rate": 2.2930371799975593e-07,
      "loss": 0.031,
      "step": 1384
    },
    {
      "epoch": 1.8271767810026385,
      "grad_norm": 0.5793799621353137,
      "learning_rate": 2.258688786051022e-07,
      "loss": 0.03,
      "step": 1385
    },
    {
      "epoch": 1.8284960422163588,
      "grad_norm": 0.4228166482758251,
      "learning_rate": 2.2245936512606314e-07,
      "loss": 0.0299,
      "step": 1386
    },
    {
      "epoch": 1.8298153034300793,
      "grad_norm": 0.47912620143754475,
      "learning_rate": 2.1907519564950075e-07,
      "loss": 0.0448,
      "step": 1387
    },
    {
      "epoch": 1.8311345646437993,
      "grad_norm": 0.46622053446154005,
      "learning_rate": 2.1571638812783125e-07,
      "loss": 0.0451,
      "step": 1388
    },
    {
      "epoch": 1.8324538258575198,
      "grad_norm": 0.478821469184056,
      "learning_rate": 2.123829603789307e-07,
      "loss": 0.0359,
      "step": 1389
    },
    {
      "epoch": 1.83377308707124,
      "grad_norm": 0.5485233771158393,
      "learning_rate": 2.0907493008604007e-07,
      "loss": 0.0458,
      "step": 1390
    },
    {
      "epoch": 1.8350923482849604,
      "grad_norm": 0.641372675811971,
      "learning_rate": 2.0579231479767093e-07,
      "loss": 0.0366,
      "step": 1391
    },
    {
      "epoch": 1.8364116094986809,
      "grad_norm": 0.6700473910897512,
      "learning_rate": 2.0253513192751374e-07,
      "loss": 0.0583,
      "step": 1392
    },
    {
      "epoch": 1.837730870712401,
      "grad_norm": 0.8939302910448729,
      "learning_rate": 1.993033987543419e-07,
      "loss": 0.0455,
      "step": 1393
    },
    {
      "epoch": 1.8390501319261214,
      "grad_norm": 0.3237337238677659,
      "learning_rate": 1.960971324219263e-07,
      "loss": 0.0249,
      "step": 1394
    },
    {
      "epoch": 1.8403693931398417,
      "grad_norm": 0.5188592511707015,
      "learning_rate": 1.9291634993893803e-07,
      "loss": 0.0405,
      "step": 1395
    },
    {
      "epoch": 1.841688654353562,
      "grad_norm": 0.711598934804882,
      "learning_rate": 1.8976106817886197e-07,
      "loss": 0.0393,
      "step": 1396
    },
    {
      "epoch": 1.8430079155672823,
      "grad_norm": 0.5158747451274224,
      "learning_rate": 1.8663130387990612e-07,
      "loss": 0.0495,
      "step": 1397
    },
    {
      "epoch": 1.8443271767810026,
      "grad_norm": 0.5051814837715082,
      "learning_rate": 1.8352707364491352e-07,
      "loss": 0.0408,
      "step": 1398
    },
    {
      "epoch": 1.845646437994723,
      "grad_norm": 0.364833204684728,
      "learning_rate": 1.804483939412721e-07,
      "loss": 0.0305,
      "step": 1399
    },
    {
      "epoch": 1.8469656992084431,
      "grad_norm": 0.4926948614968836,
      "learning_rate": 1.7739528110083003e-07,
      "loss": 0.0462,
      "step": 1400
    },
    {
      "epoch": 1.8482849604221636,
      "grad_norm": 0.4060254995491697,
      "learning_rate": 1.7436775131980665e-07,
      "loss": 0.0311,
      "step": 1401
    },
    {
      "epoch": 1.849604221635884,
      "grad_norm": 0.463198005789052,
      "learning_rate": 1.7136582065870876e-07,
      "loss": 0.0426,
      "step": 1402
    },
    {
      "epoch": 1.8509234828496042,
      "grad_norm": 0.36889044907936963,
      "learning_rate": 1.683895050422446e-07,
      "loss": 0.0263,
      "step": 1403
    },
    {
      "epoch": 1.8522427440633247,
      "grad_norm": 0.659379596842805,
      "learning_rate": 1.6543882025923884e-07,
      "loss": 0.0362,
      "step": 1404
    },
    {
      "epoch": 1.8535620052770447,
      "grad_norm": 0.48285367378926597,
      "learning_rate": 1.6251378196254775e-07,
      "loss": 0.0315,
      "step": 1405
    },
    {
      "epoch": 1.8548812664907652,
      "grad_norm": 0.31072920298807033,
      "learning_rate": 1.5961440566897913e-07,
      "loss": 0.0275,
      "step": 1406
    },
    {
      "epoch": 1.8562005277044855,
      "grad_norm": 0.38114528622335464,
      "learning_rate": 1.5674070675920805e-07,
      "loss": 0.0368,
      "step": 1407
    },
    {
      "epoch": 1.8575197889182058,
      "grad_norm": 0.3847157043647642,
      "learning_rate": 1.5389270047769578e-07,
      "loss": 0.0345,
      "step": 1408
    },
    {
      "epoch": 1.858839050131926,
      "grad_norm": 0.34769161488585415,
      "learning_rate": 1.5107040193260814e-07,
      "loss": 0.0275,
      "step": 1409
    },
    {
      "epoch": 1.8601583113456464,
      "grad_norm": 0.5361485360942428,
      "learning_rate": 1.482738260957378e-07,
      "loss": 0.0509,
      "step": 1410
    },
    {
      "epoch": 1.8614775725593669,
      "grad_norm": 0.3634369135567901,
      "learning_rate": 1.4550298780242e-07,
      "loss": 0.0355,
      "step": 1411
    },
    {
      "epoch": 1.862796833773087,
      "grad_norm": 0.46123718043387246,
      "learning_rate": 1.427579017514591e-07,
      "loss": 0.0441,
      "step": 1412
    },
    {
      "epoch": 1.8641160949868074,
      "grad_norm": 0.749043814853032,
      "learning_rate": 1.400385825050482e-07,
      "loss": 0.0374,
      "step": 1413
    },
    {
      "epoch": 1.8654353562005277,
      "grad_norm": 0.4128932447195577,
      "learning_rate": 1.3734504448869147e-07,
      "loss": 0.0316,
      "step": 1414
    },
    {
      "epoch": 1.866754617414248,
      "grad_norm": 0.3877671226172349,
      "learning_rate": 1.346773019911285e-07,
      "loss": 0.0302,
      "step": 1415
    },
    {
      "epoch": 1.8680738786279685,
      "grad_norm": 0.44772392186436794,
      "learning_rate": 1.3203536916425842e-07,
      "loss": 0.0364,
      "step": 1416
    },
    {
      "epoch": 1.8693931398416885,
      "grad_norm": 0.3061199809339557,
      "learning_rate": 1.2941926002306536e-07,
      "loss": 0.0252,
      "step": 1417
    },
    {
      "epoch": 1.870712401055409,
      "grad_norm": 0.42164260692116234,
      "learning_rate": 1.2682898844554093e-07,
      "loss": 0.0232,
      "step": 1418
    },
    {
      "epoch": 1.8720316622691293,
      "grad_norm": 0.34059144856622964,
      "learning_rate": 1.2426456817261513e-07,
      "loss": 0.0234,
      "step": 1419
    },
    {
      "epoch": 1.8733509234828496,
      "grad_norm": 0.3976682954800572,
      "learning_rate": 1.217260128080816e-07,
      "loss": 0.042,
      "step": 1420
    },
    {
      "epoch": 1.8746701846965699,
      "grad_norm": 0.46610358415516434,
      "learning_rate": 1.192133358185238e-07,
      "loss": 0.0435,
      "step": 1421
    },
    {
      "epoch": 1.8759894459102902,
      "grad_norm": 0.6845908917213743,
      "learning_rate": 1.1672655053324655e-07,
      "loss": 0.0398,
      "step": 1422
    },
    {
      "epoch": 1.8773087071240107,
      "grad_norm": 0.4319665923504084,
      "learning_rate": 1.1426567014420297e-07,
      "loss": 0.0218,
      "step": 1423
    },
    {
      "epoch": 1.8786279683377307,
      "grad_norm": 0.33201568533751724,
      "learning_rate": 1.1183070770592442e-07,
      "loss": 0.0287,
      "step": 1424
    },
    {
      "epoch": 1.8799472295514512,
      "grad_norm": 0.41576390147862746,
      "learning_rate": 1.094216761354544e-07,
      "loss": 0.0508,
      "step": 1425
    },
    {
      "epoch": 1.8812664907651715,
      "grad_norm": 0.4440946282610569,
      "learning_rate": 1.0703858821227541e-07,
      "loss": 0.0413,
      "step": 1426
    },
    {
      "epoch": 1.8825857519788918,
      "grad_norm": 0.4666316377271166,
      "learning_rate": 1.0468145657824558e-07,
      "loss": 0.0291,
      "step": 1427
    },
    {
      "epoch": 1.8839050131926123,
      "grad_norm": 0.5848051758884389,
      "learning_rate": 1.0235029373752758e-07,
      "loss": 0.0364,
      "step": 1428
    },
    {
      "epoch": 1.8852242744063323,
      "grad_norm": 0.38398899757542004,
      "learning_rate": 1.0004511205652656e-07,
      "loss": 0.0334,
      "step": 1429
    },
    {
      "epoch": 1.8865435356200528,
      "grad_norm": 0.36496273056510814,
      "learning_rate": 9.776592376381955e-08,
      "loss": 0.0336,
      "step": 1430
    },
    {
      "epoch": 1.8878627968337731,
      "grad_norm": 0.3788551185262741,
      "learning_rate": 9.551274095009444e-08,
      "loss": 0.0308,
      "step": 1431
    },
    {
      "epoch": 1.8891820580474934,
      "grad_norm": 0.4022971579863129,
      "learning_rate": 9.32855755680867e-08,
      "loss": 0.036,
      "step": 1432
    },
    {
      "epoch": 1.8905013192612137,
      "grad_norm": 0.393215166622088,
      "learning_rate": 9.108443943251055e-08,
      "loss": 0.041,
      "step": 1433
    },
    {
      "epoch": 1.891820580474934,
      "grad_norm": 0.5447231643725466,
      "learning_rate": 8.89093442200023e-08,
      "loss": 0.0428,
      "step": 1434
    },
    {
      "epoch": 1.8931398416886545,
      "grad_norm": 0.3790315267419849,
      "learning_rate": 8.676030146905434e-08,
      "loss": 0.0288,
      "step": 1435
    },
    {
      "epoch": 1.8944591029023745,
      "grad_norm": 0.37530226335630346,
      "learning_rate": 8.463732257995571e-08,
      "loss": 0.0277,
      "step": 1436
    },
    {
      "epoch": 1.895778364116095,
      "grad_norm": 0.37514039760418133,
      "learning_rate": 8.254041881473163e-08,
      "loss": 0.031,
      "step": 1437
    },
    {
      "epoch": 1.8970976253298153,
      "grad_norm": 0.35567700062264085,
      "learning_rate": 8.046960129708348e-08,
      "loss": 0.0258,
      "step": 1438
    },
    {
      "epoch": 1.8984168865435356,
      "grad_norm": 0.31803214794946344,
      "learning_rate": 7.842488101232893e-08,
      "loss": 0.0221,
      "step": 1439
    },
    {
      "epoch": 1.899736147757256,
      "grad_norm": 0.42855694638621145,
      "learning_rate": 7.640626880734581e-08,
      "loss": 0.0287,
      "step": 1440
    },
    {
      "epoch": 1.9010554089709761,
      "grad_norm": 0.44050404423122047,
      "learning_rate": 7.441377539051165e-08,
      "loss": 0.0231,
      "step": 1441
    },
    {
      "epoch": 1.9023746701846966,
      "grad_norm": 0.4006135382220008,
      "learning_rate": 7.244741133164979e-08,
      "loss": 0.0416,
      "step": 1442
    },
    {
      "epoch": 1.903693931398417,
      "grad_norm": 0.5035650996719838,
      "learning_rate": 7.050718706197168e-08,
      "loss": 0.0441,
      "step": 1443
    },
    {
      "epoch": 1.9050131926121372,
      "grad_norm": 0.38877200706439374,
      "learning_rate": 6.859311287402081e-08,
      "loss": 0.0383,
      "step": 1444
    },
    {
      "epoch": 1.9063324538258575,
      "grad_norm": 0.4710132583661197,
      "learning_rate": 6.670519892162053e-08,
      "loss": 0.0333,
      "step": 1445
    },
    {
      "epoch": 1.9076517150395778,
      "grad_norm": 0.39111866383182126,
      "learning_rate": 6.48434552198185e-08,
      "loss": 0.0344,
      "step": 1446
    },
    {
      "epoch": 1.9089709762532983,
      "grad_norm": 0.3369746744134288,
      "learning_rate": 6.30078916448329e-08,
      "loss": 0.0262,
      "step": 1447
    },
    {
      "epoch": 1.9102902374670183,
      "grad_norm": 0.3672285966035807,
      "learning_rate": 6.119851793400188e-08,
      "loss": 0.0357,
      "step": 1448
    },
    {
      "epoch": 1.9116094986807388,
      "grad_norm": 0.501282704651796,
      "learning_rate": 5.941534368573143e-08,
      "loss": 0.0513,
      "step": 1449
    },
    {
      "epoch": 1.912928759894459,
      "grad_norm": 0.515286690814764,
      "learning_rate": 5.7658378359443104e-08,
      "loss": 0.0261,
      "step": 1450
    },
    {
      "epoch": 1.9142480211081794,
      "grad_norm": 0.4150045067689141,
      "learning_rate": 5.59276312755247e-08,
      "loss": 0.0395,
      "step": 1451
    },
    {
      "epoch": 1.9155672823218999,
      "grad_norm": 0.3637171809597343,
      "learning_rate": 5.4223111615281935e-08,
      "loss": 0.0349,
      "step": 1452
    },
    {
      "epoch": 1.91688654353562,
      "grad_norm": 0.405991825012232,
      "learning_rate": 5.254482842088793e-08,
      "loss": 0.0302,
      "step": 1453
    },
    {
      "epoch": 1.9182058047493404,
      "grad_norm": 0.6027202091909221,
      "learning_rate": 5.089279059533658e-08,
      "loss": 0.0537,
      "step": 1454
    },
    {
      "epoch": 1.9195250659630607,
      "grad_norm": 0.3826886359954292,
      "learning_rate": 4.926700690239372e-08,
      "loss": 0.0325,
      "step": 1455
    },
    {
      "epoch": 1.920844327176781,
      "grad_norm": 0.5813404321771318,
      "learning_rate": 4.766748596655268e-08,
      "loss": 0.0368,
      "step": 1456
    },
    {
      "epoch": 1.9221635883905013,
      "grad_norm": 0.455817012611782,
      "learning_rate": 4.609423627298715e-08,
      "loss": 0.0237,
      "step": 1457
    },
    {
      "epoch": 1.9234828496042216,
      "grad_norm": 0.5487232651165294,
      "learning_rate": 4.4547266167507264e-08,
      "loss": 0.034,
      "step": 1458
    },
    {
      "epoch": 1.924802110817942,
      "grad_norm": 0.3923882257847083,
      "learning_rate": 4.302658385651359e-08,
      "loss": 0.0305,
      "step": 1459
    },
    {
      "epoch": 1.9261213720316621,
      "grad_norm": 0.361930421531541,
      "learning_rate": 4.1532197406954357e-08,
      "loss": 0.027,
      "step": 1460
    },
    {
      "epoch": 1.9274406332453826,
      "grad_norm": 0.3410974255034189,
      "learning_rate": 4.006411474628491e-08,
      "loss": 0.0257,
      "step": 1461
    },
    {
      "epoch": 1.928759894459103,
      "grad_norm": 0.3391437722549354,
      "learning_rate": 3.862234366242168e-08,
      "loss": 0.0257,
      "step": 1462
    },
    {
      "epoch": 1.9300791556728232,
      "grad_norm": 0.7996017631556153,
      "learning_rate": 3.720689180370329e-08,
      "loss": 0.0353,
      "step": 1463
    },
    {
      "epoch": 1.9313984168865437,
      "grad_norm": 0.3903649276794555,
      "learning_rate": 3.581776667885062e-08,
      "loss": 0.0317,
      "step": 1464
    },
    {
      "epoch": 1.9327176781002637,
      "grad_norm": 0.46536412104370617,
      "learning_rate": 3.445497565692457e-08,
      "loss": 0.0351,
      "step": 1465
    },
    {
      "epoch": 1.9340369393139842,
      "grad_norm": 0.4529803508470128,
      "learning_rate": 3.311852596728948e-08,
      "loss": 0.0339,
      "step": 1466
    },
    {
      "epoch": 1.9353562005277045,
      "grad_norm": 0.680215979452562,
      "learning_rate": 3.1808424699572014e-08,
      "loss": 0.0311,
      "step": 1467
    },
    {
      "epoch": 1.9366754617414248,
      "grad_norm": 0.409978370192348,
      "learning_rate": 3.052467880362675e-08,
      "loss": 0.0435,
      "step": 1468
    },
    {
      "epoch": 1.937994722955145,
      "grad_norm": 0.3182373919865013,
      "learning_rate": 2.9267295089497327e-08,
      "loss": 0.0344,
      "step": 1469
    },
    {
      "epoch": 1.9393139841688654,
      "grad_norm": 1.0684008641939509,
      "learning_rate": 2.8036280227379808e-08,
      "loss": 0.0382,
      "step": 1470
    },
    {
      "epoch": 1.9406332453825859,
      "grad_norm": 0.8240642381272889,
      "learning_rate": 2.6831640747589925e-08,
      "loss": 0.0253,
      "step": 1471
    },
    {
      "epoch": 1.941952506596306,
      "grad_norm": 0.39242837732937735,
      "learning_rate": 2.5653383040524228e-08,
      "loss": 0.0368,
      "step": 1472
    },
    {
      "epoch": 1.9432717678100264,
      "grad_norm": 0.433259089337428,
      "learning_rate": 2.4501513356631202e-08,
      "loss": 0.0296,
      "step": 1473
    },
    {
      "epoch": 1.9445910290237467,
      "grad_norm": 0.728610866735138,
      "learning_rate": 2.3376037806374097e-08,
      "loss": 0.0412,
      "step": 1474
    },
    {
      "epoch": 1.945910290237467,
      "grad_norm": 0.3805477215998381,
      "learning_rate": 2.2276962360200383e-08,
      "loss": 0.0297,
      "step": 1475
    },
    {
      "epoch": 1.9472295514511875,
      "grad_norm": 0.3452238003115353,
      "learning_rate": 2.1204292848509557e-08,
      "loss": 0.0303,
      "step": 1476
    },
    {
      "epoch": 1.9485488126649075,
      "grad_norm": 0.4659613637601965,
      "learning_rate": 2.0158034961622607e-08,
      "loss": 0.041,
      "step": 1477
    },
    {
      "epoch": 1.949868073878628,
      "grad_norm": 0.33093130150957134,
      "learning_rate": 1.9138194249750386e-08,
      "loss": 0.0346,
      "step": 1478
    },
    {
      "epoch": 1.9511873350923483,
      "grad_norm": 1.0830612352894715,
      "learning_rate": 1.81447761229675e-08,
      "loss": 0.0551,
      "step": 1479
    },
    {
      "epoch": 1.9525065963060686,
      "grad_norm": 0.4573678054826342,
      "learning_rate": 1.7177785851180127e-08,
      "loss": 0.0468,
      "step": 1480
    },
    {
      "epoch": 1.9538258575197889,
      "grad_norm": 0.36034300277622483,
      "learning_rate": 1.6237228564098818e-08,
      "loss": 0.0276,
      "step": 1481
    },
    {
      "epoch": 1.9551451187335092,
      "grad_norm": 0.4269174174424741,
      "learning_rate": 1.532310925121294e-08,
      "loss": 0.0334,
      "step": 1482
    },
    {
      "epoch": 1.9564643799472297,
      "grad_norm": 0.5534031509176094,
      "learning_rate": 1.4435432761762958e-08,
      "loss": 0.0454,
      "step": 1483
    },
    {
      "epoch": 1.9577836411609497,
      "grad_norm": 0.48672300692494025,
      "learning_rate": 1.3574203804713748e-08,
      "loss": 0.047,
      "step": 1484
    },
    {
      "epoch": 1.9591029023746702,
      "grad_norm": 0.3089514019798877,
      "learning_rate": 1.2739426948732426e-08,
      "loss": 0.0222,
      "step": 1485
    },
    {
      "epoch": 1.9604221635883905,
      "grad_norm": 0.45341517194027414,
      "learning_rate": 1.1931106622161127e-08,
      "loss": 0.0297,
      "step": 1486
    },
    {
      "epoch": 1.9617414248021108,
      "grad_norm": 0.5583483850941126,
      "learning_rate": 1.1149247112995365e-08,
      "loss": 0.0463,
      "step": 1487
    },
    {
      "epoch": 1.9630606860158313,
      "grad_norm": 0.3564316234881558,
      "learning_rate": 1.0393852568860718e-08,
      "loss": 0.0203,
      "step": 1488
    },
    {
      "epoch": 1.9643799472295513,
      "grad_norm": 0.44367974596971,
      "learning_rate": 9.664926996991176e-09,
      "loss": 0.0275,
      "step": 1489
    },
    {
      "epoch": 1.9656992084432718,
      "grad_norm": 0.39152274879775784,
      "learning_rate": 8.962474264206378e-09,
      "loss": 0.0315,
      "step": 1490
    },
    {
      "epoch": 1.9670184696569921,
      "grad_norm": 0.3557589209853032,
      "learning_rate": 8.286498096893304e-09,
      "loss": 0.0252,
      "step": 1491
    },
    {
      "epoch": 1.9683377308707124,
      "grad_norm": 0.41200361634618793,
      "learning_rate": 7.637002080985167e-09,
      "loss": 0.0302,
      "step": 1492
    },
    {
      "epoch": 1.9696569920844327,
      "grad_norm": 0.7439654263189737,
      "learning_rate": 7.013989661942555e-09,
      "loss": 0.042,
      "step": 1493
    },
    {
      "epoch": 1.970976253298153,
      "grad_norm": 0.410572155045626,
      "learning_rate": 6.417464144736208e-09,
      "loss": 0.0433,
      "step": 1494
    },
    {
      "epoch": 1.9722955145118735,
      "grad_norm": 0.43767939437112174,
      "learning_rate": 5.847428693826484e-09,
      "loss": 0.0322,
      "step": 1495
    },
    {
      "epoch": 1.9736147757255935,
      "grad_norm": 0.3220957811650404,
      "learning_rate": 5.303886333151154e-09,
      "loss": 0.0245,
      "step": 1496
    },
    {
      "epoch": 1.974934036939314,
      "grad_norm": 0.39372689260064875,
      "learning_rate": 4.786839946104849e-09,
      "loss": 0.0255,
      "step": 1497
    },
    {
      "epoch": 1.9762532981530343,
      "grad_norm": 0.45658707250148867,
      "learning_rate": 4.296292275526859e-09,
      "loss": 0.0387,
      "step": 1498
    },
    {
      "epoch": 1.9775725593667546,
      "grad_norm": 0.4758183336377512,
      "learning_rate": 3.8322459236850296e-09,
      "loss": 0.037,
      "step": 1499
    },
    {
      "epoch": 1.978891820580475,
      "grad_norm": 0.334075873932478,
      "learning_rate": 3.394703352263551e-09,
      "loss": 0.0277,
      "step": 1500
    },
    {
      "epoch": 1.9802110817941951,
      "grad_norm": 0.7256974162727998,
      "learning_rate": 2.9836668823468583e-09,
      "loss": 0.0993,
      "step": 1501
    },
    {
      "epoch": 1.9815303430079156,
      "grad_norm": 0.5370951547540892,
      "learning_rate": 2.5991386944107524e-09,
      "loss": 0.0401,
      "step": 1502
    },
    {
      "epoch": 1.982849604221636,
      "grad_norm": 0.5128469752391834,
      "learning_rate": 2.241120828308518e-09,
      "loss": 0.0429,
      "step": 1503
    },
    {
      "epoch": 1.9841688654353562,
      "grad_norm": 0.5835488080744212,
      "learning_rate": 1.9096151832609378e-09,
      "loss": 0.0305,
      "step": 1504
    },
    {
      "epoch": 1.9854881266490765,
      "grad_norm": 0.7672436878231969,
      "learning_rate": 1.6046235178474034e-09,
      "loss": 0.0621,
      "step": 1505
    },
    {
      "epoch": 1.9868073878627968,
      "grad_norm": 0.42712213180129055,
      "learning_rate": 1.326147449993709e-09,
      "loss": 0.0277,
      "step": 1506
    },
    {
      "epoch": 1.9881266490765173,
      "grad_norm": 0.443092608668473,
      "learning_rate": 1.0741884569659412e-09,
      "loss": 0.0455,
      "step": 1507
    },
    {
      "epoch": 1.9894459102902373,
      "grad_norm": 0.5669527089886233,
      "learning_rate": 8.487478753615997e-10,
      "loss": 0.0511,
      "step": 1508
    },
    {
      "epoch": 1.9907651715039578,
      "grad_norm": 0.49600894039730775,
      "learning_rate": 6.498269011029346e-10,
      "loss": 0.0536,
      "step": 1509
    },
    {
      "epoch": 1.992084432717678,
      "grad_norm": 0.4435904800966189,
      "learning_rate": 4.774265894302854e-10,
      "loss": 0.0298,
      "step": 1510
    },
    {
      "epoch": 1.9934036939313984,
      "grad_norm": 0.3237649659830547,
      "learning_rate": 3.3154785489653006e-10,
      "loss": 0.0244,
      "step": 1511
    },
    {
      "epoch": 1.9947229551451189,
      "grad_norm": 0.7407042869403138,
      "learning_rate": 2.1219147136264383e-10,
      "loss": 0.0434,
      "step": 1512
    },
    {
      "epoch": 1.996042216358839,
      "grad_norm": 0.4675300736701374,
      "learning_rate": 1.1935807199270343e-10,
      "loss": 0.0324,
      "step": 1513
    },
    {
      "epoch": 1.9973614775725594,
      "grad_norm": 0.29583225995771156,
      "learning_rate": 5.3048149251111456e-11,
      "loss": 0.0207,
      "step": 1514
    },
    {
      "epoch": 1.9986807387862797,
      "grad_norm": 0.3329196213750932,
      "learning_rate": 1.3262054900931021e-11,
      "loss": 0.0302,
      "step": 1515
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.3715122507168322,
      "learning_rate": 0.0,
      "loss": 0.0308,
      "step": 1516
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.04380889981985092,
      "eval_runtime": 125.8608,
      "eval_samples_per_second": 40.561,
      "eval_steps_per_second": 1.271,
      "step": 1516
    },
    {
      "epoch": 2.0,
      "step": 1516,
      "total_flos": 4.548718205631201e+17,
      "train_loss": 0.0573945701893881,
      "train_runtime": 16154.4675,
      "train_samples_per_second": 12.007,
      "train_steps_per_second": 0.094
    }
  ],
  "logging_steps": 1,
  "max_steps": 1516,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "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": 4.548718205631201e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}