{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 2112,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0014204545454545455,
      "grad_norm": 57.174587192423886,
      "learning_rate": 2.3584905660377358e-07,
      "loss": 11.1055,
      "step": 1
    },
    {
      "epoch": 0.002840909090909091,
      "grad_norm": 54.95948078088431,
      "learning_rate": 4.7169811320754717e-07,
      "loss": 11.0966,
      "step": 2
    },
    {
      "epoch": 0.004261363636363636,
      "grad_norm": 54.06610347121592,
      "learning_rate": 7.075471698113208e-07,
      "loss": 11.1551,
      "step": 3
    },
    {
      "epoch": 0.005681818181818182,
      "grad_norm": 54.46356141732358,
      "learning_rate": 9.433962264150943e-07,
      "loss": 11.1966,
      "step": 4
    },
    {
      "epoch": 0.007102272727272727,
      "grad_norm": 55.879848023051714,
      "learning_rate": 1.179245283018868e-06,
      "loss": 11.0245,
      "step": 5
    },
    {
      "epoch": 0.008522727272727272,
      "grad_norm": 57.64823630461894,
      "learning_rate": 1.4150943396226415e-06,
      "loss": 11.0271,
      "step": 6
    },
    {
      "epoch": 0.009943181818181818,
      "grad_norm": 56.37355169534874,
      "learning_rate": 1.650943396226415e-06,
      "loss": 10.8065,
      "step": 7
    },
    {
      "epoch": 0.011363636363636364,
      "grad_norm": 59.797214793501276,
      "learning_rate": 1.8867924528301887e-06,
      "loss": 10.8,
      "step": 8
    },
    {
      "epoch": 0.01278409090909091,
      "grad_norm": 61.7136856689601,
      "learning_rate": 2.1226415094339624e-06,
      "loss": 10.6342,
      "step": 9
    },
    {
      "epoch": 0.014204545454545454,
      "grad_norm": 86.1011365741843,
      "learning_rate": 2.358490566037736e-06,
      "loss": 9.3562,
      "step": 10
    },
    {
      "epoch": 0.015625,
      "grad_norm": 84.44052281975253,
      "learning_rate": 2.5943396226415095e-06,
      "loss": 9.4571,
      "step": 11
    },
    {
      "epoch": 0.017045454545454544,
      "grad_norm": 101.54300447854155,
      "learning_rate": 2.830188679245283e-06,
      "loss": 8.7379,
      "step": 12
    },
    {
      "epoch": 0.018465909090909092,
      "grad_norm": 84.82895473757196,
      "learning_rate": 3.0660377358490567e-06,
      "loss": 4.3785,
      "step": 13
    },
    {
      "epoch": 0.019886363636363636,
      "grad_norm": 59.11186600192468,
      "learning_rate": 3.30188679245283e-06,
      "loss": 3.3681,
      "step": 14
    },
    {
      "epoch": 0.02130681818181818,
      "grad_norm": 55.15275893889262,
      "learning_rate": 3.5377358490566038e-06,
      "loss": 3.1861,
      "step": 15
    },
    {
      "epoch": 0.022727272727272728,
      "grad_norm": 36.8525096118668,
      "learning_rate": 3.7735849056603773e-06,
      "loss": 2.5442,
      "step": 16
    },
    {
      "epoch": 0.024147727272727272,
      "grad_norm": 30.43715811222529,
      "learning_rate": 4.009433962264151e-06,
      "loss": 2.3048,
      "step": 17
    },
    {
      "epoch": 0.02556818181818182,
      "grad_norm": 6.840826072316582,
      "learning_rate": 4.245283018867925e-06,
      "loss": 1.3869,
      "step": 18
    },
    {
      "epoch": 0.026988636363636364,
      "grad_norm": 5.281364302528376,
      "learning_rate": 4.481132075471698e-06,
      "loss": 1.3419,
      "step": 19
    },
    {
      "epoch": 0.028409090909090908,
      "grad_norm": 4.2914326060636565,
      "learning_rate": 4.716981132075472e-06,
      "loss": 1.2589,
      "step": 20
    },
    {
      "epoch": 0.029829545454545456,
      "grad_norm": 3.1094800327161733,
      "learning_rate": 4.952830188679246e-06,
      "loss": 1.2378,
      "step": 21
    },
    {
      "epoch": 0.03125,
      "grad_norm": 2.4404069766948857,
      "learning_rate": 5.188679245283019e-06,
      "loss": 1.1218,
      "step": 22
    },
    {
      "epoch": 0.032670454545454544,
      "grad_norm": 2.0474678157939836,
      "learning_rate": 5.424528301886793e-06,
      "loss": 1.0831,
      "step": 23
    },
    {
      "epoch": 0.03409090909090909,
      "grad_norm": 1.5139698113730216,
      "learning_rate": 5.660377358490566e-06,
      "loss": 0.9814,
      "step": 24
    },
    {
      "epoch": 0.03551136363636364,
      "grad_norm": 15.883001930310485,
      "learning_rate": 5.89622641509434e-06,
      "loss": 0.9541,
      "step": 25
    },
    {
      "epoch": 0.036931818181818184,
      "grad_norm": 31.98696125739845,
      "learning_rate": 6.132075471698113e-06,
      "loss": 0.9491,
      "step": 26
    },
    {
      "epoch": 0.03835227272727273,
      "grad_norm": 2.009614720097698,
      "learning_rate": 6.367924528301887e-06,
      "loss": 0.905,
      "step": 27
    },
    {
      "epoch": 0.03977272727272727,
      "grad_norm": 1.2976847184101623,
      "learning_rate": 6.60377358490566e-06,
      "loss": 0.9435,
      "step": 28
    },
    {
      "epoch": 0.041193181818181816,
      "grad_norm": 1.0504443062741082,
      "learning_rate": 6.839622641509434e-06,
      "loss": 0.89,
      "step": 29
    },
    {
      "epoch": 0.04261363636363636,
      "grad_norm": 0.9355121657062315,
      "learning_rate": 7.0754716981132075e-06,
      "loss": 0.8324,
      "step": 30
    },
    {
      "epoch": 0.04403409090909091,
      "grad_norm": 0.8924274438989973,
      "learning_rate": 7.3113207547169815e-06,
      "loss": 0.8179,
      "step": 31
    },
    {
      "epoch": 0.045454545454545456,
      "grad_norm": 0.8036797799868524,
      "learning_rate": 7.547169811320755e-06,
      "loss": 0.7912,
      "step": 32
    },
    {
      "epoch": 0.046875,
      "grad_norm": 0.7633403910370843,
      "learning_rate": 7.783018867924528e-06,
      "loss": 0.7614,
      "step": 33
    },
    {
      "epoch": 0.048295454545454544,
      "grad_norm": 0.9486995935518266,
      "learning_rate": 8.018867924528302e-06,
      "loss": 0.7684,
      "step": 34
    },
    {
      "epoch": 0.04971590909090909,
      "grad_norm": 0.6358639740749938,
      "learning_rate": 8.254716981132076e-06,
      "loss": 0.7747,
      "step": 35
    },
    {
      "epoch": 0.05113636363636364,
      "grad_norm": 0.6072786942709014,
      "learning_rate": 8.49056603773585e-06,
      "loss": 0.7216,
      "step": 36
    },
    {
      "epoch": 0.052556818181818184,
      "grad_norm": 0.7664662958711278,
      "learning_rate": 8.726415094339622e-06,
      "loss": 0.7527,
      "step": 37
    },
    {
      "epoch": 0.05397727272727273,
      "grad_norm": 0.6720674914399464,
      "learning_rate": 8.962264150943396e-06,
      "loss": 0.7219,
      "step": 38
    },
    {
      "epoch": 0.05539772727272727,
      "grad_norm": 0.5978596701357648,
      "learning_rate": 9.19811320754717e-06,
      "loss": 0.7343,
      "step": 39
    },
    {
      "epoch": 0.056818181818181816,
      "grad_norm": 0.547895535760375,
      "learning_rate": 9.433962264150944e-06,
      "loss": 0.6985,
      "step": 40
    },
    {
      "epoch": 0.05823863636363636,
      "grad_norm": 0.5773797536290137,
      "learning_rate": 9.669811320754718e-06,
      "loss": 0.6762,
      "step": 41
    },
    {
      "epoch": 0.05965909090909091,
      "grad_norm": 0.5274226790438865,
      "learning_rate": 9.905660377358492e-06,
      "loss": 0.7009,
      "step": 42
    },
    {
      "epoch": 0.061079545454545456,
      "grad_norm": 0.5301486471844916,
      "learning_rate": 1.0141509433962266e-05,
      "loss": 0.6745,
      "step": 43
    },
    {
      "epoch": 0.0625,
      "grad_norm": 0.5018820473061312,
      "learning_rate": 1.0377358490566038e-05,
      "loss": 0.6725,
      "step": 44
    },
    {
      "epoch": 0.06392045454545454,
      "grad_norm": 1.194157723480261,
      "learning_rate": 1.0613207547169812e-05,
      "loss": 0.6463,
      "step": 45
    },
    {
      "epoch": 0.06534090909090909,
      "grad_norm": 0.45706127559808823,
      "learning_rate": 1.0849056603773586e-05,
      "loss": 0.6679,
      "step": 46
    },
    {
      "epoch": 0.06676136363636363,
      "grad_norm": 0.47246265748959365,
      "learning_rate": 1.108490566037736e-05,
      "loss": 0.6273,
      "step": 47
    },
    {
      "epoch": 0.06818181818181818,
      "grad_norm": 0.48283524528585403,
      "learning_rate": 1.1320754716981132e-05,
      "loss": 0.6254,
      "step": 48
    },
    {
      "epoch": 0.06960227272727272,
      "grad_norm": 0.4400119355398842,
      "learning_rate": 1.1556603773584906e-05,
      "loss": 0.633,
      "step": 49
    },
    {
      "epoch": 0.07102272727272728,
      "grad_norm": 0.39937757189530515,
      "learning_rate": 1.179245283018868e-05,
      "loss": 0.6492,
      "step": 50
    },
    {
      "epoch": 0.07244318181818182,
      "grad_norm": 0.6146486361375509,
      "learning_rate": 1.2028301886792454e-05,
      "loss": 0.6298,
      "step": 51
    },
    {
      "epoch": 0.07386363636363637,
      "grad_norm": 0.3625031799691764,
      "learning_rate": 1.2264150943396227e-05,
      "loss": 0.6405,
      "step": 52
    },
    {
      "epoch": 0.07528409090909091,
      "grad_norm": 0.451002632148673,
      "learning_rate": 1.25e-05,
      "loss": 0.6307,
      "step": 53
    },
    {
      "epoch": 0.07670454545454546,
      "grad_norm": 0.39300659830895485,
      "learning_rate": 1.2735849056603775e-05,
      "loss": 0.6352,
      "step": 54
    },
    {
      "epoch": 0.078125,
      "grad_norm": 0.31301746051685164,
      "learning_rate": 1.2971698113207547e-05,
      "loss": 0.6051,
      "step": 55
    },
    {
      "epoch": 0.07954545454545454,
      "grad_norm": 0.37461072254553773,
      "learning_rate": 1.320754716981132e-05,
      "loss": 0.6055,
      "step": 56
    },
    {
      "epoch": 0.08096590909090909,
      "grad_norm": 0.3233013964405309,
      "learning_rate": 1.3443396226415095e-05,
      "loss": 0.5949,
      "step": 57
    },
    {
      "epoch": 0.08238636363636363,
      "grad_norm": 0.3335963406349363,
      "learning_rate": 1.3679245283018869e-05,
      "loss": 0.5835,
      "step": 58
    },
    {
      "epoch": 0.08380681818181818,
      "grad_norm": 0.37671141575892,
      "learning_rate": 1.3915094339622641e-05,
      "loss": 0.5989,
      "step": 59
    },
    {
      "epoch": 0.08522727272727272,
      "grad_norm": 0.3122935621131345,
      "learning_rate": 1.4150943396226415e-05,
      "loss": 0.6058,
      "step": 60
    },
    {
      "epoch": 0.08664772727272728,
      "grad_norm": 0.3367020576133394,
      "learning_rate": 1.4386792452830189e-05,
      "loss": 0.5987,
      "step": 61
    },
    {
      "epoch": 0.08806818181818182,
      "grad_norm": 0.29874488354071604,
      "learning_rate": 1.4622641509433963e-05,
      "loss": 0.5713,
      "step": 62
    },
    {
      "epoch": 0.08948863636363637,
      "grad_norm": 0.3268781717342126,
      "learning_rate": 1.4858490566037735e-05,
      "loss": 0.6029,
      "step": 63
    },
    {
      "epoch": 0.09090909090909091,
      "grad_norm": 0.28962933135779173,
      "learning_rate": 1.509433962264151e-05,
      "loss": 0.5812,
      "step": 64
    },
    {
      "epoch": 0.09232954545454546,
      "grad_norm": 0.2686485077913842,
      "learning_rate": 1.5330188679245283e-05,
      "loss": 0.5837,
      "step": 65
    },
    {
      "epoch": 0.09375,
      "grad_norm": 0.2989999801958056,
      "learning_rate": 1.5566037735849056e-05,
      "loss": 0.583,
      "step": 66
    },
    {
      "epoch": 0.09517045454545454,
      "grad_norm": 0.30000938142262656,
      "learning_rate": 1.580188679245283e-05,
      "loss": 0.6088,
      "step": 67
    },
    {
      "epoch": 0.09659090909090909,
      "grad_norm": 0.4032041359312091,
      "learning_rate": 1.6037735849056604e-05,
      "loss": 0.5995,
      "step": 68
    },
    {
      "epoch": 0.09801136363636363,
      "grad_norm": 0.28930777432200405,
      "learning_rate": 1.6273584905660376e-05,
      "loss": 0.6061,
      "step": 69
    },
    {
      "epoch": 0.09943181818181818,
      "grad_norm": 0.25443324004477114,
      "learning_rate": 1.650943396226415e-05,
      "loss": 0.5781,
      "step": 70
    },
    {
      "epoch": 0.10085227272727272,
      "grad_norm": 0.2945044892132117,
      "learning_rate": 1.6745283018867924e-05,
      "loss": 0.5798,
      "step": 71
    },
    {
      "epoch": 0.10227272727272728,
      "grad_norm": 0.2818376719064825,
      "learning_rate": 1.69811320754717e-05,
      "loss": 0.5985,
      "step": 72
    },
    {
      "epoch": 0.10369318181818182,
      "grad_norm": 0.2737424104101698,
      "learning_rate": 1.7216981132075472e-05,
      "loss": 0.58,
      "step": 73
    },
    {
      "epoch": 0.10511363636363637,
      "grad_norm": 0.3023733810362363,
      "learning_rate": 1.7452830188679244e-05,
      "loss": 0.5726,
      "step": 74
    },
    {
      "epoch": 0.10653409090909091,
      "grad_norm": 0.2614846268144454,
      "learning_rate": 1.768867924528302e-05,
      "loss": 0.541,
      "step": 75
    },
    {
      "epoch": 0.10795454545454546,
      "grad_norm": 0.26213186817426776,
      "learning_rate": 1.7924528301886792e-05,
      "loss": 0.5384,
      "step": 76
    },
    {
      "epoch": 0.109375,
      "grad_norm": 0.27268238202011263,
      "learning_rate": 1.8160377358490564e-05,
      "loss": 0.5823,
      "step": 77
    },
    {
      "epoch": 0.11079545454545454,
      "grad_norm": 0.271456797925649,
      "learning_rate": 1.839622641509434e-05,
      "loss": 0.5524,
      "step": 78
    },
    {
      "epoch": 0.11221590909090909,
      "grad_norm": 0.27894358573972255,
      "learning_rate": 1.8632075471698112e-05,
      "loss": 0.5587,
      "step": 79
    },
    {
      "epoch": 0.11363636363636363,
      "grad_norm": 0.2526155414563606,
      "learning_rate": 1.8867924528301888e-05,
      "loss": 0.5516,
      "step": 80
    },
    {
      "epoch": 0.11505681818181818,
      "grad_norm": 0.2640035389598287,
      "learning_rate": 1.9103773584905664e-05,
      "loss": 0.5703,
      "step": 81
    },
    {
      "epoch": 0.11647727272727272,
      "grad_norm": 0.2579787869642489,
      "learning_rate": 1.9339622641509436e-05,
      "loss": 0.5678,
      "step": 82
    },
    {
      "epoch": 0.11789772727272728,
      "grad_norm": 0.2812429289225403,
      "learning_rate": 1.9575471698113208e-05,
      "loss": 0.5447,
      "step": 83
    },
    {
      "epoch": 0.11931818181818182,
      "grad_norm": 0.2923032550439248,
      "learning_rate": 1.9811320754716984e-05,
      "loss": 0.5675,
      "step": 84
    },
    {
      "epoch": 0.12073863636363637,
      "grad_norm": 0.2615855931756729,
      "learning_rate": 2.0047169811320756e-05,
      "loss": 0.5206,
      "step": 85
    },
    {
      "epoch": 0.12215909090909091,
      "grad_norm": 0.2993650529520495,
      "learning_rate": 2.0283018867924532e-05,
      "loss": 0.5538,
      "step": 86
    },
    {
      "epoch": 0.12357954545454546,
      "grad_norm": 0.25083237803402875,
      "learning_rate": 2.0518867924528304e-05,
      "loss": 0.5357,
      "step": 87
    },
    {
      "epoch": 0.125,
      "grad_norm": 2.5161326447730805,
      "learning_rate": 2.0754716981132076e-05,
      "loss": 0.5645,
      "step": 88
    },
    {
      "epoch": 0.12642045454545456,
      "grad_norm": 0.32211310803705934,
      "learning_rate": 2.0990566037735852e-05,
      "loss": 0.5502,
      "step": 89
    },
    {
      "epoch": 0.1278409090909091,
      "grad_norm": 0.29381763735211924,
      "learning_rate": 2.1226415094339624e-05,
      "loss": 0.5533,
      "step": 90
    },
    {
      "epoch": 0.12926136363636365,
      "grad_norm": 0.3191934336080857,
      "learning_rate": 2.1462264150943397e-05,
      "loss": 0.5395,
      "step": 91
    },
    {
      "epoch": 0.13068181818181818,
      "grad_norm": 0.3161128755847218,
      "learning_rate": 2.1698113207547172e-05,
      "loss": 0.5454,
      "step": 92
    },
    {
      "epoch": 0.13210227272727273,
      "grad_norm": 0.309480241116538,
      "learning_rate": 2.1933962264150945e-05,
      "loss": 0.5419,
      "step": 93
    },
    {
      "epoch": 0.13352272727272727,
      "grad_norm": 0.3209464321676852,
      "learning_rate": 2.216981132075472e-05,
      "loss": 0.5439,
      "step": 94
    },
    {
      "epoch": 0.13494318181818182,
      "grad_norm": 0.3012199535255356,
      "learning_rate": 2.2405660377358493e-05,
      "loss": 0.5588,
      "step": 95
    },
    {
      "epoch": 0.13636363636363635,
      "grad_norm": 0.33328550926209605,
      "learning_rate": 2.2641509433962265e-05,
      "loss": 0.5627,
      "step": 96
    },
    {
      "epoch": 0.1377840909090909,
      "grad_norm": 0.318525882773988,
      "learning_rate": 2.287735849056604e-05,
      "loss": 0.5536,
      "step": 97
    },
    {
      "epoch": 0.13920454545454544,
      "grad_norm": 0.2869869019544255,
      "learning_rate": 2.3113207547169813e-05,
      "loss": 0.542,
      "step": 98
    },
    {
      "epoch": 0.140625,
      "grad_norm": 0.3173423525879084,
      "learning_rate": 2.3349056603773585e-05,
      "loss": 0.5251,
      "step": 99
    },
    {
      "epoch": 0.14204545454545456,
      "grad_norm": 0.3078057087584305,
      "learning_rate": 2.358490566037736e-05,
      "loss": 0.5191,
      "step": 100
    },
    {
      "epoch": 0.1434659090909091,
      "grad_norm": 0.3359357438023514,
      "learning_rate": 2.3820754716981133e-05,
      "loss": 0.5221,
      "step": 101
    },
    {
      "epoch": 0.14488636363636365,
      "grad_norm": 0.31652546000151144,
      "learning_rate": 2.405660377358491e-05,
      "loss": 0.5371,
      "step": 102
    },
    {
      "epoch": 0.14630681818181818,
      "grad_norm": 0.2797533996478957,
      "learning_rate": 2.429245283018868e-05,
      "loss": 0.5217,
      "step": 103
    },
    {
      "epoch": 0.14772727272727273,
      "grad_norm": 0.33914390512383774,
      "learning_rate": 2.4528301886792453e-05,
      "loss": 0.5239,
      "step": 104
    },
    {
      "epoch": 0.14914772727272727,
      "grad_norm": 0.3534983625120479,
      "learning_rate": 2.476415094339623e-05,
      "loss": 0.5358,
      "step": 105
    },
    {
      "epoch": 0.15056818181818182,
      "grad_norm": 0.33548811483738605,
      "learning_rate": 2.5e-05,
      "loss": 0.5452,
      "step": 106
    },
    {
      "epoch": 0.15198863636363635,
      "grad_norm": 0.32354876306615044,
      "learning_rate": 2.5235849056603777e-05,
      "loss": 0.5217,
      "step": 107
    },
    {
      "epoch": 0.1534090909090909,
      "grad_norm": 0.3703652747216364,
      "learning_rate": 2.547169811320755e-05,
      "loss": 0.5264,
      "step": 108
    },
    {
      "epoch": 0.15482954545454544,
      "grad_norm": 0.29847306613098507,
      "learning_rate": 2.5707547169811325e-05,
      "loss": 0.5163,
      "step": 109
    },
    {
      "epoch": 0.15625,
      "grad_norm": 0.4366993388525997,
      "learning_rate": 2.5943396226415094e-05,
      "loss": 0.522,
      "step": 110
    },
    {
      "epoch": 0.15767045454545456,
      "grad_norm": 0.3672762494485308,
      "learning_rate": 2.6179245283018873e-05,
      "loss": 0.5418,
      "step": 111
    },
    {
      "epoch": 0.1590909090909091,
      "grad_norm": 0.4008755094232181,
      "learning_rate": 2.641509433962264e-05,
      "loss": 0.5261,
      "step": 112
    },
    {
      "epoch": 0.16051136363636365,
      "grad_norm": 0.43812901459158643,
      "learning_rate": 2.6650943396226417e-05,
      "loss": 0.5301,
      "step": 113
    },
    {
      "epoch": 0.16193181818181818,
      "grad_norm": 0.4084182791105245,
      "learning_rate": 2.688679245283019e-05,
      "loss": 0.5183,
      "step": 114
    },
    {
      "epoch": 0.16335227272727273,
      "grad_norm": 0.41155498450006367,
      "learning_rate": 2.7122641509433965e-05,
      "loss": 0.5184,
      "step": 115
    },
    {
      "epoch": 0.16477272727272727,
      "grad_norm": 0.41014739268836575,
      "learning_rate": 2.7358490566037738e-05,
      "loss": 0.5313,
      "step": 116
    },
    {
      "epoch": 0.16619318181818182,
      "grad_norm": 0.3407187576468219,
      "learning_rate": 2.7594339622641513e-05,
      "loss": 0.5188,
      "step": 117
    },
    {
      "epoch": 0.16761363636363635,
      "grad_norm": 0.36291129049080234,
      "learning_rate": 2.7830188679245282e-05,
      "loss": 0.5182,
      "step": 118
    },
    {
      "epoch": 0.1690340909090909,
      "grad_norm": 0.3339379643513558,
      "learning_rate": 2.806603773584906e-05,
      "loss": 0.5189,
      "step": 119
    },
    {
      "epoch": 0.17045454545454544,
      "grad_norm": 0.36896480633085016,
      "learning_rate": 2.830188679245283e-05,
      "loss": 0.5091,
      "step": 120
    },
    {
      "epoch": 0.171875,
      "grad_norm": 0.36512120511586627,
      "learning_rate": 2.8537735849056606e-05,
      "loss": 0.5041,
      "step": 121
    },
    {
      "epoch": 0.17329545454545456,
      "grad_norm": 0.4582599343757955,
      "learning_rate": 2.8773584905660378e-05,
      "loss": 0.524,
      "step": 122
    },
    {
      "epoch": 0.1747159090909091,
      "grad_norm": 0.40095197985225417,
      "learning_rate": 2.9009433962264154e-05,
      "loss": 0.5244,
      "step": 123
    },
    {
      "epoch": 0.17613636363636365,
      "grad_norm": 0.37188778720718213,
      "learning_rate": 2.9245283018867926e-05,
      "loss": 0.4994,
      "step": 124
    },
    {
      "epoch": 0.17755681818181818,
      "grad_norm": 0.4208092294932796,
      "learning_rate": 2.9481132075471702e-05,
      "loss": 0.5062,
      "step": 125
    },
    {
      "epoch": 0.17897727272727273,
      "grad_norm": 0.36078312688929426,
      "learning_rate": 2.971698113207547e-05,
      "loss": 0.4955,
      "step": 126
    },
    {
      "epoch": 0.18039772727272727,
      "grad_norm": 0.38074450574793756,
      "learning_rate": 2.995283018867925e-05,
      "loss": 0.5022,
      "step": 127
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 0.48828149180456576,
      "learning_rate": 3.018867924528302e-05,
      "loss": 0.5143,
      "step": 128
    },
    {
      "epoch": 0.18323863636363635,
      "grad_norm": 0.3657939551306818,
      "learning_rate": 3.0424528301886794e-05,
      "loss": 0.5342,
      "step": 129
    },
    {
      "epoch": 0.1846590909090909,
      "grad_norm": 0.5538003442959814,
      "learning_rate": 3.0660377358490567e-05,
      "loss": 0.4959,
      "step": 130
    },
    {
      "epoch": 0.18607954545454544,
      "grad_norm": 0.4086387901227155,
      "learning_rate": 3.0896226415094346e-05,
      "loss": 0.5021,
      "step": 131
    },
    {
      "epoch": 0.1875,
      "grad_norm": 0.4202075727471731,
      "learning_rate": 3.113207547169811e-05,
      "loss": 0.5172,
      "step": 132
    },
    {
      "epoch": 0.18892045454545456,
      "grad_norm": 0.502870022468119,
      "learning_rate": 3.136792452830189e-05,
      "loss": 0.5065,
      "step": 133
    },
    {
      "epoch": 0.1903409090909091,
      "grad_norm": 0.3723601550164188,
      "learning_rate": 3.160377358490566e-05,
      "loss": 0.5152,
      "step": 134
    },
    {
      "epoch": 0.19176136363636365,
      "grad_norm": 0.44684604891147883,
      "learning_rate": 3.1839622641509435e-05,
      "loss": 0.5081,
      "step": 135
    },
    {
      "epoch": 0.19318181818181818,
      "grad_norm": 0.5147511714464937,
      "learning_rate": 3.207547169811321e-05,
      "loss": 0.5382,
      "step": 136
    },
    {
      "epoch": 0.19460227272727273,
      "grad_norm": 0.38702334910521086,
      "learning_rate": 3.2311320754716986e-05,
      "loss": 0.4959,
      "step": 137
    },
    {
      "epoch": 0.19602272727272727,
      "grad_norm": 0.5469805636197131,
      "learning_rate": 3.254716981132075e-05,
      "loss": 0.5224,
      "step": 138
    },
    {
      "epoch": 0.19744318181818182,
      "grad_norm": 0.3597107095440647,
      "learning_rate": 3.278301886792453e-05,
      "loss": 0.4852,
      "step": 139
    },
    {
      "epoch": 0.19886363636363635,
      "grad_norm": 0.5032372592627841,
      "learning_rate": 3.30188679245283e-05,
      "loss": 0.5068,
      "step": 140
    },
    {
      "epoch": 0.2002840909090909,
      "grad_norm": 0.4131142868746613,
      "learning_rate": 3.3254716981132075e-05,
      "loss": 0.5066,
      "step": 141
    },
    {
      "epoch": 0.20170454545454544,
      "grad_norm": 0.4535422787375006,
      "learning_rate": 3.349056603773585e-05,
      "loss": 0.4883,
      "step": 142
    },
    {
      "epoch": 0.203125,
      "grad_norm": 0.44094551464390075,
      "learning_rate": 3.3726415094339627e-05,
      "loss": 0.4938,
      "step": 143
    },
    {
      "epoch": 0.20454545454545456,
      "grad_norm": 0.4201427402946494,
      "learning_rate": 3.39622641509434e-05,
      "loss": 0.5028,
      "step": 144
    },
    {
      "epoch": 0.2059659090909091,
      "grad_norm": 0.4660626594745151,
      "learning_rate": 3.419811320754717e-05,
      "loss": 0.5231,
      "step": 145
    },
    {
      "epoch": 0.20738636363636365,
      "grad_norm": 0.3807708341739199,
      "learning_rate": 3.4433962264150943e-05,
      "loss": 0.4952,
      "step": 146
    },
    {
      "epoch": 0.20880681818181818,
      "grad_norm": 0.4204324949486002,
      "learning_rate": 3.466981132075472e-05,
      "loss": 0.5276,
      "step": 147
    },
    {
      "epoch": 0.21022727272727273,
      "grad_norm": 0.4712703262206408,
      "learning_rate": 3.490566037735849e-05,
      "loss": 0.5061,
      "step": 148
    },
    {
      "epoch": 0.21164772727272727,
      "grad_norm": 0.3998703512012177,
      "learning_rate": 3.514150943396227e-05,
      "loss": 0.4925,
      "step": 149
    },
    {
      "epoch": 0.21306818181818182,
      "grad_norm": 0.4846931837889489,
      "learning_rate": 3.537735849056604e-05,
      "loss": 0.5078,
      "step": 150
    },
    {
      "epoch": 0.21448863636363635,
      "grad_norm": 0.38604433277600286,
      "learning_rate": 3.561320754716981e-05,
      "loss": 0.4823,
      "step": 151
    },
    {
      "epoch": 0.2159090909090909,
      "grad_norm": 0.5818849755369959,
      "learning_rate": 3.5849056603773584e-05,
      "loss": 0.4886,
      "step": 152
    },
    {
      "epoch": 0.21732954545454544,
      "grad_norm": 0.3420537391506494,
      "learning_rate": 3.608490566037736e-05,
      "loss": 0.5208,
      "step": 153
    },
    {
      "epoch": 0.21875,
      "grad_norm": 0.6399065514163873,
      "learning_rate": 3.632075471698113e-05,
      "loss": 0.4992,
      "step": 154
    },
    {
      "epoch": 0.22017045454545456,
      "grad_norm": 0.3674008446717818,
      "learning_rate": 3.655660377358491e-05,
      "loss": 0.5024,
      "step": 155
    },
    {
      "epoch": 0.2215909090909091,
      "grad_norm": 0.5601590281778308,
      "learning_rate": 3.679245283018868e-05,
      "loss": 0.4801,
      "step": 156
    },
    {
      "epoch": 0.22301136363636365,
      "grad_norm": 0.4617194361453004,
      "learning_rate": 3.702830188679245e-05,
      "loss": 0.4916,
      "step": 157
    },
    {
      "epoch": 0.22443181818181818,
      "grad_norm": 0.4450048047867621,
      "learning_rate": 3.7264150943396224e-05,
      "loss": 0.4906,
      "step": 158
    },
    {
      "epoch": 0.22585227272727273,
      "grad_norm": 0.46634707443247975,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.4916,
      "step": 159
    },
    {
      "epoch": 0.22727272727272727,
      "grad_norm": 0.4317325375822225,
      "learning_rate": 3.7735849056603776e-05,
      "loss": 0.489,
      "step": 160
    },
    {
      "epoch": 0.22869318181818182,
      "grad_norm": 0.41800755689037533,
      "learning_rate": 3.797169811320755e-05,
      "loss": 0.4722,
      "step": 161
    },
    {
      "epoch": 0.23011363636363635,
      "grad_norm": 0.4765724625813527,
      "learning_rate": 3.820754716981133e-05,
      "loss": 0.4885,
      "step": 162
    },
    {
      "epoch": 0.2315340909090909,
      "grad_norm": 0.4605907744173066,
      "learning_rate": 3.844339622641509e-05,
      "loss": 0.5115,
      "step": 163
    },
    {
      "epoch": 0.23295454545454544,
      "grad_norm": 0.4536647003593576,
      "learning_rate": 3.867924528301887e-05,
      "loss": 0.5226,
      "step": 164
    },
    {
      "epoch": 0.234375,
      "grad_norm": 0.45725726017704665,
      "learning_rate": 3.8915094339622644e-05,
      "loss": 0.4902,
      "step": 165
    },
    {
      "epoch": 0.23579545454545456,
      "grad_norm": 0.35738638665262795,
      "learning_rate": 3.9150943396226416e-05,
      "loss": 0.495,
      "step": 166
    },
    {
      "epoch": 0.2372159090909091,
      "grad_norm": 0.42239334597125705,
      "learning_rate": 3.938679245283019e-05,
      "loss": 0.5068,
      "step": 167
    },
    {
      "epoch": 0.23863636363636365,
      "grad_norm": 0.383196409732039,
      "learning_rate": 3.962264150943397e-05,
      "loss": 0.4768,
      "step": 168
    },
    {
      "epoch": 0.24005681818181818,
      "grad_norm": 0.4343427413651625,
      "learning_rate": 3.985849056603774e-05,
      "loss": 0.5182,
      "step": 169
    },
    {
      "epoch": 0.24147727272727273,
      "grad_norm": 0.37996912589056914,
      "learning_rate": 4.009433962264151e-05,
      "loss": 0.5017,
      "step": 170
    },
    {
      "epoch": 0.24289772727272727,
      "grad_norm": 0.46513692962972714,
      "learning_rate": 4.0330188679245284e-05,
      "loss": 0.4837,
      "step": 171
    },
    {
      "epoch": 0.24431818181818182,
      "grad_norm": 0.5022329387126359,
      "learning_rate": 4.0566037735849064e-05,
      "loss": 0.5035,
      "step": 172
    },
    {
      "epoch": 0.24573863636363635,
      "grad_norm": 0.3643438231153161,
      "learning_rate": 4.080188679245283e-05,
      "loss": 0.497,
      "step": 173
    },
    {
      "epoch": 0.2471590909090909,
      "grad_norm": 0.3855725737909487,
      "learning_rate": 4.103773584905661e-05,
      "loss": 0.5007,
      "step": 174
    },
    {
      "epoch": 0.24857954545454544,
      "grad_norm": 0.473643487488157,
      "learning_rate": 4.127358490566038e-05,
      "loss": 0.5176,
      "step": 175
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4902902281603711,
      "learning_rate": 4.150943396226415e-05,
      "loss": 0.4956,
      "step": 176
    },
    {
      "epoch": 0.25142045454545453,
      "grad_norm": 0.39551255844148264,
      "learning_rate": 4.1745283018867925e-05,
      "loss": 0.4778,
      "step": 177
    },
    {
      "epoch": 0.2528409090909091,
      "grad_norm": 0.44016268161488037,
      "learning_rate": 4.1981132075471704e-05,
      "loss": 0.4868,
      "step": 178
    },
    {
      "epoch": 0.25426136363636365,
      "grad_norm": 0.4858438754778914,
      "learning_rate": 4.221698113207547e-05,
      "loss": 0.5051,
      "step": 179
    },
    {
      "epoch": 0.2556818181818182,
      "grad_norm": 0.4222577744809464,
      "learning_rate": 4.245283018867925e-05,
      "loss": 0.5054,
      "step": 180
    },
    {
      "epoch": 0.2571022727272727,
      "grad_norm": 0.47305443246779116,
      "learning_rate": 4.268867924528302e-05,
      "loss": 0.4871,
      "step": 181
    },
    {
      "epoch": 0.2585227272727273,
      "grad_norm": 0.39490852894282275,
      "learning_rate": 4.292452830188679e-05,
      "loss": 0.4976,
      "step": 182
    },
    {
      "epoch": 0.2599431818181818,
      "grad_norm": 0.5227847253611165,
      "learning_rate": 4.3160377358490565e-05,
      "loss": 0.4857,
      "step": 183
    },
    {
      "epoch": 0.26136363636363635,
      "grad_norm": 0.5488926142742995,
      "learning_rate": 4.3396226415094345e-05,
      "loss": 0.5189,
      "step": 184
    },
    {
      "epoch": 0.2627840909090909,
      "grad_norm": 0.4522053350330469,
      "learning_rate": 4.363207547169812e-05,
      "loss": 0.4818,
      "step": 185
    },
    {
      "epoch": 0.26420454545454547,
      "grad_norm": 0.5103041933620538,
      "learning_rate": 4.386792452830189e-05,
      "loss": 0.4978,
      "step": 186
    },
    {
      "epoch": 0.265625,
      "grad_norm": 0.45100412955307045,
      "learning_rate": 4.410377358490566e-05,
      "loss": 0.4847,
      "step": 187
    },
    {
      "epoch": 0.26704545454545453,
      "grad_norm": 0.4976892941738726,
      "learning_rate": 4.433962264150944e-05,
      "loss": 0.4791,
      "step": 188
    },
    {
      "epoch": 0.2684659090909091,
      "grad_norm": 0.4203249700826567,
      "learning_rate": 4.4575471698113206e-05,
      "loss": 0.4756,
      "step": 189
    },
    {
      "epoch": 0.26988636363636365,
      "grad_norm": 0.5294337499764774,
      "learning_rate": 4.4811320754716985e-05,
      "loss": 0.4904,
      "step": 190
    },
    {
      "epoch": 0.2713068181818182,
      "grad_norm": 0.49961161848903474,
      "learning_rate": 4.504716981132076e-05,
      "loss": 0.4791,
      "step": 191
    },
    {
      "epoch": 0.2727272727272727,
      "grad_norm": 0.4623103280579072,
      "learning_rate": 4.528301886792453e-05,
      "loss": 0.4972,
      "step": 192
    },
    {
      "epoch": 0.2741477272727273,
      "grad_norm": 0.36791543025016576,
      "learning_rate": 4.55188679245283e-05,
      "loss": 0.4806,
      "step": 193
    },
    {
      "epoch": 0.2755681818181818,
      "grad_norm": 0.4318595072411048,
      "learning_rate": 4.575471698113208e-05,
      "loss": 0.4901,
      "step": 194
    },
    {
      "epoch": 0.27698863636363635,
      "grad_norm": 0.5234665150862851,
      "learning_rate": 4.5990566037735846e-05,
      "loss": 0.5214,
      "step": 195
    },
    {
      "epoch": 0.2784090909090909,
      "grad_norm": 0.4844820260402496,
      "learning_rate": 4.6226415094339625e-05,
      "loss": 0.4904,
      "step": 196
    },
    {
      "epoch": 0.27982954545454547,
      "grad_norm": 0.41287644442646665,
      "learning_rate": 4.64622641509434e-05,
      "loss": 0.4779,
      "step": 197
    },
    {
      "epoch": 0.28125,
      "grad_norm": 0.3868155691538918,
      "learning_rate": 4.669811320754717e-05,
      "loss": 0.5026,
      "step": 198
    },
    {
      "epoch": 0.28267045454545453,
      "grad_norm": 0.4084256516851678,
      "learning_rate": 4.693396226415094e-05,
      "loss": 0.4836,
      "step": 199
    },
    {
      "epoch": 0.2840909090909091,
      "grad_norm": 0.44015252347882244,
      "learning_rate": 4.716981132075472e-05,
      "loss": 0.5009,
      "step": 200
    },
    {
      "epoch": 0.28551136363636365,
      "grad_norm": 0.5014025060737504,
      "learning_rate": 4.7405660377358494e-05,
      "loss": 0.5076,
      "step": 201
    },
    {
      "epoch": 0.2869318181818182,
      "grad_norm": 0.4182655622797628,
      "learning_rate": 4.7641509433962266e-05,
      "loss": 0.4783,
      "step": 202
    },
    {
      "epoch": 0.2883522727272727,
      "grad_norm": 0.3696114567993049,
      "learning_rate": 4.787735849056604e-05,
      "loss": 0.4823,
      "step": 203
    },
    {
      "epoch": 0.2897727272727273,
      "grad_norm": 0.4666423727962833,
      "learning_rate": 4.811320754716982e-05,
      "loss": 0.4863,
      "step": 204
    },
    {
      "epoch": 0.2911931818181818,
      "grad_norm": 0.45860361328036575,
      "learning_rate": 4.834905660377358e-05,
      "loss": 0.49,
      "step": 205
    },
    {
      "epoch": 0.29261363636363635,
      "grad_norm": 0.4514617274489974,
      "learning_rate": 4.858490566037736e-05,
      "loss": 0.4847,
      "step": 206
    },
    {
      "epoch": 0.2940340909090909,
      "grad_norm": 0.42269999089328214,
      "learning_rate": 4.8820754716981134e-05,
      "loss": 0.4802,
      "step": 207
    },
    {
      "epoch": 0.29545454545454547,
      "grad_norm": 0.5206490473722941,
      "learning_rate": 4.9056603773584906e-05,
      "loss": 0.4689,
      "step": 208
    },
    {
      "epoch": 0.296875,
      "grad_norm": 0.815000165422235,
      "learning_rate": 4.929245283018868e-05,
      "loss": 0.4923,
      "step": 209
    },
    {
      "epoch": 0.29829545454545453,
      "grad_norm": 0.6996159279450798,
      "learning_rate": 4.952830188679246e-05,
      "loss": 0.4912,
      "step": 210
    },
    {
      "epoch": 0.2997159090909091,
      "grad_norm": 0.5016484278714396,
      "learning_rate": 4.976415094339622e-05,
      "loss": 0.4939,
      "step": 211
    },
    {
      "epoch": 0.30113636363636365,
      "grad_norm": 0.5408181232692402,
      "learning_rate": 5e-05,
      "loss": 0.4896,
      "step": 212
    },
    {
      "epoch": 0.3025568181818182,
      "grad_norm": 0.5597259232152089,
      "learning_rate": 4.9973684210526314e-05,
      "loss": 0.492,
      "step": 213
    },
    {
      "epoch": 0.3039772727272727,
      "grad_norm": 0.5756220325054142,
      "learning_rate": 4.994736842105263e-05,
      "loss": 0.4719,
      "step": 214
    },
    {
      "epoch": 0.3053977272727273,
      "grad_norm": 0.44308465658964863,
      "learning_rate": 4.992105263157895e-05,
      "loss": 0.4866,
      "step": 215
    },
    {
      "epoch": 0.3068181818181818,
      "grad_norm": 0.5342941064700469,
      "learning_rate": 4.989473684210527e-05,
      "loss": 0.4858,
      "step": 216
    },
    {
      "epoch": 0.30823863636363635,
      "grad_norm": 0.4607384838180216,
      "learning_rate": 4.986842105263158e-05,
      "loss": 0.4671,
      "step": 217
    },
    {
      "epoch": 0.3096590909090909,
      "grad_norm": 0.4305169824921994,
      "learning_rate": 4.984210526315789e-05,
      "loss": 0.4735,
      "step": 218
    },
    {
      "epoch": 0.31107954545454547,
      "grad_norm": 0.5011775758398513,
      "learning_rate": 4.981578947368422e-05,
      "loss": 0.4711,
      "step": 219
    },
    {
      "epoch": 0.3125,
      "grad_norm": 0.46134351134447044,
      "learning_rate": 4.978947368421053e-05,
      "loss": 0.4819,
      "step": 220
    },
    {
      "epoch": 0.31392045454545453,
      "grad_norm": 0.5443889829264801,
      "learning_rate": 4.976315789473685e-05,
      "loss": 0.4923,
      "step": 221
    },
    {
      "epoch": 0.3153409090909091,
      "grad_norm": 0.43051768015172315,
      "learning_rate": 4.973684210526316e-05,
      "loss": 0.4711,
      "step": 222
    },
    {
      "epoch": 0.31676136363636365,
      "grad_norm": 0.4935372667139101,
      "learning_rate": 4.971052631578948e-05,
      "loss": 0.4904,
      "step": 223
    },
    {
      "epoch": 0.3181818181818182,
      "grad_norm": 0.45660366075587117,
      "learning_rate": 4.9684210526315796e-05,
      "loss": 0.4995,
      "step": 224
    },
    {
      "epoch": 0.3196022727272727,
      "grad_norm": 0.4741209382901837,
      "learning_rate": 4.965789473684211e-05,
      "loss": 0.5035,
      "step": 225
    },
    {
      "epoch": 0.3210227272727273,
      "grad_norm": 0.5329136759798478,
      "learning_rate": 4.9631578947368426e-05,
      "loss": 0.4829,
      "step": 226
    },
    {
      "epoch": 0.3224431818181818,
      "grad_norm": 0.5307831005858525,
      "learning_rate": 4.960526315789474e-05,
      "loss": 0.4782,
      "step": 227
    },
    {
      "epoch": 0.32386363636363635,
      "grad_norm": 0.7288629764935626,
      "learning_rate": 4.9578947368421055e-05,
      "loss": 0.4559,
      "step": 228
    },
    {
      "epoch": 0.3252840909090909,
      "grad_norm": 0.37769535374881186,
      "learning_rate": 4.9552631578947374e-05,
      "loss": 0.469,
      "step": 229
    },
    {
      "epoch": 0.32670454545454547,
      "grad_norm": 0.6850263233558431,
      "learning_rate": 4.9526315789473685e-05,
      "loss": 0.4877,
      "step": 230
    },
    {
      "epoch": 0.328125,
      "grad_norm": 0.7507927737488479,
      "learning_rate": 4.9500000000000004e-05,
      "loss": 0.4712,
      "step": 231
    },
    {
      "epoch": 0.32954545454545453,
      "grad_norm": 0.5187863776427161,
      "learning_rate": 4.9473684210526315e-05,
      "loss": 0.4817,
      "step": 232
    },
    {
      "epoch": 0.3309659090909091,
      "grad_norm": 0.6994073848803729,
      "learning_rate": 4.9447368421052634e-05,
      "loss": 0.4674,
      "step": 233
    },
    {
      "epoch": 0.33238636363636365,
      "grad_norm": 0.8761464099294459,
      "learning_rate": 4.942105263157895e-05,
      "loss": 0.4954,
      "step": 234
    },
    {
      "epoch": 0.3338068181818182,
      "grad_norm": 0.5080341963961035,
      "learning_rate": 4.9394736842105264e-05,
      "loss": 0.4818,
      "step": 235
    },
    {
      "epoch": 0.3352272727272727,
      "grad_norm": 0.536231233159386,
      "learning_rate": 4.936842105263158e-05,
      "loss": 0.495,
      "step": 236
    },
    {
      "epoch": 0.3366477272727273,
      "grad_norm": 0.566908382561944,
      "learning_rate": 4.9342105263157894e-05,
      "loss": 0.5001,
      "step": 237
    },
    {
      "epoch": 0.3380681818181818,
      "grad_norm": 0.4287857949347938,
      "learning_rate": 4.931578947368421e-05,
      "loss": 0.4767,
      "step": 238
    },
    {
      "epoch": 0.33948863636363635,
      "grad_norm": 0.573740701256386,
      "learning_rate": 4.928947368421053e-05,
      "loss": 0.4737,
      "step": 239
    },
    {
      "epoch": 0.3409090909090909,
      "grad_norm": 0.5877167227944023,
      "learning_rate": 4.926315789473684e-05,
      "loss": 0.519,
      "step": 240
    },
    {
      "epoch": 0.34232954545454547,
      "grad_norm": 0.6161619622789617,
      "learning_rate": 4.923684210526316e-05,
      "loss": 0.4698,
      "step": 241
    },
    {
      "epoch": 0.34375,
      "grad_norm": 0.7570155247853391,
      "learning_rate": 4.921052631578947e-05,
      "loss": 0.4726,
      "step": 242
    },
    {
      "epoch": 0.34517045454545453,
      "grad_norm": 0.46051958573466517,
      "learning_rate": 4.91842105263158e-05,
      "loss": 0.4884,
      "step": 243
    },
    {
      "epoch": 0.3465909090909091,
      "grad_norm": 0.7235381327366881,
      "learning_rate": 4.915789473684211e-05,
      "loss": 0.4775,
      "step": 244
    },
    {
      "epoch": 0.34801136363636365,
      "grad_norm": 0.5807380806816569,
      "learning_rate": 4.913157894736842e-05,
      "loss": 0.4689,
      "step": 245
    },
    {
      "epoch": 0.3494318181818182,
      "grad_norm": 0.5146799296670865,
      "learning_rate": 4.910526315789474e-05,
      "loss": 0.4782,
      "step": 246
    },
    {
      "epoch": 0.3508522727272727,
      "grad_norm": 0.7027954093606257,
      "learning_rate": 4.907894736842106e-05,
      "loss": 0.4842,
      "step": 247
    },
    {
      "epoch": 0.3522727272727273,
      "grad_norm": 0.5286098675412559,
      "learning_rate": 4.9052631578947375e-05,
      "loss": 0.4976,
      "step": 248
    },
    {
      "epoch": 0.3536931818181818,
      "grad_norm": 0.49120938116698787,
      "learning_rate": 4.902631578947369e-05,
      "loss": 0.468,
      "step": 249
    },
    {
      "epoch": 0.35511363636363635,
      "grad_norm": 0.5307282832808795,
      "learning_rate": 4.9e-05,
      "loss": 0.4746,
      "step": 250
    },
    {
      "epoch": 0.3565340909090909,
      "grad_norm": 0.4102318783096874,
      "learning_rate": 4.897368421052632e-05,
      "loss": 0.4615,
      "step": 251
    },
    {
      "epoch": 0.35795454545454547,
      "grad_norm": 0.6029326040666113,
      "learning_rate": 4.8947368421052635e-05,
      "loss": 0.4752,
      "step": 252
    },
    {
      "epoch": 0.359375,
      "grad_norm": 0.5398226229862518,
      "learning_rate": 4.8921052631578953e-05,
      "loss": 0.4577,
      "step": 253
    },
    {
      "epoch": 0.36079545454545453,
      "grad_norm": 0.6120036144900686,
      "learning_rate": 4.8894736842105265e-05,
      "loss": 0.4818,
      "step": 254
    },
    {
      "epoch": 0.3622159090909091,
      "grad_norm": 0.5167651144845148,
      "learning_rate": 4.886842105263158e-05,
      "loss": 0.4755,
      "step": 255
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 0.5625224625629944,
      "learning_rate": 4.8842105263157895e-05,
      "loss": 0.5007,
      "step": 256
    },
    {
      "epoch": 0.3650568181818182,
      "grad_norm": 0.44492712751039537,
      "learning_rate": 4.881578947368421e-05,
      "loss": 0.4697,
      "step": 257
    },
    {
      "epoch": 0.3664772727272727,
      "grad_norm": 0.582482439417126,
      "learning_rate": 4.878947368421053e-05,
      "loss": 0.499,
      "step": 258
    },
    {
      "epoch": 0.3678977272727273,
      "grad_norm": 0.46960996014868633,
      "learning_rate": 4.876315789473684e-05,
      "loss": 0.4859,
      "step": 259
    },
    {
      "epoch": 0.3693181818181818,
      "grad_norm": 0.5234400773865269,
      "learning_rate": 4.873684210526316e-05,
      "loss": 0.4886,
      "step": 260
    },
    {
      "epoch": 0.37073863636363635,
      "grad_norm": 0.4335086871407323,
      "learning_rate": 4.871052631578948e-05,
      "loss": 0.4708,
      "step": 261
    },
    {
      "epoch": 0.3721590909090909,
      "grad_norm": 0.6007810626544366,
      "learning_rate": 4.868421052631579e-05,
      "loss": 0.4784,
      "step": 262
    },
    {
      "epoch": 0.37357954545454547,
      "grad_norm": 0.42321468079342384,
      "learning_rate": 4.865789473684211e-05,
      "loss": 0.4755,
      "step": 263
    },
    {
      "epoch": 0.375,
      "grad_norm": 0.5494653560950116,
      "learning_rate": 4.863157894736842e-05,
      "loss": 0.4707,
      "step": 264
    },
    {
      "epoch": 0.37642045454545453,
      "grad_norm": 0.5820558361410004,
      "learning_rate": 4.860526315789474e-05,
      "loss": 0.4356,
      "step": 265
    },
    {
      "epoch": 0.3778409090909091,
      "grad_norm": 0.5406557457244423,
      "learning_rate": 4.857894736842106e-05,
      "loss": 0.4895,
      "step": 266
    },
    {
      "epoch": 0.37926136363636365,
      "grad_norm": 0.48183053954034744,
      "learning_rate": 4.855263157894737e-05,
      "loss": 0.4789,
      "step": 267
    },
    {
      "epoch": 0.3806818181818182,
      "grad_norm": 0.5542937883158866,
      "learning_rate": 4.852631578947369e-05,
      "loss": 0.4787,
      "step": 268
    },
    {
      "epoch": 0.3821022727272727,
      "grad_norm": 0.5188092304158637,
      "learning_rate": 4.85e-05,
      "loss": 0.4727,
      "step": 269
    },
    {
      "epoch": 0.3835227272727273,
      "grad_norm": 0.4981975079245477,
      "learning_rate": 4.847368421052632e-05,
      "loss": 0.4638,
      "step": 270
    },
    {
      "epoch": 0.3849431818181818,
      "grad_norm": 0.4736038764160945,
      "learning_rate": 4.8447368421052637e-05,
      "loss": 0.4665,
      "step": 271
    },
    {
      "epoch": 0.38636363636363635,
      "grad_norm": 0.6653176307737855,
      "learning_rate": 4.842105263157895e-05,
      "loss": 0.4633,
      "step": 272
    },
    {
      "epoch": 0.3877840909090909,
      "grad_norm": 0.4444686863491031,
      "learning_rate": 4.8394736842105266e-05,
      "loss": 0.4782,
      "step": 273
    },
    {
      "epoch": 0.38920454545454547,
      "grad_norm": 0.46229733174970494,
      "learning_rate": 4.836842105263158e-05,
      "loss": 0.4755,
      "step": 274
    },
    {
      "epoch": 0.390625,
      "grad_norm": 0.4206833101505621,
      "learning_rate": 4.8342105263157896e-05,
      "loss": 0.4669,
      "step": 275
    },
    {
      "epoch": 0.39204545454545453,
      "grad_norm": 0.44628168210932034,
      "learning_rate": 4.8315789473684215e-05,
      "loss": 0.4774,
      "step": 276
    },
    {
      "epoch": 0.3934659090909091,
      "grad_norm": 0.3984643927762997,
      "learning_rate": 4.8289473684210526e-05,
      "loss": 0.463,
      "step": 277
    },
    {
      "epoch": 0.39488636363636365,
      "grad_norm": 0.3841925893370871,
      "learning_rate": 4.8263157894736845e-05,
      "loss": 0.4581,
      "step": 278
    },
    {
      "epoch": 0.3963068181818182,
      "grad_norm": 0.4406324488892526,
      "learning_rate": 4.8236842105263156e-05,
      "loss": 0.4835,
      "step": 279
    },
    {
      "epoch": 0.3977272727272727,
      "grad_norm": 0.37919825189854295,
      "learning_rate": 4.821052631578948e-05,
      "loss": 0.4671,
      "step": 280
    },
    {
      "epoch": 0.3991477272727273,
      "grad_norm": 0.5223956600282648,
      "learning_rate": 4.818421052631579e-05,
      "loss": 0.4688,
      "step": 281
    },
    {
      "epoch": 0.4005681818181818,
      "grad_norm": 0.37502113743094684,
      "learning_rate": 4.8157894736842105e-05,
      "loss": 0.4642,
      "step": 282
    },
    {
      "epoch": 0.40198863636363635,
      "grad_norm": 0.4653584967926116,
      "learning_rate": 4.813157894736842e-05,
      "loss": 0.4811,
      "step": 283
    },
    {
      "epoch": 0.4034090909090909,
      "grad_norm": 0.4614674731315193,
      "learning_rate": 4.8105263157894735e-05,
      "loss": 0.4625,
      "step": 284
    },
    {
      "epoch": 0.40482954545454547,
      "grad_norm": 0.4091362291946624,
      "learning_rate": 4.807894736842106e-05,
      "loss": 0.4593,
      "step": 285
    },
    {
      "epoch": 0.40625,
      "grad_norm": 0.5133821515166488,
      "learning_rate": 4.805263157894737e-05,
      "loss": 0.4802,
      "step": 286
    },
    {
      "epoch": 0.40767045454545453,
      "grad_norm": 0.40212773894399406,
      "learning_rate": 4.802631578947368e-05,
      "loss": 0.4503,
      "step": 287
    },
    {
      "epoch": 0.4090909090909091,
      "grad_norm": 0.5582657571160663,
      "learning_rate": 4.8e-05,
      "loss": 0.4845,
      "step": 288
    },
    {
      "epoch": 0.41051136363636365,
      "grad_norm": 0.43529498460911215,
      "learning_rate": 4.797368421052632e-05,
      "loss": 0.4626,
      "step": 289
    },
    {
      "epoch": 0.4119318181818182,
      "grad_norm": 0.6629838248040675,
      "learning_rate": 4.794736842105264e-05,
      "loss": 0.4781,
      "step": 290
    },
    {
      "epoch": 0.4133522727272727,
      "grad_norm": 0.47129460917821914,
      "learning_rate": 4.792105263157895e-05,
      "loss": 0.4777,
      "step": 291
    },
    {
      "epoch": 0.4147727272727273,
      "grad_norm": 0.5749878248589911,
      "learning_rate": 4.789473684210526e-05,
      "loss": 0.4298,
      "step": 292
    },
    {
      "epoch": 0.4161931818181818,
      "grad_norm": 0.38693098993857133,
      "learning_rate": 4.786842105263158e-05,
      "loss": 0.4408,
      "step": 293
    },
    {
      "epoch": 0.41761363636363635,
      "grad_norm": 0.47205887816680336,
      "learning_rate": 4.78421052631579e-05,
      "loss": 0.4712,
      "step": 294
    },
    {
      "epoch": 0.4190340909090909,
      "grad_norm": 0.4560828447848661,
      "learning_rate": 4.7815789473684216e-05,
      "loss": 0.4684,
      "step": 295
    },
    {
      "epoch": 0.42045454545454547,
      "grad_norm": 0.4240395858391297,
      "learning_rate": 4.778947368421053e-05,
      "loss": 0.4709,
      "step": 296
    },
    {
      "epoch": 0.421875,
      "grad_norm": 0.40228768083524186,
      "learning_rate": 4.7763157894736846e-05,
      "loss": 0.4766,
      "step": 297
    },
    {
      "epoch": 0.42329545454545453,
      "grad_norm": 0.3871328833215933,
      "learning_rate": 4.773684210526316e-05,
      "loss": 0.468,
      "step": 298
    },
    {
      "epoch": 0.4247159090909091,
      "grad_norm": 0.47481360310301135,
      "learning_rate": 4.7710526315789476e-05,
      "loss": 0.4687,
      "step": 299
    },
    {
      "epoch": 0.42613636363636365,
      "grad_norm": 0.41353416110144886,
      "learning_rate": 4.7684210526315794e-05,
      "loss": 0.4601,
      "step": 300
    },
    {
      "epoch": 0.4275568181818182,
      "grad_norm": 0.34982575569178254,
      "learning_rate": 4.7657894736842106e-05,
      "loss": 0.4487,
      "step": 301
    },
    {
      "epoch": 0.4289772727272727,
      "grad_norm": 0.4604820214265997,
      "learning_rate": 4.7631578947368424e-05,
      "loss": 0.4773,
      "step": 302
    },
    {
      "epoch": 0.4303977272727273,
      "grad_norm": 0.4526129141571692,
      "learning_rate": 4.760526315789474e-05,
      "loss": 0.4871,
      "step": 303
    },
    {
      "epoch": 0.4318181818181818,
      "grad_norm": 0.4635500699824651,
      "learning_rate": 4.7578947368421054e-05,
      "loss": 0.454,
      "step": 304
    },
    {
      "epoch": 0.43323863636363635,
      "grad_norm": 0.33250600434929356,
      "learning_rate": 4.755263157894737e-05,
      "loss": 0.4377,
      "step": 305
    },
    {
      "epoch": 0.4346590909090909,
      "grad_norm": 0.3854930671742253,
      "learning_rate": 4.7526315789473684e-05,
      "loss": 0.4828,
      "step": 306
    },
    {
      "epoch": 0.43607954545454547,
      "grad_norm": 0.35831263000372304,
      "learning_rate": 4.75e-05,
      "loss": 0.445,
      "step": 307
    },
    {
      "epoch": 0.4375,
      "grad_norm": 0.33179531579529115,
      "learning_rate": 4.747368421052632e-05,
      "loss": 0.4481,
      "step": 308
    },
    {
      "epoch": 0.43892045454545453,
      "grad_norm": 0.443902189364721,
      "learning_rate": 4.744736842105263e-05,
      "loss": 0.4683,
      "step": 309
    },
    {
      "epoch": 0.4403409090909091,
      "grad_norm": 0.40820143170425943,
      "learning_rate": 4.742105263157895e-05,
      "loss": 0.4959,
      "step": 310
    },
    {
      "epoch": 0.44176136363636365,
      "grad_norm": 0.46658445516984226,
      "learning_rate": 4.739473684210526e-05,
      "loss": 0.447,
      "step": 311
    },
    {
      "epoch": 0.4431818181818182,
      "grad_norm": 0.34973736712438586,
      "learning_rate": 4.736842105263158e-05,
      "loss": 0.4531,
      "step": 312
    },
    {
      "epoch": 0.4446022727272727,
      "grad_norm": 0.4724337224616794,
      "learning_rate": 4.73421052631579e-05,
      "loss": 0.4392,
      "step": 313
    },
    {
      "epoch": 0.4460227272727273,
      "grad_norm": 0.4210968323215773,
      "learning_rate": 4.731578947368421e-05,
      "loss": 0.4675,
      "step": 314
    },
    {
      "epoch": 0.4474431818181818,
      "grad_norm": 0.5630868315436802,
      "learning_rate": 4.728947368421053e-05,
      "loss": 0.4743,
      "step": 315
    },
    {
      "epoch": 0.44886363636363635,
      "grad_norm": 0.4206407448174499,
      "learning_rate": 4.726315789473684e-05,
      "loss": 0.4765,
      "step": 316
    },
    {
      "epoch": 0.4502840909090909,
      "grad_norm": 0.4614390235430736,
      "learning_rate": 4.7236842105263166e-05,
      "loss": 0.4504,
      "step": 317
    },
    {
      "epoch": 0.45170454545454547,
      "grad_norm": 0.41761449043679494,
      "learning_rate": 4.721052631578948e-05,
      "loss": 0.4558,
      "step": 318
    },
    {
      "epoch": 0.453125,
      "grad_norm": 0.6226231451104491,
      "learning_rate": 4.718421052631579e-05,
      "loss": 0.478,
      "step": 319
    },
    {
      "epoch": 0.45454545454545453,
      "grad_norm": 0.5275835245563256,
      "learning_rate": 4.715789473684211e-05,
      "loss": 0.4779,
      "step": 320
    },
    {
      "epoch": 0.4559659090909091,
      "grad_norm": 0.47443465634455223,
      "learning_rate": 4.713157894736842e-05,
      "loss": 0.469,
      "step": 321
    },
    {
      "epoch": 0.45738636363636365,
      "grad_norm": 0.5258097643145201,
      "learning_rate": 4.7105263157894744e-05,
      "loss": 0.4381,
      "step": 322
    },
    {
      "epoch": 0.4588068181818182,
      "grad_norm": 0.4908532774686488,
      "learning_rate": 4.7078947368421056e-05,
      "loss": 0.4636,
      "step": 323
    },
    {
      "epoch": 0.4602272727272727,
      "grad_norm": 0.4380538898064296,
      "learning_rate": 4.705263157894737e-05,
      "loss": 0.4573,
      "step": 324
    },
    {
      "epoch": 0.4616477272727273,
      "grad_norm": 0.4211322287406041,
      "learning_rate": 4.7026315789473686e-05,
      "loss": 0.4502,
      "step": 325
    },
    {
      "epoch": 0.4630681818181818,
      "grad_norm": 0.4509962057643071,
      "learning_rate": 4.7e-05,
      "loss": 0.4514,
      "step": 326
    },
    {
      "epoch": 0.46448863636363635,
      "grad_norm": 0.45333958361355725,
      "learning_rate": 4.697368421052632e-05,
      "loss": 0.4654,
      "step": 327
    },
    {
      "epoch": 0.4659090909090909,
      "grad_norm": 0.4827264263468752,
      "learning_rate": 4.6947368421052634e-05,
      "loss": 0.4527,
      "step": 328
    },
    {
      "epoch": 0.46732954545454547,
      "grad_norm": 0.5136080409775614,
      "learning_rate": 4.6921052631578946e-05,
      "loss": 0.464,
      "step": 329
    },
    {
      "epoch": 0.46875,
      "grad_norm": 0.5166994674491183,
      "learning_rate": 4.6894736842105264e-05,
      "loss": 0.4518,
      "step": 330
    },
    {
      "epoch": 0.47017045454545453,
      "grad_norm": 0.4591232289098012,
      "learning_rate": 4.686842105263158e-05,
      "loss": 0.4664,
      "step": 331
    },
    {
      "epoch": 0.4715909090909091,
      "grad_norm": 0.449178594469784,
      "learning_rate": 4.68421052631579e-05,
      "loss": 0.4588,
      "step": 332
    },
    {
      "epoch": 0.47301136363636365,
      "grad_norm": 0.45322941918203974,
      "learning_rate": 4.681578947368421e-05,
      "loss": 0.4426,
      "step": 333
    },
    {
      "epoch": 0.4744318181818182,
      "grad_norm": 0.44667265829043384,
      "learning_rate": 4.678947368421053e-05,
      "loss": 0.4429,
      "step": 334
    },
    {
      "epoch": 0.4758522727272727,
      "grad_norm": 0.49172695697490043,
      "learning_rate": 4.676315789473684e-05,
      "loss": 0.4449,
      "step": 335
    },
    {
      "epoch": 0.4772727272727273,
      "grad_norm": 0.4053270036469298,
      "learning_rate": 4.673684210526316e-05,
      "loss": 0.446,
      "step": 336
    },
    {
      "epoch": 0.4786931818181818,
      "grad_norm": 0.49345161348573013,
      "learning_rate": 4.671052631578948e-05,
      "loss": 0.4764,
      "step": 337
    },
    {
      "epoch": 0.48011363636363635,
      "grad_norm": 0.3963676472427591,
      "learning_rate": 4.668421052631579e-05,
      "loss": 0.4732,
      "step": 338
    },
    {
      "epoch": 0.4815340909090909,
      "grad_norm": 0.5364613301410666,
      "learning_rate": 4.665789473684211e-05,
      "loss": 0.4722,
      "step": 339
    },
    {
      "epoch": 0.48295454545454547,
      "grad_norm": 0.3789032257137947,
      "learning_rate": 4.663157894736842e-05,
      "loss": 0.4774,
      "step": 340
    },
    {
      "epoch": 0.484375,
      "grad_norm": 0.46785470498471365,
      "learning_rate": 4.660526315789474e-05,
      "loss": 0.4959,
      "step": 341
    },
    {
      "epoch": 0.48579545454545453,
      "grad_norm": 0.3752741266058422,
      "learning_rate": 4.657894736842106e-05,
      "loss": 0.4627,
      "step": 342
    },
    {
      "epoch": 0.4872159090909091,
      "grad_norm": 0.44622209427888565,
      "learning_rate": 4.655263157894737e-05,
      "loss": 0.4539,
      "step": 343
    },
    {
      "epoch": 0.48863636363636365,
      "grad_norm": 0.444810449628708,
      "learning_rate": 4.652631578947369e-05,
      "loss": 0.4658,
      "step": 344
    },
    {
      "epoch": 0.4900568181818182,
      "grad_norm": 0.3733983180544888,
      "learning_rate": 4.6500000000000005e-05,
      "loss": 0.4769,
      "step": 345
    },
    {
      "epoch": 0.4914772727272727,
      "grad_norm": 0.3662400417842176,
      "learning_rate": 4.647368421052632e-05,
      "loss": 0.4277,
      "step": 346
    },
    {
      "epoch": 0.4928977272727273,
      "grad_norm": 0.3692706408775056,
      "learning_rate": 4.6447368421052635e-05,
      "loss": 0.449,
      "step": 347
    },
    {
      "epoch": 0.4943181818181818,
      "grad_norm": 0.42024351456136216,
      "learning_rate": 4.642105263157895e-05,
      "loss": 0.4679,
      "step": 348
    },
    {
      "epoch": 0.49573863636363635,
      "grad_norm": 0.38013288368384596,
      "learning_rate": 4.6394736842105265e-05,
      "loss": 0.4764,
      "step": 349
    },
    {
      "epoch": 0.4971590909090909,
      "grad_norm": 0.40566882072361826,
      "learning_rate": 4.6368421052631584e-05,
      "loss": 0.4479,
      "step": 350
    },
    {
      "epoch": 0.49857954545454547,
      "grad_norm": 0.4110219431392234,
      "learning_rate": 4.6342105263157895e-05,
      "loss": 0.4514,
      "step": 351
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3595270423872282,
      "learning_rate": 4.6315789473684214e-05,
      "loss": 0.4485,
      "step": 352
    },
    {
      "epoch": 0.5014204545454546,
      "grad_norm": 0.4336681143256855,
      "learning_rate": 4.6289473684210525e-05,
      "loss": 0.4715,
      "step": 353
    },
    {
      "epoch": 0.5028409090909091,
      "grad_norm": 0.40163052055222737,
      "learning_rate": 4.6263157894736844e-05,
      "loss": 0.4359,
      "step": 354
    },
    {
      "epoch": 0.5042613636363636,
      "grad_norm": 0.3473207755775183,
      "learning_rate": 4.623684210526316e-05,
      "loss": 0.4541,
      "step": 355
    },
    {
      "epoch": 0.5056818181818182,
      "grad_norm": 0.3911360756673772,
      "learning_rate": 4.6210526315789473e-05,
      "loss": 0.4569,
      "step": 356
    },
    {
      "epoch": 0.5071022727272727,
      "grad_norm": 0.38602420058739106,
      "learning_rate": 4.618421052631579e-05,
      "loss": 0.4564,
      "step": 357
    },
    {
      "epoch": 0.5085227272727273,
      "grad_norm": 0.3658003970695547,
      "learning_rate": 4.6157894736842103e-05,
      "loss": 0.4566,
      "step": 358
    },
    {
      "epoch": 0.5099431818181818,
      "grad_norm": 0.43351503356553794,
      "learning_rate": 4.613157894736843e-05,
      "loss": 0.4553,
      "step": 359
    },
    {
      "epoch": 0.5113636363636364,
      "grad_norm": 0.3913006205840493,
      "learning_rate": 4.610526315789474e-05,
      "loss": 0.4468,
      "step": 360
    },
    {
      "epoch": 0.5127840909090909,
      "grad_norm": 0.3448666674929392,
      "learning_rate": 4.607894736842105e-05,
      "loss": 0.4467,
      "step": 361
    },
    {
      "epoch": 0.5142045454545454,
      "grad_norm": 0.4291915889540231,
      "learning_rate": 4.605263157894737e-05,
      "loss": 0.4599,
      "step": 362
    },
    {
      "epoch": 0.515625,
      "grad_norm": 0.40484348256322633,
      "learning_rate": 4.602631578947368e-05,
      "loss": 0.4805,
      "step": 363
    },
    {
      "epoch": 0.5170454545454546,
      "grad_norm": 0.4196624774808661,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.465,
      "step": 364
    },
    {
      "epoch": 0.5184659090909091,
      "grad_norm": 0.3760134295020477,
      "learning_rate": 4.597368421052632e-05,
      "loss": 0.4782,
      "step": 365
    },
    {
      "epoch": 0.5198863636363636,
      "grad_norm": 0.3998800544890235,
      "learning_rate": 4.594736842105264e-05,
      "loss": 0.453,
      "step": 366
    },
    {
      "epoch": 0.5213068181818182,
      "grad_norm": 0.35086020714691535,
      "learning_rate": 4.592105263157895e-05,
      "loss": 0.4631,
      "step": 367
    },
    {
      "epoch": 0.5227272727272727,
      "grad_norm": 0.33538244655917604,
      "learning_rate": 4.589473684210526e-05,
      "loss": 0.4711,
      "step": 368
    },
    {
      "epoch": 0.5241477272727273,
      "grad_norm": 0.3944626795937277,
      "learning_rate": 4.5868421052631585e-05,
      "loss": 0.4759,
      "step": 369
    },
    {
      "epoch": 0.5255681818181818,
      "grad_norm": 0.452608952114295,
      "learning_rate": 4.58421052631579e-05,
      "loss": 0.4601,
      "step": 370
    },
    {
      "epoch": 0.5269886363636364,
      "grad_norm": 0.34139849866813166,
      "learning_rate": 4.5815789473684215e-05,
      "loss": 0.4548,
      "step": 371
    },
    {
      "epoch": 0.5284090909090909,
      "grad_norm": 0.4091615894083717,
      "learning_rate": 4.5789473684210527e-05,
      "loss": 0.4368,
      "step": 372
    },
    {
      "epoch": 0.5298295454545454,
      "grad_norm": 0.3494541951127624,
      "learning_rate": 4.5763157894736845e-05,
      "loss": 0.46,
      "step": 373
    },
    {
      "epoch": 0.53125,
      "grad_norm": 0.3758079408401562,
      "learning_rate": 4.573684210526316e-05,
      "loss": 0.4572,
      "step": 374
    },
    {
      "epoch": 0.5326704545454546,
      "grad_norm": 0.3931802556499837,
      "learning_rate": 4.5710526315789475e-05,
      "loss": 0.4457,
      "step": 375
    },
    {
      "epoch": 0.5340909090909091,
      "grad_norm": 0.40264751425586986,
      "learning_rate": 4.568421052631579e-05,
      "loss": 0.447,
      "step": 376
    },
    {
      "epoch": 0.5355113636363636,
      "grad_norm": 0.4229880169451451,
      "learning_rate": 4.5657894736842105e-05,
      "loss": 0.4539,
      "step": 377
    },
    {
      "epoch": 0.5369318181818182,
      "grad_norm": 0.5224687048568715,
      "learning_rate": 4.563157894736842e-05,
      "loss": 0.4724,
      "step": 378
    },
    {
      "epoch": 0.5383522727272727,
      "grad_norm": 0.3745502349503197,
      "learning_rate": 4.560526315789474e-05,
      "loss": 0.4584,
      "step": 379
    },
    {
      "epoch": 0.5397727272727273,
      "grad_norm": 0.38511492560974026,
      "learning_rate": 4.557894736842105e-05,
      "loss": 0.4595,
      "step": 380
    },
    {
      "epoch": 0.5411931818181818,
      "grad_norm": 0.45883677403477646,
      "learning_rate": 4.555263157894737e-05,
      "loss": 0.4619,
      "step": 381
    },
    {
      "epoch": 0.5426136363636364,
      "grad_norm": 0.3582921003676137,
      "learning_rate": 4.552631578947369e-05,
      "loss": 0.4517,
      "step": 382
    },
    {
      "epoch": 0.5440340909090909,
      "grad_norm": 0.4287431372302858,
      "learning_rate": 4.55e-05,
      "loss": 0.4501,
      "step": 383
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 0.47828685335481463,
      "learning_rate": 4.547368421052632e-05,
      "loss": 0.4579,
      "step": 384
    },
    {
      "epoch": 0.546875,
      "grad_norm": 0.4591725092943762,
      "learning_rate": 4.544736842105263e-05,
      "loss": 0.4833,
      "step": 385
    },
    {
      "epoch": 0.5482954545454546,
      "grad_norm": 0.4122564590892917,
      "learning_rate": 4.542105263157895e-05,
      "loss": 0.4456,
      "step": 386
    },
    {
      "epoch": 0.5497159090909091,
      "grad_norm": 0.4939587583986073,
      "learning_rate": 4.539473684210527e-05,
      "loss": 0.4845,
      "step": 387
    },
    {
      "epoch": 0.5511363636363636,
      "grad_norm": 0.4084654305038244,
      "learning_rate": 4.536842105263158e-05,
      "loss": 0.4931,
      "step": 388
    },
    {
      "epoch": 0.5525568181818182,
      "grad_norm": 0.3473984314387611,
      "learning_rate": 4.53421052631579e-05,
      "loss": 0.4467,
      "step": 389
    },
    {
      "epoch": 0.5539772727272727,
      "grad_norm": 0.3804099958911269,
      "learning_rate": 4.531578947368421e-05,
      "loss": 0.4675,
      "step": 390
    },
    {
      "epoch": 0.5553977272727273,
      "grad_norm": 0.36798755191984506,
      "learning_rate": 4.528947368421053e-05,
      "loss": 0.4499,
      "step": 391
    },
    {
      "epoch": 0.5568181818181818,
      "grad_norm": 0.38230910129789436,
      "learning_rate": 4.5263157894736846e-05,
      "loss": 0.4531,
      "step": 392
    },
    {
      "epoch": 0.5582386363636364,
      "grad_norm": 0.4070626431636088,
      "learning_rate": 4.523684210526316e-05,
      "loss": 0.463,
      "step": 393
    },
    {
      "epoch": 0.5596590909090909,
      "grad_norm": 0.36393375155949836,
      "learning_rate": 4.5210526315789476e-05,
      "loss": 0.4547,
      "step": 394
    },
    {
      "epoch": 0.5610795454545454,
      "grad_norm": 0.43371947452057663,
      "learning_rate": 4.518421052631579e-05,
      "loss": 0.4717,
      "step": 395
    },
    {
      "epoch": 0.5625,
      "grad_norm": 0.37073894474811026,
      "learning_rate": 4.515789473684211e-05,
      "loss": 0.4422,
      "step": 396
    },
    {
      "epoch": 0.5639204545454546,
      "grad_norm": 0.4028796939504037,
      "learning_rate": 4.5131578947368425e-05,
      "loss": 0.4486,
      "step": 397
    },
    {
      "epoch": 0.5653409090909091,
      "grad_norm": 0.5173497352389002,
      "learning_rate": 4.5105263157894736e-05,
      "loss": 0.4513,
      "step": 398
    },
    {
      "epoch": 0.5667613636363636,
      "grad_norm": 0.3622745041959469,
      "learning_rate": 4.5078947368421055e-05,
      "loss": 0.4234,
      "step": 399
    },
    {
      "epoch": 0.5681818181818182,
      "grad_norm": 0.6913291442972719,
      "learning_rate": 4.5052631578947366e-05,
      "loss": 0.4626,
      "step": 400
    },
    {
      "epoch": 0.5696022727272727,
      "grad_norm": 0.31599929624229794,
      "learning_rate": 4.502631578947369e-05,
      "loss": 0.4389,
      "step": 401
    },
    {
      "epoch": 0.5710227272727273,
      "grad_norm": 0.5912309179006723,
      "learning_rate": 4.5e-05,
      "loss": 0.4397,
      "step": 402
    },
    {
      "epoch": 0.5724431818181818,
      "grad_norm": 0.38007471165854123,
      "learning_rate": 4.497368421052632e-05,
      "loss": 0.4386,
      "step": 403
    },
    {
      "epoch": 0.5738636363636364,
      "grad_norm": 0.4167912123172323,
      "learning_rate": 4.494736842105263e-05,
      "loss": 0.4522,
      "step": 404
    },
    {
      "epoch": 0.5752840909090909,
      "grad_norm": 0.48464920130561573,
      "learning_rate": 4.4921052631578944e-05,
      "loss": 0.4529,
      "step": 405
    },
    {
      "epoch": 0.5767045454545454,
      "grad_norm": 0.3752298131247919,
      "learning_rate": 4.489473684210527e-05,
      "loss": 0.4522,
      "step": 406
    },
    {
      "epoch": 0.578125,
      "grad_norm": 0.5413468056111472,
      "learning_rate": 4.486842105263158e-05,
      "loss": 0.4602,
      "step": 407
    },
    {
      "epoch": 0.5795454545454546,
      "grad_norm": 0.36056725219520164,
      "learning_rate": 4.48421052631579e-05,
      "loss": 0.4538,
      "step": 408
    },
    {
      "epoch": 0.5809659090909091,
      "grad_norm": 0.4274865623624824,
      "learning_rate": 4.481578947368421e-05,
      "loss": 0.4469,
      "step": 409
    },
    {
      "epoch": 0.5823863636363636,
      "grad_norm": 0.39064671076971724,
      "learning_rate": 4.478947368421053e-05,
      "loss": 0.455,
      "step": 410
    },
    {
      "epoch": 0.5838068181818182,
      "grad_norm": 0.525238548242024,
      "learning_rate": 4.476315789473685e-05,
      "loss": 0.4308,
      "step": 411
    },
    {
      "epoch": 0.5852272727272727,
      "grad_norm": 0.39426486388300797,
      "learning_rate": 4.473684210526316e-05,
      "loss": 0.4674,
      "step": 412
    },
    {
      "epoch": 0.5866477272727273,
      "grad_norm": 0.5061162124098662,
      "learning_rate": 4.471052631578948e-05,
      "loss": 0.4696,
      "step": 413
    },
    {
      "epoch": 0.5880681818181818,
      "grad_norm": 0.3729533225746771,
      "learning_rate": 4.468421052631579e-05,
      "loss": 0.4601,
      "step": 414
    },
    {
      "epoch": 0.5894886363636364,
      "grad_norm": 0.36704068945906393,
      "learning_rate": 4.465789473684211e-05,
      "loss": 0.4648,
      "step": 415
    },
    {
      "epoch": 0.5909090909090909,
      "grad_norm": 0.36363487054355964,
      "learning_rate": 4.4631578947368426e-05,
      "loss": 0.4368,
      "step": 416
    },
    {
      "epoch": 0.5923295454545454,
      "grad_norm": 0.3264765472426919,
      "learning_rate": 4.460526315789474e-05,
      "loss": 0.4728,
      "step": 417
    },
    {
      "epoch": 0.59375,
      "grad_norm": 0.36810730735226976,
      "learning_rate": 4.4578947368421056e-05,
      "loss": 0.452,
      "step": 418
    },
    {
      "epoch": 0.5951704545454546,
      "grad_norm": 0.32796785535666356,
      "learning_rate": 4.455263157894737e-05,
      "loss": 0.4388,
      "step": 419
    },
    {
      "epoch": 0.5965909090909091,
      "grad_norm": 0.3727667018434383,
      "learning_rate": 4.4526315789473686e-05,
      "loss": 0.4578,
      "step": 420
    },
    {
      "epoch": 0.5980113636363636,
      "grad_norm": 0.3342662905742609,
      "learning_rate": 4.4500000000000004e-05,
      "loss": 0.4353,
      "step": 421
    },
    {
      "epoch": 0.5994318181818182,
      "grad_norm": 0.5149895753015494,
      "learning_rate": 4.4473684210526316e-05,
      "loss": 0.46,
      "step": 422
    },
    {
      "epoch": 0.6008522727272727,
      "grad_norm": 0.3153401921148877,
      "learning_rate": 4.4447368421052634e-05,
      "loss": 0.4389,
      "step": 423
    },
    {
      "epoch": 0.6022727272727273,
      "grad_norm": 0.4988548039070881,
      "learning_rate": 4.442105263157895e-05,
      "loss": 0.4499,
      "step": 424
    },
    {
      "epoch": 0.6036931818181818,
      "grad_norm": 0.48154394786568083,
      "learning_rate": 4.4394736842105264e-05,
      "loss": 0.4704,
      "step": 425
    },
    {
      "epoch": 0.6051136363636364,
      "grad_norm": 0.37106450090055865,
      "learning_rate": 4.436842105263158e-05,
      "loss": 0.4541,
      "step": 426
    },
    {
      "epoch": 0.6065340909090909,
      "grad_norm": 0.4475370992680974,
      "learning_rate": 4.4342105263157894e-05,
      "loss": 0.444,
      "step": 427
    },
    {
      "epoch": 0.6079545454545454,
      "grad_norm": 0.3898743976778191,
      "learning_rate": 4.431578947368421e-05,
      "loss": 0.4566,
      "step": 428
    },
    {
      "epoch": 0.609375,
      "grad_norm": 0.35760072545685284,
      "learning_rate": 4.428947368421053e-05,
      "loss": 0.4325,
      "step": 429
    },
    {
      "epoch": 0.6107954545454546,
      "grad_norm": 0.4557905384541361,
      "learning_rate": 4.426315789473684e-05,
      "loss": 0.4724,
      "step": 430
    },
    {
      "epoch": 0.6122159090909091,
      "grad_norm": 0.3922883956824982,
      "learning_rate": 4.423684210526316e-05,
      "loss": 0.4449,
      "step": 431
    },
    {
      "epoch": 0.6136363636363636,
      "grad_norm": 0.3626495011919139,
      "learning_rate": 4.421052631578947e-05,
      "loss": 0.4747,
      "step": 432
    },
    {
      "epoch": 0.6150568181818182,
      "grad_norm": 0.5378760941017343,
      "learning_rate": 4.418421052631579e-05,
      "loss": 0.4751,
      "step": 433
    },
    {
      "epoch": 0.6164772727272727,
      "grad_norm": 0.3227889211318601,
      "learning_rate": 4.415789473684211e-05,
      "loss": 0.4438,
      "step": 434
    },
    {
      "epoch": 0.6178977272727273,
      "grad_norm": 0.6181162856033461,
      "learning_rate": 4.413157894736842e-05,
      "loss": 0.4513,
      "step": 435
    },
    {
      "epoch": 0.6193181818181818,
      "grad_norm": 0.4649264574949569,
      "learning_rate": 4.410526315789474e-05,
      "loss": 0.467,
      "step": 436
    },
    {
      "epoch": 0.6207386363636364,
      "grad_norm": 0.3990879746208028,
      "learning_rate": 4.407894736842105e-05,
      "loss": 0.4344,
      "step": 437
    },
    {
      "epoch": 0.6221590909090909,
      "grad_norm": 0.44745396442654356,
      "learning_rate": 4.4052631578947376e-05,
      "loss": 0.4615,
      "step": 438
    },
    {
      "epoch": 0.6235795454545454,
      "grad_norm": 0.3747908994217127,
      "learning_rate": 4.402631578947369e-05,
      "loss": 0.4388,
      "step": 439
    },
    {
      "epoch": 0.625,
      "grad_norm": 0.48097640869230496,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.4522,
      "step": 440
    },
    {
      "epoch": 0.6264204545454546,
      "grad_norm": 0.6281995070999654,
      "learning_rate": 4.397368421052632e-05,
      "loss": 0.4465,
      "step": 441
    },
    {
      "epoch": 0.6278409090909091,
      "grad_norm": 0.34531604278717526,
      "learning_rate": 4.394736842105263e-05,
      "loss": 0.4346,
      "step": 442
    },
    {
      "epoch": 0.6292613636363636,
      "grad_norm": 0.6766158388004772,
      "learning_rate": 4.3921052631578954e-05,
      "loss": 0.4686,
      "step": 443
    },
    {
      "epoch": 0.6306818181818182,
      "grad_norm": 0.4952487507556938,
      "learning_rate": 4.3894736842105266e-05,
      "loss": 0.4588,
      "step": 444
    },
    {
      "epoch": 0.6321022727272727,
      "grad_norm": 0.5172058806902666,
      "learning_rate": 4.3868421052631584e-05,
      "loss": 0.4527,
      "step": 445
    },
    {
      "epoch": 0.6335227272727273,
      "grad_norm": 0.6137503722770618,
      "learning_rate": 4.3842105263157895e-05,
      "loss": 0.4187,
      "step": 446
    },
    {
      "epoch": 0.6349431818181818,
      "grad_norm": 0.423999732048008,
      "learning_rate": 4.381578947368421e-05,
      "loss": 0.4419,
      "step": 447
    },
    {
      "epoch": 0.6363636363636364,
      "grad_norm": 0.5912813181849411,
      "learning_rate": 4.378947368421053e-05,
      "loss": 0.4179,
      "step": 448
    },
    {
      "epoch": 0.6377840909090909,
      "grad_norm": 0.3839265201320063,
      "learning_rate": 4.3763157894736844e-05,
      "loss": 0.4393,
      "step": 449
    },
    {
      "epoch": 0.6392045454545454,
      "grad_norm": 0.682740318870617,
      "learning_rate": 4.373684210526316e-05,
      "loss": 0.4507,
      "step": 450
    },
    {
      "epoch": 0.640625,
      "grad_norm": 0.5046196233057137,
      "learning_rate": 4.3710526315789474e-05,
      "loss": 0.4477,
      "step": 451
    },
    {
      "epoch": 0.6420454545454546,
      "grad_norm": 0.5018021598310706,
      "learning_rate": 4.368421052631579e-05,
      "loss": 0.4497,
      "step": 452
    },
    {
      "epoch": 0.6434659090909091,
      "grad_norm": 0.5093467644773783,
      "learning_rate": 4.365789473684211e-05,
      "loss": 0.4445,
      "step": 453
    },
    {
      "epoch": 0.6448863636363636,
      "grad_norm": 0.5471654590306478,
      "learning_rate": 4.363157894736842e-05,
      "loss": 0.4372,
      "step": 454
    },
    {
      "epoch": 0.6463068181818182,
      "grad_norm": 0.39247038727146855,
      "learning_rate": 4.360526315789474e-05,
      "loss": 0.457,
      "step": 455
    },
    {
      "epoch": 0.6477272727272727,
      "grad_norm": 0.48980134016250737,
      "learning_rate": 4.357894736842105e-05,
      "loss": 0.4325,
      "step": 456
    },
    {
      "epoch": 0.6491477272727273,
      "grad_norm": 0.34622749106101264,
      "learning_rate": 4.355263157894737e-05,
      "loss": 0.4303,
      "step": 457
    },
    {
      "epoch": 0.6505681818181818,
      "grad_norm": 0.47291545376072436,
      "learning_rate": 4.352631578947369e-05,
      "loss": 0.4594,
      "step": 458
    },
    {
      "epoch": 0.6519886363636364,
      "grad_norm": 0.37064514395094,
      "learning_rate": 4.35e-05,
      "loss": 0.4435,
      "step": 459
    },
    {
      "epoch": 0.6534090909090909,
      "grad_norm": 0.42090057516553697,
      "learning_rate": 4.347368421052632e-05,
      "loss": 0.437,
      "step": 460
    },
    {
      "epoch": 0.6548295454545454,
      "grad_norm": 0.5002718706841304,
      "learning_rate": 4.344736842105263e-05,
      "loss": 0.4553,
      "step": 461
    },
    {
      "epoch": 0.65625,
      "grad_norm": 0.3752698201311098,
      "learning_rate": 4.342105263157895e-05,
      "loss": 0.4303,
      "step": 462
    },
    {
      "epoch": 0.6576704545454546,
      "grad_norm": 0.4235004691760855,
      "learning_rate": 4.339473684210527e-05,
      "loss": 0.4646,
      "step": 463
    },
    {
      "epoch": 0.6590909090909091,
      "grad_norm": 0.33906752745904567,
      "learning_rate": 4.336842105263158e-05,
      "loss": 0.4386,
      "step": 464
    },
    {
      "epoch": 0.6605113636363636,
      "grad_norm": 0.41860763687223285,
      "learning_rate": 4.33421052631579e-05,
      "loss": 0.4408,
      "step": 465
    },
    {
      "epoch": 0.6619318181818182,
      "grad_norm": 0.3872923163379341,
      "learning_rate": 4.3315789473684215e-05,
      "loss": 0.4591,
      "step": 466
    },
    {
      "epoch": 0.6633522727272727,
      "grad_norm": 0.3447828859355511,
      "learning_rate": 4.328947368421053e-05,
      "loss": 0.4317,
      "step": 467
    },
    {
      "epoch": 0.6647727272727273,
      "grad_norm": 0.37933295436249453,
      "learning_rate": 4.3263157894736845e-05,
      "loss": 0.4416,
      "step": 468
    },
    {
      "epoch": 0.6661931818181818,
      "grad_norm": 0.34860323977059193,
      "learning_rate": 4.323684210526316e-05,
      "loss": 0.4365,
      "step": 469
    },
    {
      "epoch": 0.6676136363636364,
      "grad_norm": 0.3825290614459148,
      "learning_rate": 4.3210526315789475e-05,
      "loss": 0.4343,
      "step": 470
    },
    {
      "epoch": 0.6690340909090909,
      "grad_norm": 0.4787253673000609,
      "learning_rate": 4.3184210526315793e-05,
      "loss": 0.4788,
      "step": 471
    },
    {
      "epoch": 0.6704545454545454,
      "grad_norm": 0.39431808606802915,
      "learning_rate": 4.3157894736842105e-05,
      "loss": 0.4511,
      "step": 472
    },
    {
      "epoch": 0.671875,
      "grad_norm": 0.44507746644306423,
      "learning_rate": 4.3131578947368423e-05,
      "loss": 0.465,
      "step": 473
    },
    {
      "epoch": 0.6732954545454546,
      "grad_norm": 0.405716645358215,
      "learning_rate": 4.3105263157894735e-05,
      "loss": 0.4487,
      "step": 474
    },
    {
      "epoch": 0.6747159090909091,
      "grad_norm": 0.41461196210343965,
      "learning_rate": 4.307894736842105e-05,
      "loss": 0.4673,
      "step": 475
    },
    {
      "epoch": 0.6761363636363636,
      "grad_norm": 0.39908764622398,
      "learning_rate": 4.305263157894737e-05,
      "loss": 0.447,
      "step": 476
    },
    {
      "epoch": 0.6775568181818182,
      "grad_norm": 0.4059549120120126,
      "learning_rate": 4.302631578947369e-05,
      "loss": 0.4578,
      "step": 477
    },
    {
      "epoch": 0.6789772727272727,
      "grad_norm": 0.31895051467823665,
      "learning_rate": 4.3e-05,
      "loss": 0.4394,
      "step": 478
    },
    {
      "epoch": 0.6803977272727273,
      "grad_norm": 0.3761773686216746,
      "learning_rate": 4.297368421052631e-05,
      "loss": 0.4331,
      "step": 479
    },
    {
      "epoch": 0.6818181818181818,
      "grad_norm": 0.33652977526848027,
      "learning_rate": 4.294736842105264e-05,
      "loss": 0.4351,
      "step": 480
    },
    {
      "epoch": 0.6832386363636364,
      "grad_norm": 0.3541781733512332,
      "learning_rate": 4.292105263157895e-05,
      "loss": 0.4443,
      "step": 481
    },
    {
      "epoch": 0.6846590909090909,
      "grad_norm": 0.36232078758040803,
      "learning_rate": 4.289473684210527e-05,
      "loss": 0.4392,
      "step": 482
    },
    {
      "epoch": 0.6860795454545454,
      "grad_norm": 0.42369354103859913,
      "learning_rate": 4.286842105263158e-05,
      "loss": 0.4332,
      "step": 483
    },
    {
      "epoch": 0.6875,
      "grad_norm": 0.36346225370343505,
      "learning_rate": 4.284210526315789e-05,
      "loss": 0.4496,
      "step": 484
    },
    {
      "epoch": 0.6889204545454546,
      "grad_norm": 0.3933391174114948,
      "learning_rate": 4.281578947368422e-05,
      "loss": 0.4494,
      "step": 485
    },
    {
      "epoch": 0.6903409090909091,
      "grad_norm": 0.39989001693119874,
      "learning_rate": 4.278947368421053e-05,
      "loss": 0.4375,
      "step": 486
    },
    {
      "epoch": 0.6917613636363636,
      "grad_norm": 0.3457492093182776,
      "learning_rate": 4.2763157894736847e-05,
      "loss": 0.4491,
      "step": 487
    },
    {
      "epoch": 0.6931818181818182,
      "grad_norm": 0.3594537826266957,
      "learning_rate": 4.273684210526316e-05,
      "loss": 0.418,
      "step": 488
    },
    {
      "epoch": 0.6946022727272727,
      "grad_norm": 0.3439406485864203,
      "learning_rate": 4.271052631578947e-05,
      "loss": 0.4547,
      "step": 489
    },
    {
      "epoch": 0.6960227272727273,
      "grad_norm": 0.4268193771400363,
      "learning_rate": 4.2684210526315795e-05,
      "loss": 0.4538,
      "step": 490
    },
    {
      "epoch": 0.6974431818181818,
      "grad_norm": 0.34676010305276483,
      "learning_rate": 4.2657894736842106e-05,
      "loss": 0.4403,
      "step": 491
    },
    {
      "epoch": 0.6988636363636364,
      "grad_norm": 0.35001229418387636,
      "learning_rate": 4.2631578947368425e-05,
      "loss": 0.4432,
      "step": 492
    },
    {
      "epoch": 0.7002840909090909,
      "grad_norm": 0.49023872191655726,
      "learning_rate": 4.2605263157894736e-05,
      "loss": 0.4297,
      "step": 493
    },
    {
      "epoch": 0.7017045454545454,
      "grad_norm": 0.31480573912271076,
      "learning_rate": 4.2578947368421055e-05,
      "loss": 0.4251,
      "step": 494
    },
    {
      "epoch": 0.703125,
      "grad_norm": 0.32595300564735175,
      "learning_rate": 4.255263157894737e-05,
      "loss": 0.4415,
      "step": 495
    },
    {
      "epoch": 0.7045454545454546,
      "grad_norm": 0.36580604246079124,
      "learning_rate": 4.2526315789473685e-05,
      "loss": 0.4314,
      "step": 496
    },
    {
      "epoch": 0.7059659090909091,
      "grad_norm": 0.3571870745328893,
      "learning_rate": 4.25e-05,
      "loss": 0.4219,
      "step": 497
    },
    {
      "epoch": 0.7073863636363636,
      "grad_norm": 0.3808345169606755,
      "learning_rate": 4.2473684210526315e-05,
      "loss": 0.4367,
      "step": 498
    },
    {
      "epoch": 0.7088068181818182,
      "grad_norm": 0.48507935679218994,
      "learning_rate": 4.244736842105263e-05,
      "loss": 0.4684,
      "step": 499
    },
    {
      "epoch": 0.7102272727272727,
      "grad_norm": 0.42606971135627986,
      "learning_rate": 4.242105263157895e-05,
      "loss": 0.4432,
      "step": 500
    },
    {
      "epoch": 0.7116477272727273,
      "grad_norm": 0.44332516132128275,
      "learning_rate": 4.239473684210526e-05,
      "loss": 0.4349,
      "step": 501
    },
    {
      "epoch": 0.7130681818181818,
      "grad_norm": 0.3703330710460626,
      "learning_rate": 4.236842105263158e-05,
      "loss": 0.4344,
      "step": 502
    },
    {
      "epoch": 0.7144886363636364,
      "grad_norm": 0.4274874021941476,
      "learning_rate": 4.234210526315789e-05,
      "loss": 0.4343,
      "step": 503
    },
    {
      "epoch": 0.7159090909090909,
      "grad_norm": 0.35730924286155014,
      "learning_rate": 4.231578947368421e-05,
      "loss": 0.4427,
      "step": 504
    },
    {
      "epoch": 0.7173295454545454,
      "grad_norm": 0.43735255249937904,
      "learning_rate": 4.228947368421053e-05,
      "loss": 0.4285,
      "step": 505
    },
    {
      "epoch": 0.71875,
      "grad_norm": 0.34166421492755333,
      "learning_rate": 4.226315789473684e-05,
      "loss": 0.4367,
      "step": 506
    },
    {
      "epoch": 0.7201704545454546,
      "grad_norm": 0.47275946094879284,
      "learning_rate": 4.223684210526316e-05,
      "loss": 0.4465,
      "step": 507
    },
    {
      "epoch": 0.7215909090909091,
      "grad_norm": 0.3391658965563464,
      "learning_rate": 4.221052631578948e-05,
      "loss": 0.4512,
      "step": 508
    },
    {
      "epoch": 0.7230113636363636,
      "grad_norm": 0.3736717893466626,
      "learning_rate": 4.218421052631579e-05,
      "loss": 0.4356,
      "step": 509
    },
    {
      "epoch": 0.7244318181818182,
      "grad_norm": 0.44525108514497724,
      "learning_rate": 4.215789473684211e-05,
      "loss": 0.4504,
      "step": 510
    },
    {
      "epoch": 0.7258522727272727,
      "grad_norm": 0.33131519264385484,
      "learning_rate": 4.213157894736842e-05,
      "loss": 0.4288,
      "step": 511
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 0.3799960659232681,
      "learning_rate": 4.210526315789474e-05,
      "loss": 0.4306,
      "step": 512
    },
    {
      "epoch": 0.7286931818181818,
      "grad_norm": 0.32118927178240336,
      "learning_rate": 4.2078947368421056e-05,
      "loss": 0.4311,
      "step": 513
    },
    {
      "epoch": 0.7301136363636364,
      "grad_norm": 0.4452262921918583,
      "learning_rate": 4.2052631578947375e-05,
      "loss": 0.4455,
      "step": 514
    },
    {
      "epoch": 0.7315340909090909,
      "grad_norm": 0.32407131891184654,
      "learning_rate": 4.2026315789473686e-05,
      "loss": 0.4321,
      "step": 515
    },
    {
      "epoch": 0.7329545454545454,
      "grad_norm": 0.35335240236723353,
      "learning_rate": 4.2e-05,
      "loss": 0.4545,
      "step": 516
    },
    {
      "epoch": 0.734375,
      "grad_norm": 0.3942348474311365,
      "learning_rate": 4.1973684210526316e-05,
      "loss": 0.4586,
      "step": 517
    },
    {
      "epoch": 0.7357954545454546,
      "grad_norm": 0.358914376832063,
      "learning_rate": 4.1947368421052634e-05,
      "loss": 0.426,
      "step": 518
    },
    {
      "epoch": 0.7372159090909091,
      "grad_norm": 0.38742054622241945,
      "learning_rate": 4.192105263157895e-05,
      "loss": 0.4214,
      "step": 519
    },
    {
      "epoch": 0.7386363636363636,
      "grad_norm": 0.4223684399663011,
      "learning_rate": 4.1894736842105264e-05,
      "loss": 0.4473,
      "step": 520
    },
    {
      "epoch": 0.7400568181818182,
      "grad_norm": 0.39622831211404397,
      "learning_rate": 4.1868421052631576e-05,
      "loss": 0.4428,
      "step": 521
    },
    {
      "epoch": 0.7414772727272727,
      "grad_norm": 0.4002936234798669,
      "learning_rate": 4.18421052631579e-05,
      "loss": 0.464,
      "step": 522
    },
    {
      "epoch": 0.7428977272727273,
      "grad_norm": 0.3518467057986245,
      "learning_rate": 4.181578947368421e-05,
      "loss": 0.45,
      "step": 523
    },
    {
      "epoch": 0.7443181818181818,
      "grad_norm": 0.4547849872780403,
      "learning_rate": 4.178947368421053e-05,
      "loss": 0.4486,
      "step": 524
    },
    {
      "epoch": 0.7457386363636364,
      "grad_norm": 0.35644049916317966,
      "learning_rate": 4.176315789473684e-05,
      "loss": 0.4466,
      "step": 525
    },
    {
      "epoch": 0.7471590909090909,
      "grad_norm": 0.3643051727741714,
      "learning_rate": 4.1736842105263154e-05,
      "loss": 0.4404,
      "step": 526
    },
    {
      "epoch": 0.7485795454545454,
      "grad_norm": 0.3634878675565472,
      "learning_rate": 4.171052631578948e-05,
      "loss": 0.4429,
      "step": 527
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36346586438314143,
      "learning_rate": 4.168421052631579e-05,
      "loss": 0.4357,
      "step": 528
    },
    {
      "epoch": 0.7514204545454546,
      "grad_norm": 0.3577309141003053,
      "learning_rate": 4.165789473684211e-05,
      "loss": 0.4507,
      "step": 529
    },
    {
      "epoch": 0.7528409090909091,
      "grad_norm": 0.4328543350765463,
      "learning_rate": 4.163157894736842e-05,
      "loss": 0.4379,
      "step": 530
    },
    {
      "epoch": 0.7542613636363636,
      "grad_norm": 0.3664737900006198,
      "learning_rate": 4.160526315789474e-05,
      "loss": 0.4609,
      "step": 531
    },
    {
      "epoch": 0.7556818181818182,
      "grad_norm": 0.361559650909458,
      "learning_rate": 4.157894736842106e-05,
      "loss": 0.4501,
      "step": 532
    },
    {
      "epoch": 0.7571022727272727,
      "grad_norm": 0.3675610046427887,
      "learning_rate": 4.155263157894737e-05,
      "loss": 0.4529,
      "step": 533
    },
    {
      "epoch": 0.7585227272727273,
      "grad_norm": 0.34574092912054083,
      "learning_rate": 4.152631578947369e-05,
      "loss": 0.4327,
      "step": 534
    },
    {
      "epoch": 0.7599431818181818,
      "grad_norm": 0.3314420520913311,
      "learning_rate": 4.15e-05,
      "loss": 0.449,
      "step": 535
    },
    {
      "epoch": 0.7613636363636364,
      "grad_norm": 0.37758528132763436,
      "learning_rate": 4.147368421052632e-05,
      "loss": 0.4366,
      "step": 536
    },
    {
      "epoch": 0.7627840909090909,
      "grad_norm": 0.37299081468892664,
      "learning_rate": 4.1447368421052636e-05,
      "loss": 0.4291,
      "step": 537
    },
    {
      "epoch": 0.7642045454545454,
      "grad_norm": 0.346849539104924,
      "learning_rate": 4.142105263157895e-05,
      "loss": 0.4314,
      "step": 538
    },
    {
      "epoch": 0.765625,
      "grad_norm": 0.38268850916084507,
      "learning_rate": 4.1394736842105266e-05,
      "loss": 0.4364,
      "step": 539
    },
    {
      "epoch": 0.7670454545454546,
      "grad_norm": 0.41576375562283135,
      "learning_rate": 4.136842105263158e-05,
      "loss": 0.434,
      "step": 540
    },
    {
      "epoch": 0.7684659090909091,
      "grad_norm": 0.4100340998957794,
      "learning_rate": 4.1342105263157896e-05,
      "loss": 0.4354,
      "step": 541
    },
    {
      "epoch": 0.7698863636363636,
      "grad_norm": 0.3412658452708126,
      "learning_rate": 4.1315789473684214e-05,
      "loss": 0.4539,
      "step": 542
    },
    {
      "epoch": 0.7713068181818182,
      "grad_norm": 0.3847761548615778,
      "learning_rate": 4.1289473684210526e-05,
      "loss": 0.4651,
      "step": 543
    },
    {
      "epoch": 0.7727272727272727,
      "grad_norm": 0.3519101783492497,
      "learning_rate": 4.1263157894736844e-05,
      "loss": 0.4304,
      "step": 544
    },
    {
      "epoch": 0.7741477272727273,
      "grad_norm": 0.3956715920309705,
      "learning_rate": 4.123684210526316e-05,
      "loss": 0.4229,
      "step": 545
    },
    {
      "epoch": 0.7755681818181818,
      "grad_norm": 0.4154400236797322,
      "learning_rate": 4.1210526315789474e-05,
      "loss": 0.4346,
      "step": 546
    },
    {
      "epoch": 0.7769886363636364,
      "grad_norm": 0.36103792316845945,
      "learning_rate": 4.118421052631579e-05,
      "loss": 0.4433,
      "step": 547
    },
    {
      "epoch": 0.7784090909090909,
      "grad_norm": 0.35318956763040643,
      "learning_rate": 4.1157894736842104e-05,
      "loss": 0.4388,
      "step": 548
    },
    {
      "epoch": 0.7798295454545454,
      "grad_norm": 0.37793112073017276,
      "learning_rate": 4.113157894736842e-05,
      "loss": 0.4367,
      "step": 549
    },
    {
      "epoch": 0.78125,
      "grad_norm": 0.34021873228286564,
      "learning_rate": 4.110526315789474e-05,
      "loss": 0.4389,
      "step": 550
    },
    {
      "epoch": 0.7826704545454546,
      "grad_norm": 0.35373004942355485,
      "learning_rate": 4.107894736842106e-05,
      "loss": 0.4426,
      "step": 551
    },
    {
      "epoch": 0.7840909090909091,
      "grad_norm": 0.40265438549752497,
      "learning_rate": 4.105263157894737e-05,
      "loss": 0.4257,
      "step": 552
    },
    {
      "epoch": 0.7855113636363636,
      "grad_norm": 0.33407966129453315,
      "learning_rate": 4.102631578947368e-05,
      "loss": 0.4405,
      "step": 553
    },
    {
      "epoch": 0.7869318181818182,
      "grad_norm": 0.5103952269578559,
      "learning_rate": 4.1e-05,
      "loss": 0.4487,
      "step": 554
    },
    {
      "epoch": 0.7883522727272727,
      "grad_norm": 0.3947058960330454,
      "learning_rate": 4.097368421052632e-05,
      "loss": 0.4593,
      "step": 555
    },
    {
      "epoch": 0.7897727272727273,
      "grad_norm": 0.42795495812988754,
      "learning_rate": 4.094736842105264e-05,
      "loss": 0.4351,
      "step": 556
    },
    {
      "epoch": 0.7911931818181818,
      "grad_norm": 0.4396660461351393,
      "learning_rate": 4.092105263157895e-05,
      "loss": 0.4258,
      "step": 557
    },
    {
      "epoch": 0.7926136363636364,
      "grad_norm": 0.4398793346762401,
      "learning_rate": 4.089473684210526e-05,
      "loss": 0.4392,
      "step": 558
    },
    {
      "epoch": 0.7940340909090909,
      "grad_norm": 0.3832023406842286,
      "learning_rate": 4.0868421052631585e-05,
      "loss": 0.4406,
      "step": 559
    },
    {
      "epoch": 0.7954545454545454,
      "grad_norm": 0.44525330180300193,
      "learning_rate": 4.08421052631579e-05,
      "loss": 0.4686,
      "step": 560
    },
    {
      "epoch": 0.796875,
      "grad_norm": 0.5391658132465169,
      "learning_rate": 4.0815789473684215e-05,
      "loss": 0.4458,
      "step": 561
    },
    {
      "epoch": 0.7982954545454546,
      "grad_norm": 0.3653081352599652,
      "learning_rate": 4.078947368421053e-05,
      "loss": 0.4231,
      "step": 562
    },
    {
      "epoch": 0.7997159090909091,
      "grad_norm": 0.4667532612815708,
      "learning_rate": 4.076315789473684e-05,
      "loss": 0.4494,
      "step": 563
    },
    {
      "epoch": 0.8011363636363636,
      "grad_norm": 0.425013959387646,
      "learning_rate": 4.0736842105263164e-05,
      "loss": 0.4366,
      "step": 564
    },
    {
      "epoch": 0.8025568181818182,
      "grad_norm": 0.4859547865412071,
      "learning_rate": 4.0710526315789475e-05,
      "loss": 0.4284,
      "step": 565
    },
    {
      "epoch": 0.8039772727272727,
      "grad_norm": 0.35281368931923013,
      "learning_rate": 4.0684210526315794e-05,
      "loss": 0.4518,
      "step": 566
    },
    {
      "epoch": 0.8053977272727273,
      "grad_norm": 0.3966282623423134,
      "learning_rate": 4.0657894736842105e-05,
      "loss": 0.4314,
      "step": 567
    },
    {
      "epoch": 0.8068181818181818,
      "grad_norm": 0.37827578401097356,
      "learning_rate": 4.0631578947368424e-05,
      "loss": 0.4247,
      "step": 568
    },
    {
      "epoch": 0.8082386363636364,
      "grad_norm": 0.3250091335211562,
      "learning_rate": 4.060526315789474e-05,
      "loss": 0.4399,
      "step": 569
    },
    {
      "epoch": 0.8096590909090909,
      "grad_norm": 0.39849303245154755,
      "learning_rate": 4.0578947368421054e-05,
      "loss": 0.4218,
      "step": 570
    },
    {
      "epoch": 0.8110795454545454,
      "grad_norm": 0.3446078368092656,
      "learning_rate": 4.055263157894737e-05,
      "loss": 0.4494,
      "step": 571
    },
    {
      "epoch": 0.8125,
      "grad_norm": 0.31816399095772235,
      "learning_rate": 4.0526315789473684e-05,
      "loss": 0.4404,
      "step": 572
    },
    {
      "epoch": 0.8139204545454546,
      "grad_norm": 0.43639843761417824,
      "learning_rate": 4.05e-05,
      "loss": 0.4627,
      "step": 573
    },
    {
      "epoch": 0.8153409090909091,
      "grad_norm": 0.3465399150539147,
      "learning_rate": 4.047368421052632e-05,
      "loss": 0.4518,
      "step": 574
    },
    {
      "epoch": 0.8167613636363636,
      "grad_norm": 0.3854204630510848,
      "learning_rate": 4.044736842105263e-05,
      "loss": 0.4589,
      "step": 575
    },
    {
      "epoch": 0.8181818181818182,
      "grad_norm": 0.3820398290361902,
      "learning_rate": 4.042105263157895e-05,
      "loss": 0.4447,
      "step": 576
    },
    {
      "epoch": 0.8196022727272727,
      "grad_norm": 0.3008276248163918,
      "learning_rate": 4.039473684210526e-05,
      "loss": 0.4209,
      "step": 577
    },
    {
      "epoch": 0.8210227272727273,
      "grad_norm": 0.4244115963221484,
      "learning_rate": 4.036842105263158e-05,
      "loss": 0.4351,
      "step": 578
    },
    {
      "epoch": 0.8224431818181818,
      "grad_norm": 0.3631460138555324,
      "learning_rate": 4.03421052631579e-05,
      "loss": 0.4387,
      "step": 579
    },
    {
      "epoch": 0.8238636363636364,
      "grad_norm": 0.39425364483919134,
      "learning_rate": 4.031578947368421e-05,
      "loss": 0.4502,
      "step": 580
    },
    {
      "epoch": 0.8252840909090909,
      "grad_norm": 0.44111055826781564,
      "learning_rate": 4.028947368421053e-05,
      "loss": 0.4294,
      "step": 581
    },
    {
      "epoch": 0.8267045454545454,
      "grad_norm": 0.3946185922124401,
      "learning_rate": 4.026315789473684e-05,
      "loss": 0.4493,
      "step": 582
    },
    {
      "epoch": 0.828125,
      "grad_norm": 0.4868296134633893,
      "learning_rate": 4.023684210526316e-05,
      "loss": 0.4327,
      "step": 583
    },
    {
      "epoch": 0.8295454545454546,
      "grad_norm": 0.36031720909060233,
      "learning_rate": 4.021052631578948e-05,
      "loss": 0.4205,
      "step": 584
    },
    {
      "epoch": 0.8309659090909091,
      "grad_norm": 0.3797735804276419,
      "learning_rate": 4.018421052631579e-05,
      "loss": 0.4385,
      "step": 585
    },
    {
      "epoch": 0.8323863636363636,
      "grad_norm": 0.3852017758296038,
      "learning_rate": 4.015789473684211e-05,
      "loss": 0.459,
      "step": 586
    },
    {
      "epoch": 0.8338068181818182,
      "grad_norm": 0.4070511527526212,
      "learning_rate": 4.0131578947368425e-05,
      "loss": 0.4559,
      "step": 587
    },
    {
      "epoch": 0.8352272727272727,
      "grad_norm": 0.33863130024696003,
      "learning_rate": 4.010526315789474e-05,
      "loss": 0.433,
      "step": 588
    },
    {
      "epoch": 0.8366477272727273,
      "grad_norm": 0.3867588405259923,
      "learning_rate": 4.0078947368421055e-05,
      "loss": 0.4561,
      "step": 589
    },
    {
      "epoch": 0.8380681818181818,
      "grad_norm": 0.3974823526879778,
      "learning_rate": 4.0052631578947367e-05,
      "loss": 0.4191,
      "step": 590
    },
    {
      "epoch": 0.8394886363636364,
      "grad_norm": 0.35024820598950995,
      "learning_rate": 4.0026315789473685e-05,
      "loss": 0.4283,
      "step": 591
    },
    {
      "epoch": 0.8409090909090909,
      "grad_norm": 0.42932176674452094,
      "learning_rate": 4e-05,
      "loss": 0.4237,
      "step": 592
    },
    {
      "epoch": 0.8423295454545454,
      "grad_norm": 0.3218635964707305,
      "learning_rate": 3.997368421052632e-05,
      "loss": 0.423,
      "step": 593
    },
    {
      "epoch": 0.84375,
      "grad_norm": 0.4341555930367751,
      "learning_rate": 3.994736842105263e-05,
      "loss": 0.4402,
      "step": 594
    },
    {
      "epoch": 0.8451704545454546,
      "grad_norm": 0.3501284373913336,
      "learning_rate": 3.9921052631578945e-05,
      "loss": 0.4484,
      "step": 595
    },
    {
      "epoch": 0.8465909090909091,
      "grad_norm": 0.42018143968186805,
      "learning_rate": 3.989473684210526e-05,
      "loss": 0.4344,
      "step": 596
    },
    {
      "epoch": 0.8480113636363636,
      "grad_norm": 0.4217676945865392,
      "learning_rate": 3.986842105263158e-05,
      "loss": 0.4456,
      "step": 597
    },
    {
      "epoch": 0.8494318181818182,
      "grad_norm": 0.3894775348520525,
      "learning_rate": 3.98421052631579e-05,
      "loss": 0.4369,
      "step": 598
    },
    {
      "epoch": 0.8508522727272727,
      "grad_norm": 0.4244081859225972,
      "learning_rate": 3.981578947368421e-05,
      "loss": 0.4234,
      "step": 599
    },
    {
      "epoch": 0.8522727272727273,
      "grad_norm": 0.34412467651209067,
      "learning_rate": 3.978947368421053e-05,
      "loss": 0.4288,
      "step": 600
    },
    {
      "epoch": 0.8536931818181818,
      "grad_norm": 0.3550938702762472,
      "learning_rate": 3.976315789473685e-05,
      "loss": 0.4336,
      "step": 601
    },
    {
      "epoch": 0.8551136363636364,
      "grad_norm": 0.3620991314798494,
      "learning_rate": 3.973684210526316e-05,
      "loss": 0.4194,
      "step": 602
    },
    {
      "epoch": 0.8565340909090909,
      "grad_norm": 0.34538271443402085,
      "learning_rate": 3.971052631578948e-05,
      "loss": 0.4385,
      "step": 603
    },
    {
      "epoch": 0.8579545454545454,
      "grad_norm": 0.3606019172538275,
      "learning_rate": 3.968421052631579e-05,
      "loss": 0.452,
      "step": 604
    },
    {
      "epoch": 0.859375,
      "grad_norm": 0.33776006085329724,
      "learning_rate": 3.965789473684211e-05,
      "loss": 0.4478,
      "step": 605
    },
    {
      "epoch": 0.8607954545454546,
      "grad_norm": 0.31991288825634856,
      "learning_rate": 3.9631578947368426e-05,
      "loss": 0.4289,
      "step": 606
    },
    {
      "epoch": 0.8622159090909091,
      "grad_norm": 0.322782096676271,
      "learning_rate": 3.960526315789474e-05,
      "loss": 0.4287,
      "step": 607
    },
    {
      "epoch": 0.8636363636363636,
      "grad_norm": 0.4093386639087924,
      "learning_rate": 3.9578947368421056e-05,
      "loss": 0.4487,
      "step": 608
    },
    {
      "epoch": 0.8650568181818182,
      "grad_norm": 0.3699621473279953,
      "learning_rate": 3.955263157894737e-05,
      "loss": 0.4518,
      "step": 609
    },
    {
      "epoch": 0.8664772727272727,
      "grad_norm": 0.435129068825064,
      "learning_rate": 3.9526315789473686e-05,
      "loss": 0.4533,
      "step": 610
    },
    {
      "epoch": 0.8678977272727273,
      "grad_norm": 0.37207056969317975,
      "learning_rate": 3.9500000000000005e-05,
      "loss": 0.4184,
      "step": 611
    },
    {
      "epoch": 0.8693181818181818,
      "grad_norm": 0.3968201639349512,
      "learning_rate": 3.9473684210526316e-05,
      "loss": 0.4289,
      "step": 612
    },
    {
      "epoch": 0.8707386363636364,
      "grad_norm": 0.3125269871884003,
      "learning_rate": 3.9447368421052635e-05,
      "loss": 0.4232,
      "step": 613
    },
    {
      "epoch": 0.8721590909090909,
      "grad_norm": 0.36709664214530136,
      "learning_rate": 3.9421052631578946e-05,
      "loss": 0.4494,
      "step": 614
    },
    {
      "epoch": 0.8735795454545454,
      "grad_norm": 0.44891233561241894,
      "learning_rate": 3.9394736842105265e-05,
      "loss": 0.4611,
      "step": 615
    },
    {
      "epoch": 0.875,
      "grad_norm": 0.33965242298697523,
      "learning_rate": 3.936842105263158e-05,
      "loss": 0.4215,
      "step": 616
    },
    {
      "epoch": 0.8764204545454546,
      "grad_norm": 0.42525913452086933,
      "learning_rate": 3.9342105263157895e-05,
      "loss": 0.4409,
      "step": 617
    },
    {
      "epoch": 0.8778409090909091,
      "grad_norm": 0.3455756967353047,
      "learning_rate": 3.931578947368421e-05,
      "loss": 0.4296,
      "step": 618
    },
    {
      "epoch": 0.8792613636363636,
      "grad_norm": 0.3726311885613981,
      "learning_rate": 3.9289473684210524e-05,
      "loss": 0.463,
      "step": 619
    },
    {
      "epoch": 0.8806818181818182,
      "grad_norm": 0.413286491575025,
      "learning_rate": 3.926315789473684e-05,
      "loss": 0.4331,
      "step": 620
    },
    {
      "epoch": 0.8821022727272727,
      "grad_norm": 0.35109601314421185,
      "learning_rate": 3.923684210526316e-05,
      "loss": 0.4226,
      "step": 621
    },
    {
      "epoch": 0.8835227272727273,
      "grad_norm": 0.4492256571020852,
      "learning_rate": 3.921052631578947e-05,
      "loss": 0.4352,
      "step": 622
    },
    {
      "epoch": 0.8849431818181818,
      "grad_norm": 0.30857451597352903,
      "learning_rate": 3.918421052631579e-05,
      "loss": 0.4306,
      "step": 623
    },
    {
      "epoch": 0.8863636363636364,
      "grad_norm": 0.5126855390416218,
      "learning_rate": 3.91578947368421e-05,
      "loss": 0.4528,
      "step": 624
    },
    {
      "epoch": 0.8877840909090909,
      "grad_norm": 0.30889628993540863,
      "learning_rate": 3.913157894736843e-05,
      "loss": 0.3964,
      "step": 625
    },
    {
      "epoch": 0.8892045454545454,
      "grad_norm": 0.4698713820709422,
      "learning_rate": 3.910526315789474e-05,
      "loss": 0.4385,
      "step": 626
    },
    {
      "epoch": 0.890625,
      "grad_norm": 0.3817772832177049,
      "learning_rate": 3.907894736842105e-05,
      "loss": 0.4252,
      "step": 627
    },
    {
      "epoch": 0.8920454545454546,
      "grad_norm": 0.3724428710231016,
      "learning_rate": 3.905263157894737e-05,
      "loss": 0.4325,
      "step": 628
    },
    {
      "epoch": 0.8934659090909091,
      "grad_norm": 0.3793213557513137,
      "learning_rate": 3.902631578947369e-05,
      "loss": 0.4507,
      "step": 629
    },
    {
      "epoch": 0.8948863636363636,
      "grad_norm": 0.4769106683477844,
      "learning_rate": 3.9000000000000006e-05,
      "loss": 0.4382,
      "step": 630
    },
    {
      "epoch": 0.8963068181818182,
      "grad_norm": 0.5116969438168982,
      "learning_rate": 3.897368421052632e-05,
      "loss": 0.4396,
      "step": 631
    },
    {
      "epoch": 0.8977272727272727,
      "grad_norm": 0.43673574169087176,
      "learning_rate": 3.894736842105263e-05,
      "loss": 0.4619,
      "step": 632
    },
    {
      "epoch": 0.8991477272727273,
      "grad_norm": 0.5202998044986062,
      "learning_rate": 3.892105263157895e-05,
      "loss": 0.4532,
      "step": 633
    },
    {
      "epoch": 0.9005681818181818,
      "grad_norm": 0.3631028298113453,
      "learning_rate": 3.8894736842105266e-05,
      "loss": 0.4377,
      "step": 634
    },
    {
      "epoch": 0.9019886363636364,
      "grad_norm": 0.4760913953757132,
      "learning_rate": 3.8868421052631584e-05,
      "loss": 0.4339,
      "step": 635
    },
    {
      "epoch": 0.9034090909090909,
      "grad_norm": 0.33709412546065526,
      "learning_rate": 3.8842105263157896e-05,
      "loss": 0.4246,
      "step": 636
    },
    {
      "epoch": 0.9048295454545454,
      "grad_norm": 0.43430638252633336,
      "learning_rate": 3.8815789473684214e-05,
      "loss": 0.4403,
      "step": 637
    },
    {
      "epoch": 0.90625,
      "grad_norm": 0.3979424478292365,
      "learning_rate": 3.8789473684210526e-05,
      "loss": 0.4253,
      "step": 638
    },
    {
      "epoch": 0.9076704545454546,
      "grad_norm": 0.33423106891280757,
      "learning_rate": 3.8763157894736844e-05,
      "loss": 0.4273,
      "step": 639
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 0.4958643900387246,
      "learning_rate": 3.873684210526316e-05,
      "loss": 0.4439,
      "step": 640
    },
    {
      "epoch": 0.9105113636363636,
      "grad_norm": 0.36832554933601597,
      "learning_rate": 3.8710526315789474e-05,
      "loss": 0.4496,
      "step": 641
    },
    {
      "epoch": 0.9119318181818182,
      "grad_norm": 0.4115210677151824,
      "learning_rate": 3.868421052631579e-05,
      "loss": 0.4353,
      "step": 642
    },
    {
      "epoch": 0.9133522727272727,
      "grad_norm": 0.309810297635015,
      "learning_rate": 3.865789473684211e-05,
      "loss": 0.4261,
      "step": 643
    },
    {
      "epoch": 0.9147727272727273,
      "grad_norm": 0.3375990444634134,
      "learning_rate": 3.863157894736842e-05,
      "loss": 0.4304,
      "step": 644
    },
    {
      "epoch": 0.9161931818181818,
      "grad_norm": 0.3948236005479118,
      "learning_rate": 3.860526315789474e-05,
      "loss": 0.4369,
      "step": 645
    },
    {
      "epoch": 0.9176136363636364,
      "grad_norm": 0.3427128832961127,
      "learning_rate": 3.857894736842105e-05,
      "loss": 0.4569,
      "step": 646
    },
    {
      "epoch": 0.9190340909090909,
      "grad_norm": 0.3639220316080414,
      "learning_rate": 3.855263157894737e-05,
      "loss": 0.4177,
      "step": 647
    },
    {
      "epoch": 0.9204545454545454,
      "grad_norm": 0.40310505376608274,
      "learning_rate": 3.852631578947369e-05,
      "loss": 0.4246,
      "step": 648
    },
    {
      "epoch": 0.921875,
      "grad_norm": 0.35951115791243904,
      "learning_rate": 3.85e-05,
      "loss": 0.4402,
      "step": 649
    },
    {
      "epoch": 0.9232954545454546,
      "grad_norm": 0.4333781411291178,
      "learning_rate": 3.847368421052632e-05,
      "loss": 0.4378,
      "step": 650
    },
    {
      "epoch": 0.9247159090909091,
      "grad_norm": 0.3672319667330691,
      "learning_rate": 3.844736842105263e-05,
      "loss": 0.4535,
      "step": 651
    },
    {
      "epoch": 0.9261363636363636,
      "grad_norm": 0.43035852601252866,
      "learning_rate": 3.842105263157895e-05,
      "loss": 0.4327,
      "step": 652
    },
    {
      "epoch": 0.9275568181818182,
      "grad_norm": 0.3832208567843732,
      "learning_rate": 3.839473684210527e-05,
      "loss": 0.4385,
      "step": 653
    },
    {
      "epoch": 0.9289772727272727,
      "grad_norm": 0.3944129310920762,
      "learning_rate": 3.836842105263158e-05,
      "loss": 0.4438,
      "step": 654
    },
    {
      "epoch": 0.9303977272727273,
      "grad_norm": 0.36960477545592,
      "learning_rate": 3.83421052631579e-05,
      "loss": 0.4268,
      "step": 655
    },
    {
      "epoch": 0.9318181818181818,
      "grad_norm": 0.36442849045510145,
      "learning_rate": 3.831578947368421e-05,
      "loss": 0.4327,
      "step": 656
    },
    {
      "epoch": 0.9332386363636364,
      "grad_norm": 0.37442579367249584,
      "learning_rate": 3.828947368421053e-05,
      "loss": 0.4149,
      "step": 657
    },
    {
      "epoch": 0.9346590909090909,
      "grad_norm": 0.46774586234341203,
      "learning_rate": 3.8263157894736846e-05,
      "loss": 0.4252,
      "step": 658
    },
    {
      "epoch": 0.9360795454545454,
      "grad_norm": 0.3596167340160486,
      "learning_rate": 3.823684210526316e-05,
      "loss": 0.4413,
      "step": 659
    },
    {
      "epoch": 0.9375,
      "grad_norm": 0.38830185043016485,
      "learning_rate": 3.8210526315789476e-05,
      "loss": 0.4311,
      "step": 660
    },
    {
      "epoch": 0.9389204545454546,
      "grad_norm": 0.36290041000006296,
      "learning_rate": 3.818421052631579e-05,
      "loss": 0.437,
      "step": 661
    },
    {
      "epoch": 0.9403409090909091,
      "grad_norm": 0.3218174569639693,
      "learning_rate": 3.815789473684211e-05,
      "loss": 0.4265,
      "step": 662
    },
    {
      "epoch": 0.9417613636363636,
      "grad_norm": 0.37574508030810944,
      "learning_rate": 3.8131578947368424e-05,
      "loss": 0.4305,
      "step": 663
    },
    {
      "epoch": 0.9431818181818182,
      "grad_norm": 0.34942251263774204,
      "learning_rate": 3.8105263157894735e-05,
      "loss": 0.4297,
      "step": 664
    },
    {
      "epoch": 0.9446022727272727,
      "grad_norm": 0.3376041419082321,
      "learning_rate": 3.8078947368421054e-05,
      "loss": 0.447,
      "step": 665
    },
    {
      "epoch": 0.9460227272727273,
      "grad_norm": 0.374805699621099,
      "learning_rate": 3.8052631578947365e-05,
      "loss": 0.4235,
      "step": 666
    },
    {
      "epoch": 0.9474431818181818,
      "grad_norm": 0.3571862347256973,
      "learning_rate": 3.802631578947369e-05,
      "loss": 0.4208,
      "step": 667
    },
    {
      "epoch": 0.9488636363636364,
      "grad_norm": 0.4061597128233087,
      "learning_rate": 3.8e-05,
      "loss": 0.42,
      "step": 668
    },
    {
      "epoch": 0.9502840909090909,
      "grad_norm": 0.352539795488995,
      "learning_rate": 3.7973684210526314e-05,
      "loss": 0.409,
      "step": 669
    },
    {
      "epoch": 0.9517045454545454,
      "grad_norm": 0.4816188529387304,
      "learning_rate": 3.794736842105263e-05,
      "loss": 0.4481,
      "step": 670
    },
    {
      "epoch": 0.953125,
      "grad_norm": 0.4284774198611316,
      "learning_rate": 3.792105263157895e-05,
      "loss": 0.4362,
      "step": 671
    },
    {
      "epoch": 0.9545454545454546,
      "grad_norm": 0.382867644108465,
      "learning_rate": 3.789473684210527e-05,
      "loss": 0.4404,
      "step": 672
    },
    {
      "epoch": 0.9559659090909091,
      "grad_norm": 0.321093517034172,
      "learning_rate": 3.786842105263158e-05,
      "loss": 0.4358,
      "step": 673
    },
    {
      "epoch": 0.9573863636363636,
      "grad_norm": 0.3777574135016766,
      "learning_rate": 3.78421052631579e-05,
      "loss": 0.4496,
      "step": 674
    },
    {
      "epoch": 0.9588068181818182,
      "grad_norm": 0.32246698749133135,
      "learning_rate": 3.781578947368421e-05,
      "loss": 0.4316,
      "step": 675
    },
    {
      "epoch": 0.9602272727272727,
      "grad_norm": 0.36169705987618694,
      "learning_rate": 3.778947368421053e-05,
      "loss": 0.4256,
      "step": 676
    },
    {
      "epoch": 0.9616477272727273,
      "grad_norm": 0.3602107991213062,
      "learning_rate": 3.776315789473685e-05,
      "loss": 0.4257,
      "step": 677
    },
    {
      "epoch": 0.9630681818181818,
      "grad_norm": 0.3255175516748745,
      "learning_rate": 3.773684210526316e-05,
      "loss": 0.424,
      "step": 678
    },
    {
      "epoch": 0.9644886363636364,
      "grad_norm": 0.33262296868421265,
      "learning_rate": 3.771052631578948e-05,
      "loss": 0.4228,
      "step": 679
    },
    {
      "epoch": 0.9659090909090909,
      "grad_norm": 0.3191157950116114,
      "learning_rate": 3.768421052631579e-05,
      "loss": 0.4106,
      "step": 680
    },
    {
      "epoch": 0.9673295454545454,
      "grad_norm": 0.38483112999652846,
      "learning_rate": 3.765789473684211e-05,
      "loss": 0.421,
      "step": 681
    },
    {
      "epoch": 0.96875,
      "grad_norm": 0.2989255797797761,
      "learning_rate": 3.7631578947368425e-05,
      "loss": 0.4369,
      "step": 682
    },
    {
      "epoch": 0.9701704545454546,
      "grad_norm": 0.3433047990221525,
      "learning_rate": 3.760526315789474e-05,
      "loss": 0.4331,
      "step": 683
    },
    {
      "epoch": 0.9715909090909091,
      "grad_norm": 0.33242695633921343,
      "learning_rate": 3.7578947368421055e-05,
      "loss": 0.4466,
      "step": 684
    },
    {
      "epoch": 0.9730113636363636,
      "grad_norm": 0.3156342757385634,
      "learning_rate": 3.7552631578947374e-05,
      "loss": 0.4117,
      "step": 685
    },
    {
      "epoch": 0.9744318181818182,
      "grad_norm": 0.3257519752483392,
      "learning_rate": 3.7526315789473685e-05,
      "loss": 0.4446,
      "step": 686
    },
    {
      "epoch": 0.9758522727272727,
      "grad_norm": 0.3540905421241239,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.4183,
      "step": 687
    },
    {
      "epoch": 0.9772727272727273,
      "grad_norm": 0.34956183213929215,
      "learning_rate": 3.7473684210526315e-05,
      "loss": 0.4344,
      "step": 688
    },
    {
      "epoch": 0.9786931818181818,
      "grad_norm": 0.32819224274539127,
      "learning_rate": 3.7447368421052633e-05,
      "loss": 0.4236,
      "step": 689
    },
    {
      "epoch": 0.9801136363636364,
      "grad_norm": 0.3381134786814683,
      "learning_rate": 3.742105263157895e-05,
      "loss": 0.4273,
      "step": 690
    },
    {
      "epoch": 0.9815340909090909,
      "grad_norm": 0.3062213890567292,
      "learning_rate": 3.739473684210526e-05,
      "loss": 0.4538,
      "step": 691
    },
    {
      "epoch": 0.9829545454545454,
      "grad_norm": 0.3704115261893252,
      "learning_rate": 3.736842105263158e-05,
      "loss": 0.4156,
      "step": 692
    },
    {
      "epoch": 0.984375,
      "grad_norm": 0.31266568030397857,
      "learning_rate": 3.734210526315789e-05,
      "loss": 0.4306,
      "step": 693
    },
    {
      "epoch": 0.9857954545454546,
      "grad_norm": 0.3836440121564301,
      "learning_rate": 3.731578947368422e-05,
      "loss": 0.4505,
      "step": 694
    },
    {
      "epoch": 0.9872159090909091,
      "grad_norm": 0.28933110545060914,
      "learning_rate": 3.728947368421053e-05,
      "loss": 0.4232,
      "step": 695
    },
    {
      "epoch": 0.9886363636363636,
      "grad_norm": 0.37239845917588665,
      "learning_rate": 3.726315789473684e-05,
      "loss": 0.4481,
      "step": 696
    },
    {
      "epoch": 0.9900568181818182,
      "grad_norm": 0.3449358706113296,
      "learning_rate": 3.723684210526316e-05,
      "loss": 0.4186,
      "step": 697
    },
    {
      "epoch": 0.9914772727272727,
      "grad_norm": 0.34510278141465145,
      "learning_rate": 3.721052631578947e-05,
      "loss": 0.4319,
      "step": 698
    },
    {
      "epoch": 0.9928977272727273,
      "grad_norm": 0.3831492555929749,
      "learning_rate": 3.71842105263158e-05,
      "loss": 0.4395,
      "step": 699
    },
    {
      "epoch": 0.9943181818181818,
      "grad_norm": 0.35924072524590356,
      "learning_rate": 3.715789473684211e-05,
      "loss": 0.4412,
      "step": 700
    },
    {
      "epoch": 0.9957386363636364,
      "grad_norm": 0.4218734255108988,
      "learning_rate": 3.713157894736842e-05,
      "loss": 0.4206,
      "step": 701
    },
    {
      "epoch": 0.9971590909090909,
      "grad_norm": 0.3562500393795375,
      "learning_rate": 3.710526315789474e-05,
      "loss": 0.4356,
      "step": 702
    },
    {
      "epoch": 0.9985795454545454,
      "grad_norm": 0.42560782328139274,
      "learning_rate": 3.707894736842105e-05,
      "loss": 0.4161,
      "step": 703
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3779656206485301,
      "learning_rate": 3.7052631578947375e-05,
      "loss": 0.4141,
      "step": 704
    },
    {
      "epoch": 1.0014204545454546,
      "grad_norm": 0.4458909994591402,
      "learning_rate": 3.7026315789473687e-05,
      "loss": 0.3833,
      "step": 705
    },
    {
      "epoch": 1.0028409090909092,
      "grad_norm": 0.31392434753075454,
      "learning_rate": 3.7e-05,
      "loss": 0.3797,
      "step": 706
    },
    {
      "epoch": 1.0042613636363635,
      "grad_norm": 0.4036257745521145,
      "learning_rate": 3.6973684210526316e-05,
      "loss": 0.3721,
      "step": 707
    },
    {
      "epoch": 1.0056818181818181,
      "grad_norm": 0.3852137405247355,
      "learning_rate": 3.6947368421052635e-05,
      "loss": 0.3581,
      "step": 708
    },
    {
      "epoch": 1.0071022727272727,
      "grad_norm": 0.40108242916322384,
      "learning_rate": 3.692105263157895e-05,
      "loss": 0.3541,
      "step": 709
    },
    {
      "epoch": 1.0085227272727273,
      "grad_norm": 0.4741145773568856,
      "learning_rate": 3.6894736842105265e-05,
      "loss": 0.3568,
      "step": 710
    },
    {
      "epoch": 1.0099431818181819,
      "grad_norm": 0.38664502033422604,
      "learning_rate": 3.686842105263158e-05,
      "loss": 0.3625,
      "step": 711
    },
    {
      "epoch": 1.0113636363636365,
      "grad_norm": 0.447081636950645,
      "learning_rate": 3.6842105263157895e-05,
      "loss": 0.3508,
      "step": 712
    },
    {
      "epoch": 1.0127840909090908,
      "grad_norm": 0.3371976243826069,
      "learning_rate": 3.681578947368421e-05,
      "loss": 0.3686,
      "step": 713
    },
    {
      "epoch": 1.0142045454545454,
      "grad_norm": 0.3938421113202131,
      "learning_rate": 3.678947368421053e-05,
      "loss": 0.3427,
      "step": 714
    },
    {
      "epoch": 1.015625,
      "grad_norm": 0.3835664300508239,
      "learning_rate": 3.676315789473684e-05,
      "loss": 0.3791,
      "step": 715
    },
    {
      "epoch": 1.0170454545454546,
      "grad_norm": 0.3137561980184083,
      "learning_rate": 3.673684210526316e-05,
      "loss": 0.3701,
      "step": 716
    },
    {
      "epoch": 1.0184659090909092,
      "grad_norm": 0.4112738066544594,
      "learning_rate": 3.671052631578947e-05,
      "loss": 0.367,
      "step": 717
    },
    {
      "epoch": 1.0198863636363635,
      "grad_norm": 0.30025150386745425,
      "learning_rate": 3.668421052631579e-05,
      "loss": 0.3516,
      "step": 718
    },
    {
      "epoch": 1.0213068181818181,
      "grad_norm": 0.39419895368419244,
      "learning_rate": 3.665789473684211e-05,
      "loss": 0.3748,
      "step": 719
    },
    {
      "epoch": 1.0227272727272727,
      "grad_norm": 0.327532198716856,
      "learning_rate": 3.663157894736842e-05,
      "loss": 0.3616,
      "step": 720
    },
    {
      "epoch": 1.0241477272727273,
      "grad_norm": 0.30128030847673426,
      "learning_rate": 3.660526315789474e-05,
      "loss": 0.3649,
      "step": 721
    },
    {
      "epoch": 1.0255681818181819,
      "grad_norm": 0.33808509069883763,
      "learning_rate": 3.657894736842106e-05,
      "loss": 0.3549,
      "step": 722
    },
    {
      "epoch": 1.0269886363636365,
      "grad_norm": 0.3412107266685095,
      "learning_rate": 3.655263157894737e-05,
      "loss": 0.3873,
      "step": 723
    },
    {
      "epoch": 1.0284090909090908,
      "grad_norm": 0.3764470636892884,
      "learning_rate": 3.652631578947369e-05,
      "loss": 0.3518,
      "step": 724
    },
    {
      "epoch": 1.0298295454545454,
      "grad_norm": 0.36420097198865475,
      "learning_rate": 3.65e-05,
      "loss": 0.3479,
      "step": 725
    },
    {
      "epoch": 1.03125,
      "grad_norm": 0.34165300001866605,
      "learning_rate": 3.647368421052632e-05,
      "loss": 0.365,
      "step": 726
    },
    {
      "epoch": 1.0326704545454546,
      "grad_norm": 0.3908754215739632,
      "learning_rate": 3.6447368421052636e-05,
      "loss": 0.3709,
      "step": 727
    },
    {
      "epoch": 1.0340909090909092,
      "grad_norm": 0.32961058356443734,
      "learning_rate": 3.642105263157895e-05,
      "loss": 0.3496,
      "step": 728
    },
    {
      "epoch": 1.0355113636363635,
      "grad_norm": 0.36216226944727475,
      "learning_rate": 3.6394736842105266e-05,
      "loss": 0.3589,
      "step": 729
    },
    {
      "epoch": 1.0369318181818181,
      "grad_norm": 0.3550094061375519,
      "learning_rate": 3.636842105263158e-05,
      "loss": 0.37,
      "step": 730
    },
    {
      "epoch": 1.0383522727272727,
      "grad_norm": 0.3809581159254581,
      "learning_rate": 3.6342105263157896e-05,
      "loss": 0.3503,
      "step": 731
    },
    {
      "epoch": 1.0397727272727273,
      "grad_norm": 0.3946749865860991,
      "learning_rate": 3.6315789473684214e-05,
      "loss": 0.3646,
      "step": 732
    },
    {
      "epoch": 1.0411931818181819,
      "grad_norm": 0.33716614101881887,
      "learning_rate": 3.6289473684210526e-05,
      "loss": 0.3439,
      "step": 733
    },
    {
      "epoch": 1.0426136363636365,
      "grad_norm": 0.30747029664219433,
      "learning_rate": 3.6263157894736844e-05,
      "loss": 0.3499,
      "step": 734
    },
    {
      "epoch": 1.0440340909090908,
      "grad_norm": 0.3212420233500325,
      "learning_rate": 3.6236842105263156e-05,
      "loss": 0.3508,
      "step": 735
    },
    {
      "epoch": 1.0454545454545454,
      "grad_norm": 0.2827643661482006,
      "learning_rate": 3.621052631578948e-05,
      "loss": 0.3836,
      "step": 736
    },
    {
      "epoch": 1.046875,
      "grad_norm": 0.3943172902919471,
      "learning_rate": 3.618421052631579e-05,
      "loss": 0.3809,
      "step": 737
    },
    {
      "epoch": 1.0482954545454546,
      "grad_norm": 0.39272291115768865,
      "learning_rate": 3.6157894736842104e-05,
      "loss": 0.3794,
      "step": 738
    },
    {
      "epoch": 1.0497159090909092,
      "grad_norm": 0.36972794286753335,
      "learning_rate": 3.613157894736842e-05,
      "loss": 0.3772,
      "step": 739
    },
    {
      "epoch": 1.0511363636363635,
      "grad_norm": 0.3306038024732215,
      "learning_rate": 3.6105263157894734e-05,
      "loss": 0.3523,
      "step": 740
    },
    {
      "epoch": 1.0525568181818181,
      "grad_norm": 0.3540962645758875,
      "learning_rate": 3.607894736842106e-05,
      "loss": 0.3564,
      "step": 741
    },
    {
      "epoch": 1.0539772727272727,
      "grad_norm": 0.337951405790224,
      "learning_rate": 3.605263157894737e-05,
      "loss": 0.3737,
      "step": 742
    },
    {
      "epoch": 1.0553977272727273,
      "grad_norm": 0.36339492942394885,
      "learning_rate": 3.602631578947368e-05,
      "loss": 0.3685,
      "step": 743
    },
    {
      "epoch": 1.0568181818181819,
      "grad_norm": 0.2987269776563186,
      "learning_rate": 3.6e-05,
      "loss": 0.3599,
      "step": 744
    },
    {
      "epoch": 1.0582386363636365,
      "grad_norm": 0.4356724917743531,
      "learning_rate": 3.597368421052631e-05,
      "loss": 0.3657,
      "step": 745
    },
    {
      "epoch": 1.0596590909090908,
      "grad_norm": 0.39560222474246914,
      "learning_rate": 3.594736842105264e-05,
      "loss": 0.3767,
      "step": 746
    },
    {
      "epoch": 1.0610795454545454,
      "grad_norm": 0.3715945768522967,
      "learning_rate": 3.592105263157895e-05,
      "loss": 0.3635,
      "step": 747
    },
    {
      "epoch": 1.0625,
      "grad_norm": 0.38116680311999157,
      "learning_rate": 3.589473684210527e-05,
      "loss": 0.3663,
      "step": 748
    },
    {
      "epoch": 1.0639204545454546,
      "grad_norm": 0.34338696831324644,
      "learning_rate": 3.586842105263158e-05,
      "loss": 0.3589,
      "step": 749
    },
    {
      "epoch": 1.0653409090909092,
      "grad_norm": 0.34012801215230853,
      "learning_rate": 3.58421052631579e-05,
      "loss": 0.3644,
      "step": 750
    },
    {
      "epoch": 1.0667613636363635,
      "grad_norm": 0.3679867921107257,
      "learning_rate": 3.5815789473684216e-05,
      "loss": 0.3808,
      "step": 751
    },
    {
      "epoch": 1.0681818181818181,
      "grad_norm": 0.394879866322695,
      "learning_rate": 3.578947368421053e-05,
      "loss": 0.3782,
      "step": 752
    },
    {
      "epoch": 1.0696022727272727,
      "grad_norm": 0.3516224863956126,
      "learning_rate": 3.5763157894736846e-05,
      "loss": 0.3678,
      "step": 753
    },
    {
      "epoch": 1.0710227272727273,
      "grad_norm": 0.41128351768838994,
      "learning_rate": 3.573684210526316e-05,
      "loss": 0.3743,
      "step": 754
    },
    {
      "epoch": 1.0724431818181819,
      "grad_norm": 0.31844413377260156,
      "learning_rate": 3.5710526315789476e-05,
      "loss": 0.3507,
      "step": 755
    },
    {
      "epoch": 1.0738636363636365,
      "grad_norm": 0.377684904031508,
      "learning_rate": 3.5684210526315794e-05,
      "loss": 0.3665,
      "step": 756
    },
    {
      "epoch": 1.0752840909090908,
      "grad_norm": 0.35459576970862866,
      "learning_rate": 3.5657894736842106e-05,
      "loss": 0.3484,
      "step": 757
    },
    {
      "epoch": 1.0767045454545454,
      "grad_norm": 0.3221703956583694,
      "learning_rate": 3.5631578947368424e-05,
      "loss": 0.3696,
      "step": 758
    },
    {
      "epoch": 1.078125,
      "grad_norm": 0.35202676237961283,
      "learning_rate": 3.5605263157894736e-05,
      "loss": 0.3664,
      "step": 759
    },
    {
      "epoch": 1.0795454545454546,
      "grad_norm": 0.33481821154232366,
      "learning_rate": 3.5578947368421054e-05,
      "loss": 0.3462,
      "step": 760
    },
    {
      "epoch": 1.0809659090909092,
      "grad_norm": 0.3202746524038486,
      "learning_rate": 3.555263157894737e-05,
      "loss": 0.3415,
      "step": 761
    },
    {
      "epoch": 1.0823863636363635,
      "grad_norm": 0.3542243286708907,
      "learning_rate": 3.5526315789473684e-05,
      "loss": 0.3806,
      "step": 762
    },
    {
      "epoch": 1.0838068181818181,
      "grad_norm": 0.32760743805477344,
      "learning_rate": 3.55e-05,
      "loss": 0.3542,
      "step": 763
    },
    {
      "epoch": 1.0852272727272727,
      "grad_norm": 0.36730350771587816,
      "learning_rate": 3.547368421052632e-05,
      "loss": 0.3818,
      "step": 764
    },
    {
      "epoch": 1.0866477272727273,
      "grad_norm": 0.3057548384159778,
      "learning_rate": 3.544736842105263e-05,
      "loss": 0.3443,
      "step": 765
    },
    {
      "epoch": 1.0880681818181819,
      "grad_norm": 0.37962538770057547,
      "learning_rate": 3.542105263157895e-05,
      "loss": 0.366,
      "step": 766
    },
    {
      "epoch": 1.0894886363636365,
      "grad_norm": 0.3282115339387615,
      "learning_rate": 3.539473684210526e-05,
      "loss": 0.3575,
      "step": 767
    },
    {
      "epoch": 1.0909090909090908,
      "grad_norm": 0.36231499446093757,
      "learning_rate": 3.536842105263158e-05,
      "loss": 0.3586,
      "step": 768
    },
    {
      "epoch": 1.0923295454545454,
      "grad_norm": 0.30188384560489157,
      "learning_rate": 3.53421052631579e-05,
      "loss": 0.3651,
      "step": 769
    },
    {
      "epoch": 1.09375,
      "grad_norm": 0.3997141732888362,
      "learning_rate": 3.531578947368421e-05,
      "loss": 0.3631,
      "step": 770
    },
    {
      "epoch": 1.0951704545454546,
      "grad_norm": 0.32254044284360267,
      "learning_rate": 3.528947368421053e-05,
      "loss": 0.348,
      "step": 771
    },
    {
      "epoch": 1.0965909090909092,
      "grad_norm": 0.3517002359558627,
      "learning_rate": 3.526315789473684e-05,
      "loss": 0.3387,
      "step": 772
    },
    {
      "epoch": 1.0980113636363635,
      "grad_norm": 0.3833470301188439,
      "learning_rate": 3.523684210526316e-05,
      "loss": 0.361,
      "step": 773
    },
    {
      "epoch": 1.0994318181818181,
      "grad_norm": 0.45310234522647563,
      "learning_rate": 3.521052631578948e-05,
      "loss": 0.3613,
      "step": 774
    },
    {
      "epoch": 1.1008522727272727,
      "grad_norm": 0.41388131329254935,
      "learning_rate": 3.518421052631579e-05,
      "loss": 0.3834,
      "step": 775
    },
    {
      "epoch": 1.1022727272727273,
      "grad_norm": 0.36138874488389616,
      "learning_rate": 3.515789473684211e-05,
      "loss": 0.3575,
      "step": 776
    },
    {
      "epoch": 1.1036931818181819,
      "grad_norm": 0.41056313880475936,
      "learning_rate": 3.513157894736842e-05,
      "loss": 0.3759,
      "step": 777
    },
    {
      "epoch": 1.1051136363636365,
      "grad_norm": 0.33151297751628755,
      "learning_rate": 3.5105263157894744e-05,
      "loss": 0.3747,
      "step": 778
    },
    {
      "epoch": 1.1065340909090908,
      "grad_norm": 0.4300669243742748,
      "learning_rate": 3.5078947368421055e-05,
      "loss": 0.371,
      "step": 779
    },
    {
      "epoch": 1.1079545454545454,
      "grad_norm": 0.35399004884161855,
      "learning_rate": 3.505263157894737e-05,
      "loss": 0.3657,
      "step": 780
    },
    {
      "epoch": 1.109375,
      "grad_norm": 0.3512321026460036,
      "learning_rate": 3.5026315789473685e-05,
      "loss": 0.358,
      "step": 781
    },
    {
      "epoch": 1.1107954545454546,
      "grad_norm": 0.3974661284942752,
      "learning_rate": 3.5e-05,
      "loss": 0.3647,
      "step": 782
    },
    {
      "epoch": 1.1122159090909092,
      "grad_norm": 0.3205520765889229,
      "learning_rate": 3.497368421052632e-05,
      "loss": 0.367,
      "step": 783
    },
    {
      "epoch": 1.1136363636363635,
      "grad_norm": 0.3609412336129032,
      "learning_rate": 3.4947368421052634e-05,
      "loss": 0.3568,
      "step": 784
    },
    {
      "epoch": 1.1150568181818181,
      "grad_norm": 0.4026428242118274,
      "learning_rate": 3.492105263157895e-05,
      "loss": 0.362,
      "step": 785
    },
    {
      "epoch": 1.1164772727272727,
      "grad_norm": 0.38805588540360547,
      "learning_rate": 3.4894736842105264e-05,
      "loss": 0.3495,
      "step": 786
    },
    {
      "epoch": 1.1178977272727273,
      "grad_norm": 0.39980805923358737,
      "learning_rate": 3.4868421052631575e-05,
      "loss": 0.3592,
      "step": 787
    },
    {
      "epoch": 1.1193181818181819,
      "grad_norm": 0.3044222584592094,
      "learning_rate": 3.48421052631579e-05,
      "loss": 0.3727,
      "step": 788
    },
    {
      "epoch": 1.1207386363636365,
      "grad_norm": 0.3962665668354898,
      "learning_rate": 3.481578947368421e-05,
      "loss": 0.3419,
      "step": 789
    },
    {
      "epoch": 1.1221590909090908,
      "grad_norm": 0.3436520302294371,
      "learning_rate": 3.478947368421053e-05,
      "loss": 0.3607,
      "step": 790
    },
    {
      "epoch": 1.1235795454545454,
      "grad_norm": 0.43282948217159795,
      "learning_rate": 3.476315789473684e-05,
      "loss": 0.3713,
      "step": 791
    },
    {
      "epoch": 1.125,
      "grad_norm": 0.28517188160639556,
      "learning_rate": 3.473684210526316e-05,
      "loss": 0.3672,
      "step": 792
    },
    {
      "epoch": 1.1264204545454546,
      "grad_norm": 0.4022729228806461,
      "learning_rate": 3.471052631578948e-05,
      "loss": 0.3711,
      "step": 793
    },
    {
      "epoch": 1.1278409090909092,
      "grad_norm": 0.2857639990802249,
      "learning_rate": 3.468421052631579e-05,
      "loss": 0.3869,
      "step": 794
    },
    {
      "epoch": 1.1292613636363638,
      "grad_norm": 0.41288830220348144,
      "learning_rate": 3.465789473684211e-05,
      "loss": 0.3531,
      "step": 795
    },
    {
      "epoch": 1.1306818181818181,
      "grad_norm": 0.28915922672988736,
      "learning_rate": 3.463157894736842e-05,
      "loss": 0.3609,
      "step": 796
    },
    {
      "epoch": 1.1321022727272727,
      "grad_norm": 0.3372442583329665,
      "learning_rate": 3.460526315789474e-05,
      "loss": 0.3756,
      "step": 797
    },
    {
      "epoch": 1.1335227272727273,
      "grad_norm": 0.3789999418443824,
      "learning_rate": 3.457894736842106e-05,
      "loss": 0.3503,
      "step": 798
    },
    {
      "epoch": 1.1349431818181819,
      "grad_norm": 0.3164228845267267,
      "learning_rate": 3.455263157894737e-05,
      "loss": 0.3682,
      "step": 799
    },
    {
      "epoch": 1.1363636363636362,
      "grad_norm": 0.37266199024694824,
      "learning_rate": 3.452631578947369e-05,
      "loss": 0.3549,
      "step": 800
    },
    {
      "epoch": 1.1377840909090908,
      "grad_norm": 0.3306048000451069,
      "learning_rate": 3.45e-05,
      "loss": 0.3786,
      "step": 801
    },
    {
      "epoch": 1.1392045454545454,
      "grad_norm": 0.280629150679864,
      "learning_rate": 3.447368421052632e-05,
      "loss": 0.3694,
      "step": 802
    },
    {
      "epoch": 1.140625,
      "grad_norm": 0.3343078780307861,
      "learning_rate": 3.4447368421052635e-05,
      "loss": 0.3508,
      "step": 803
    },
    {
      "epoch": 1.1420454545454546,
      "grad_norm": 0.2959100617696805,
      "learning_rate": 3.442105263157895e-05,
      "loss": 0.3722,
      "step": 804
    },
    {
      "epoch": 1.1434659090909092,
      "grad_norm": 0.3394570544437537,
      "learning_rate": 3.4394736842105265e-05,
      "loss": 0.3743,
      "step": 805
    },
    {
      "epoch": 1.1448863636363638,
      "grad_norm": 0.3460899113890242,
      "learning_rate": 3.436842105263158e-05,
      "loss": 0.3402,
      "step": 806
    },
    {
      "epoch": 1.1463068181818181,
      "grad_norm": 0.31805160777060304,
      "learning_rate": 3.4342105263157895e-05,
      "loss": 0.3729,
      "step": 807
    },
    {
      "epoch": 1.1477272727272727,
      "grad_norm": 0.4018609059408068,
      "learning_rate": 3.431578947368421e-05,
      "loss": 0.3844,
      "step": 808
    },
    {
      "epoch": 1.1491477272727273,
      "grad_norm": 0.3221363491247782,
      "learning_rate": 3.4289473684210525e-05,
      "loss": 0.3824,
      "step": 809
    },
    {
      "epoch": 1.1505681818181819,
      "grad_norm": 0.3256592979969445,
      "learning_rate": 3.426315789473684e-05,
      "loss": 0.3349,
      "step": 810
    },
    {
      "epoch": 1.1519886363636362,
      "grad_norm": 0.35651415980035495,
      "learning_rate": 3.423684210526316e-05,
      "loss": 0.3703,
      "step": 811
    },
    {
      "epoch": 1.1534090909090908,
      "grad_norm": 0.3261760909226177,
      "learning_rate": 3.421052631578947e-05,
      "loss": 0.3809,
      "step": 812
    },
    {
      "epoch": 1.1548295454545454,
      "grad_norm": 0.37437802100089124,
      "learning_rate": 3.418421052631579e-05,
      "loss": 0.3813,
      "step": 813
    },
    {
      "epoch": 1.15625,
      "grad_norm": 0.3560274945782148,
      "learning_rate": 3.41578947368421e-05,
      "loss": 0.3685,
      "step": 814
    },
    {
      "epoch": 1.1576704545454546,
      "grad_norm": 0.31636312350075374,
      "learning_rate": 3.413157894736842e-05,
      "loss": 0.3724,
      "step": 815
    },
    {
      "epoch": 1.1590909090909092,
      "grad_norm": 0.3302480042205875,
      "learning_rate": 3.410526315789474e-05,
      "loss": 0.3533,
      "step": 816
    },
    {
      "epoch": 1.1605113636363638,
      "grad_norm": 0.32782293469468904,
      "learning_rate": 3.407894736842105e-05,
      "loss": 0.3635,
      "step": 817
    },
    {
      "epoch": 1.1619318181818181,
      "grad_norm": 0.32158684183412484,
      "learning_rate": 3.405263157894737e-05,
      "loss": 0.3944,
      "step": 818
    },
    {
      "epoch": 1.1633522727272727,
      "grad_norm": 0.40924380470680494,
      "learning_rate": 3.402631578947368e-05,
      "loss": 0.3897,
      "step": 819
    },
    {
      "epoch": 1.1647727272727273,
      "grad_norm": 0.35750435044408885,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 0.371,
      "step": 820
    },
    {
      "epoch": 1.1661931818181819,
      "grad_norm": 0.359856080600231,
      "learning_rate": 3.397368421052632e-05,
      "loss": 0.378,
      "step": 821
    },
    {
      "epoch": 1.1676136363636362,
      "grad_norm": 0.423220909670751,
      "learning_rate": 3.3947368421052636e-05,
      "loss": 0.346,
      "step": 822
    },
    {
      "epoch": 1.1690340909090908,
      "grad_norm": 0.4363145495555479,
      "learning_rate": 3.392105263157895e-05,
      "loss": 0.3684,
      "step": 823
    },
    {
      "epoch": 1.1704545454545454,
      "grad_norm": 0.3623954376799174,
      "learning_rate": 3.389473684210526e-05,
      "loss": 0.3749,
      "step": 824
    },
    {
      "epoch": 1.171875,
      "grad_norm": 0.40205933638406194,
      "learning_rate": 3.3868421052631585e-05,
      "loss": 0.3596,
      "step": 825
    },
    {
      "epoch": 1.1732954545454546,
      "grad_norm": 0.4119610726919289,
      "learning_rate": 3.3842105263157896e-05,
      "loss": 0.3667,
      "step": 826
    },
    {
      "epoch": 1.1747159090909092,
      "grad_norm": 0.35831491987165515,
      "learning_rate": 3.3815789473684215e-05,
      "loss": 0.3586,
      "step": 827
    },
    {
      "epoch": 1.1761363636363638,
      "grad_norm": 0.38800794932689525,
      "learning_rate": 3.3789473684210526e-05,
      "loss": 0.3825,
      "step": 828
    },
    {
      "epoch": 1.1775568181818181,
      "grad_norm": 0.39415401036714026,
      "learning_rate": 3.376315789473684e-05,
      "loss": 0.3757,
      "step": 829
    },
    {
      "epoch": 1.1789772727272727,
      "grad_norm": 0.4040237995011094,
      "learning_rate": 3.373684210526316e-05,
      "loss": 0.3579,
      "step": 830
    },
    {
      "epoch": 1.1803977272727273,
      "grad_norm": 0.36127579064352205,
      "learning_rate": 3.3710526315789475e-05,
      "loss": 0.3753,
      "step": 831
    },
    {
      "epoch": 1.1818181818181819,
      "grad_norm": 6.391144602828236,
      "learning_rate": 3.368421052631579e-05,
      "loss": 0.3665,
      "step": 832
    },
    {
      "epoch": 1.1832386363636362,
      "grad_norm": 0.46021828333886805,
      "learning_rate": 3.3657894736842105e-05,
      "loss": 0.3743,
      "step": 833
    },
    {
      "epoch": 1.1846590909090908,
      "grad_norm": 0.3467766221655247,
      "learning_rate": 3.363157894736842e-05,
      "loss": 0.3741,
      "step": 834
    },
    {
      "epoch": 1.1860795454545454,
      "grad_norm": 0.3647609145035779,
      "learning_rate": 3.360526315789474e-05,
      "loss": 0.3914,
      "step": 835
    },
    {
      "epoch": 1.1875,
      "grad_norm": 0.3692332624492452,
      "learning_rate": 3.357894736842105e-05,
      "loss": 0.3582,
      "step": 836
    },
    {
      "epoch": 1.1889204545454546,
      "grad_norm": 0.3271724515661657,
      "learning_rate": 3.355263157894737e-05,
      "loss": 0.3723,
      "step": 837
    },
    {
      "epoch": 1.1903409090909092,
      "grad_norm": 0.3648217407855925,
      "learning_rate": 3.352631578947368e-05,
      "loss": 0.3625,
      "step": 838
    },
    {
      "epoch": 1.1917613636363638,
      "grad_norm": 0.3308091401905853,
      "learning_rate": 3.35e-05,
      "loss": 0.3486,
      "step": 839
    },
    {
      "epoch": 1.1931818181818181,
      "grad_norm": 0.394522431327854,
      "learning_rate": 3.347368421052632e-05,
      "loss": 0.3833,
      "step": 840
    },
    {
      "epoch": 1.1946022727272727,
      "grad_norm": 0.357762474481986,
      "learning_rate": 3.344736842105263e-05,
      "loss": 0.3508,
      "step": 841
    },
    {
      "epoch": 1.1960227272727273,
      "grad_norm": 0.31413472196272446,
      "learning_rate": 3.342105263157895e-05,
      "loss": 0.3628,
      "step": 842
    },
    {
      "epoch": 1.1974431818181819,
      "grad_norm": 0.33866265672486956,
      "learning_rate": 3.339473684210526e-05,
      "loss": 0.3542,
      "step": 843
    },
    {
      "epoch": 1.1988636363636362,
      "grad_norm": 0.3114430229308768,
      "learning_rate": 3.336842105263158e-05,
      "loss": 0.3487,
      "step": 844
    },
    {
      "epoch": 1.2002840909090908,
      "grad_norm": 0.3317328701502765,
      "learning_rate": 3.33421052631579e-05,
      "loss": 0.3607,
      "step": 845
    },
    {
      "epoch": 1.2017045454545454,
      "grad_norm": 0.32688249833994304,
      "learning_rate": 3.331578947368421e-05,
      "loss": 0.3741,
      "step": 846
    },
    {
      "epoch": 1.203125,
      "grad_norm": 0.31966376682715897,
      "learning_rate": 3.328947368421053e-05,
      "loss": 0.3745,
      "step": 847
    },
    {
      "epoch": 1.2045454545454546,
      "grad_norm": 0.37012477603856775,
      "learning_rate": 3.3263157894736846e-05,
      "loss": 0.3835,
      "step": 848
    },
    {
      "epoch": 1.2059659090909092,
      "grad_norm": 0.3325691753854767,
      "learning_rate": 3.323684210526316e-05,
      "loss": 0.3544,
      "step": 849
    },
    {
      "epoch": 1.2073863636363638,
      "grad_norm": 0.322659962639263,
      "learning_rate": 3.3210526315789476e-05,
      "loss": 0.3432,
      "step": 850
    },
    {
      "epoch": 1.2088068181818181,
      "grad_norm": 0.35103950419107394,
      "learning_rate": 3.318421052631579e-05,
      "loss": 0.3654,
      "step": 851
    },
    {
      "epoch": 1.2102272727272727,
      "grad_norm": 0.39111636262860755,
      "learning_rate": 3.3157894736842106e-05,
      "loss": 0.3683,
      "step": 852
    },
    {
      "epoch": 1.2116477272727273,
      "grad_norm": 0.3336480692643348,
      "learning_rate": 3.3131578947368424e-05,
      "loss": 0.3674,
      "step": 853
    },
    {
      "epoch": 1.2130681818181819,
      "grad_norm": 0.30200485047094694,
      "learning_rate": 3.3105263157894736e-05,
      "loss": 0.3814,
      "step": 854
    },
    {
      "epoch": 1.2144886363636362,
      "grad_norm": 0.33494404558277574,
      "learning_rate": 3.3078947368421054e-05,
      "loss": 0.3579,
      "step": 855
    },
    {
      "epoch": 1.2159090909090908,
      "grad_norm": 0.36723909765935137,
      "learning_rate": 3.3052631578947366e-05,
      "loss": 0.3715,
      "step": 856
    },
    {
      "epoch": 1.2173295454545454,
      "grad_norm": 0.3140415467777849,
      "learning_rate": 3.302631578947369e-05,
      "loss": 0.3924,
      "step": 857
    },
    {
      "epoch": 1.21875,
      "grad_norm": 0.3394616706366431,
      "learning_rate": 3.3e-05,
      "loss": 0.377,
      "step": 858
    },
    {
      "epoch": 1.2201704545454546,
      "grad_norm": 0.39766506325027084,
      "learning_rate": 3.297368421052632e-05,
      "loss": 0.3746,
      "step": 859
    },
    {
      "epoch": 1.2215909090909092,
      "grad_norm": 0.326909853171842,
      "learning_rate": 3.294736842105263e-05,
      "loss": 0.3587,
      "step": 860
    },
    {
      "epoch": 1.2230113636363638,
      "grad_norm": 0.30623392684672673,
      "learning_rate": 3.2921052631578944e-05,
      "loss": 0.3743,
      "step": 861
    },
    {
      "epoch": 1.2244318181818181,
      "grad_norm": 0.3358068711134017,
      "learning_rate": 3.289473684210527e-05,
      "loss": 0.377,
      "step": 862
    },
    {
      "epoch": 1.2258522727272727,
      "grad_norm": 0.3379719181148699,
      "learning_rate": 3.286842105263158e-05,
      "loss": 0.354,
      "step": 863
    },
    {
      "epoch": 1.2272727272727273,
      "grad_norm": 0.293017086586076,
      "learning_rate": 3.28421052631579e-05,
      "loss": 0.3577,
      "step": 864
    },
    {
      "epoch": 1.2286931818181819,
      "grad_norm": 0.30730942066892475,
      "learning_rate": 3.281578947368421e-05,
      "loss": 0.3467,
      "step": 865
    },
    {
      "epoch": 1.2301136363636362,
      "grad_norm": 0.3156094581273121,
      "learning_rate": 3.278947368421052e-05,
      "loss": 0.375,
      "step": 866
    },
    {
      "epoch": 1.2315340909090908,
      "grad_norm": 0.31017762919573844,
      "learning_rate": 3.276315789473685e-05,
      "loss": 0.3651,
      "step": 867
    },
    {
      "epoch": 1.2329545454545454,
      "grad_norm": 0.3228274216667177,
      "learning_rate": 3.273684210526316e-05,
      "loss": 0.3753,
      "step": 868
    },
    {
      "epoch": 1.234375,
      "grad_norm": 0.28758078830144806,
      "learning_rate": 3.271052631578948e-05,
      "loss": 0.3841,
      "step": 869
    },
    {
      "epoch": 1.2357954545454546,
      "grad_norm": 0.2912512178596405,
      "learning_rate": 3.268421052631579e-05,
      "loss": 0.373,
      "step": 870
    },
    {
      "epoch": 1.2372159090909092,
      "grad_norm": 0.29284416696571985,
      "learning_rate": 3.265789473684211e-05,
      "loss": 0.3772,
      "step": 871
    },
    {
      "epoch": 1.2386363636363638,
      "grad_norm": 0.3163570633025465,
      "learning_rate": 3.2631578947368426e-05,
      "loss": 0.3807,
      "step": 872
    },
    {
      "epoch": 1.2400568181818181,
      "grad_norm": 0.30001593603745325,
      "learning_rate": 3.260526315789474e-05,
      "loss": 0.3882,
      "step": 873
    },
    {
      "epoch": 1.2414772727272727,
      "grad_norm": 0.3167435091385291,
      "learning_rate": 3.2578947368421056e-05,
      "loss": 0.3883,
      "step": 874
    },
    {
      "epoch": 1.2428977272727273,
      "grad_norm": 0.3268152833712173,
      "learning_rate": 3.255263157894737e-05,
      "loss": 0.3591,
      "step": 875
    },
    {
      "epoch": 1.2443181818181819,
      "grad_norm": 0.282866252376745,
      "learning_rate": 3.2526315789473686e-05,
      "loss": 0.3709,
      "step": 876
    },
    {
      "epoch": 1.2457386363636362,
      "grad_norm": 0.36608368232144756,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 0.3733,
      "step": 877
    },
    {
      "epoch": 1.2471590909090908,
      "grad_norm": 0.30813750942883555,
      "learning_rate": 3.2473684210526316e-05,
      "loss": 0.3866,
      "step": 878
    },
    {
      "epoch": 1.2485795454545454,
      "grad_norm": 0.37557832292373233,
      "learning_rate": 3.2447368421052634e-05,
      "loss": 0.3963,
      "step": 879
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.3446565803181206,
      "learning_rate": 3.2421052631578945e-05,
      "loss": 0.3755,
      "step": 880
    },
    {
      "epoch": 1.2514204545454546,
      "grad_norm": 0.35292674837908006,
      "learning_rate": 3.2394736842105264e-05,
      "loss": 0.3651,
      "step": 881
    },
    {
      "epoch": 1.2528409090909092,
      "grad_norm": 0.388560338527763,
      "learning_rate": 3.236842105263158e-05,
      "loss": 0.3592,
      "step": 882
    },
    {
      "epoch": 1.2542613636363638,
      "grad_norm": 0.37807251137782727,
      "learning_rate": 3.2342105263157894e-05,
      "loss": 0.3566,
      "step": 883
    },
    {
      "epoch": 1.2556818181818181,
      "grad_norm": 0.35254271934580145,
      "learning_rate": 3.231578947368421e-05,
      "loss": 0.3653,
      "step": 884
    },
    {
      "epoch": 1.2571022727272727,
      "grad_norm": 0.4985237199596022,
      "learning_rate": 3.228947368421053e-05,
      "loss": 0.3739,
      "step": 885
    },
    {
      "epoch": 1.2585227272727273,
      "grad_norm": 0.31549125165359104,
      "learning_rate": 3.226315789473684e-05,
      "loss": 0.3492,
      "step": 886
    },
    {
      "epoch": 1.2599431818181819,
      "grad_norm": 0.39821508372356484,
      "learning_rate": 3.223684210526316e-05,
      "loss": 0.3798,
      "step": 887
    },
    {
      "epoch": 1.2613636363636362,
      "grad_norm": 0.33346871851341164,
      "learning_rate": 3.221052631578947e-05,
      "loss": 0.3606,
      "step": 888
    },
    {
      "epoch": 1.2627840909090908,
      "grad_norm": 0.35752462715961664,
      "learning_rate": 3.218421052631579e-05,
      "loss": 0.3625,
      "step": 889
    },
    {
      "epoch": 1.2642045454545454,
      "grad_norm": 0.3145834700384594,
      "learning_rate": 3.215789473684211e-05,
      "loss": 0.3605,
      "step": 890
    },
    {
      "epoch": 1.265625,
      "grad_norm": 0.4157471134705189,
      "learning_rate": 3.213157894736842e-05,
      "loss": 0.3674,
      "step": 891
    },
    {
      "epoch": 1.2670454545454546,
      "grad_norm": 0.30952055073544105,
      "learning_rate": 3.210526315789474e-05,
      "loss": 0.3697,
      "step": 892
    },
    {
      "epoch": 1.2684659090909092,
      "grad_norm": 0.3226678525660401,
      "learning_rate": 3.207894736842105e-05,
      "loss": 0.357,
      "step": 893
    },
    {
      "epoch": 1.2698863636363638,
      "grad_norm": 0.5001149793669107,
      "learning_rate": 3.205263157894737e-05,
      "loss": 0.3568,
      "step": 894
    },
    {
      "epoch": 1.2713068181818181,
      "grad_norm": 0.3112740509368127,
      "learning_rate": 3.202631578947369e-05,
      "loss": 0.3566,
      "step": 895
    },
    {
      "epoch": 1.2727272727272727,
      "grad_norm": 0.4018117695988795,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.3624,
      "step": 896
    },
    {
      "epoch": 1.2741477272727273,
      "grad_norm": 0.4364073877343256,
      "learning_rate": 3.197368421052632e-05,
      "loss": 0.359,
      "step": 897
    },
    {
      "epoch": 1.2755681818181819,
      "grad_norm": 0.3554196965654742,
      "learning_rate": 3.194736842105263e-05,
      "loss": 0.3632,
      "step": 898
    },
    {
      "epoch": 1.2769886363636362,
      "grad_norm": 0.3882577349329965,
      "learning_rate": 3.1921052631578954e-05,
      "loss": 0.3589,
      "step": 899
    },
    {
      "epoch": 1.2784090909090908,
      "grad_norm": 0.43160903448724147,
      "learning_rate": 3.1894736842105265e-05,
      "loss": 0.3515,
      "step": 900
    },
    {
      "epoch": 1.2798295454545454,
      "grad_norm": 0.2715939987191114,
      "learning_rate": 3.1868421052631584e-05,
      "loss": 0.3676,
      "step": 901
    },
    {
      "epoch": 1.28125,
      "grad_norm": 0.450488009864911,
      "learning_rate": 3.1842105263157895e-05,
      "loss": 0.3651,
      "step": 902
    },
    {
      "epoch": 1.2826704545454546,
      "grad_norm": 0.3369518131231462,
      "learning_rate": 3.181578947368421e-05,
      "loss": 0.3681,
      "step": 903
    },
    {
      "epoch": 1.2840909090909092,
      "grad_norm": 0.37844695810332524,
      "learning_rate": 3.178947368421053e-05,
      "loss": 0.3821,
      "step": 904
    },
    {
      "epoch": 1.2855113636363638,
      "grad_norm": 0.31466907348343737,
      "learning_rate": 3.1763157894736843e-05,
      "loss": 0.3711,
      "step": 905
    },
    {
      "epoch": 1.2869318181818181,
      "grad_norm": 0.32260726669754236,
      "learning_rate": 3.173684210526316e-05,
      "loss": 0.3653,
      "step": 906
    },
    {
      "epoch": 1.2883522727272727,
      "grad_norm": 0.28962341353615323,
      "learning_rate": 3.1710526315789473e-05,
      "loss": 0.3738,
      "step": 907
    },
    {
      "epoch": 1.2897727272727273,
      "grad_norm": 0.3323969440784247,
      "learning_rate": 3.168421052631579e-05,
      "loss": 0.3592,
      "step": 908
    },
    {
      "epoch": 1.2911931818181819,
      "grad_norm": 0.2865257697426051,
      "learning_rate": 3.165789473684211e-05,
      "loss": 0.3574,
      "step": 909
    },
    {
      "epoch": 1.2926136363636362,
      "grad_norm": 0.32249905438371324,
      "learning_rate": 3.163157894736842e-05,
      "loss": 0.3646,
      "step": 910
    },
    {
      "epoch": 1.2940340909090908,
      "grad_norm": 0.2946355950186551,
      "learning_rate": 3.160526315789474e-05,
      "loss": 0.3794,
      "step": 911
    },
    {
      "epoch": 1.2954545454545454,
      "grad_norm": 0.30197076419264374,
      "learning_rate": 3.157894736842105e-05,
      "loss": 0.372,
      "step": 912
    },
    {
      "epoch": 1.296875,
      "grad_norm": 0.24721520230633395,
      "learning_rate": 3.155263157894737e-05,
      "loss": 0.3829,
      "step": 913
    },
    {
      "epoch": 1.2982954545454546,
      "grad_norm": 0.28753308675765493,
      "learning_rate": 3.152631578947369e-05,
      "loss": 0.4103,
      "step": 914
    },
    {
      "epoch": 1.2997159090909092,
      "grad_norm": 0.27038526112871736,
      "learning_rate": 3.15e-05,
      "loss": 0.3658,
      "step": 915
    },
    {
      "epoch": 1.3011363636363638,
      "grad_norm": 0.30685548393830375,
      "learning_rate": 3.147368421052632e-05,
      "loss": 0.3562,
      "step": 916
    },
    {
      "epoch": 1.3025568181818181,
      "grad_norm": 0.2757008949121105,
      "learning_rate": 3.144736842105263e-05,
      "loss": 0.3643,
      "step": 917
    },
    {
      "epoch": 1.3039772727272727,
      "grad_norm": 0.29561478414849096,
      "learning_rate": 3.142105263157895e-05,
      "loss": 0.3666,
      "step": 918
    },
    {
      "epoch": 1.3053977272727273,
      "grad_norm": 0.2968377022367156,
      "learning_rate": 3.139473684210527e-05,
      "loss": 0.3689,
      "step": 919
    },
    {
      "epoch": 1.3068181818181819,
      "grad_norm": 0.30287307229676275,
      "learning_rate": 3.136842105263158e-05,
      "loss": 0.3727,
      "step": 920
    },
    {
      "epoch": 1.3082386363636362,
      "grad_norm": 0.2935937639426277,
      "learning_rate": 3.1342105263157897e-05,
      "loss": 0.3778,
      "step": 921
    },
    {
      "epoch": 1.3096590909090908,
      "grad_norm": 0.2845312811926222,
      "learning_rate": 3.131578947368421e-05,
      "loss": 0.3365,
      "step": 922
    },
    {
      "epoch": 1.3110795454545454,
      "grad_norm": 0.31396360007873825,
      "learning_rate": 3.1289473684210526e-05,
      "loss": 0.3642,
      "step": 923
    },
    {
      "epoch": 1.3125,
      "grad_norm": 0.2861660219511999,
      "learning_rate": 3.1263157894736845e-05,
      "loss": 0.3607,
      "step": 924
    },
    {
      "epoch": 1.3139204545454546,
      "grad_norm": 0.29916013879751663,
      "learning_rate": 3.1236842105263156e-05,
      "loss": 0.3787,
      "step": 925
    },
    {
      "epoch": 1.3153409090909092,
      "grad_norm": 0.3092619961888014,
      "learning_rate": 3.1210526315789475e-05,
      "loss": 0.373,
      "step": 926
    },
    {
      "epoch": 1.3167613636363638,
      "grad_norm": 0.2714428160728827,
      "learning_rate": 3.118421052631579e-05,
      "loss": 0.3648,
      "step": 927
    },
    {
      "epoch": 1.3181818181818181,
      "grad_norm": 0.29468475867767524,
      "learning_rate": 3.115789473684211e-05,
      "loss": 0.3787,
      "step": 928
    },
    {
      "epoch": 1.3196022727272727,
      "grad_norm": 0.34668239752989277,
      "learning_rate": 3.113157894736842e-05,
      "loss": 0.3653,
      "step": 929
    },
    {
      "epoch": 1.3210227272727273,
      "grad_norm": 0.29819856609510065,
      "learning_rate": 3.1105263157894735e-05,
      "loss": 0.3712,
      "step": 930
    },
    {
      "epoch": 1.3224431818181819,
      "grad_norm": 0.31896393216509095,
      "learning_rate": 3.107894736842105e-05,
      "loss": 0.3468,
      "step": 931
    },
    {
      "epoch": 1.3238636363636362,
      "grad_norm": 0.2998386213374353,
      "learning_rate": 3.105263157894737e-05,
      "loss": 0.3613,
      "step": 932
    },
    {
      "epoch": 1.3252840909090908,
      "grad_norm": 0.2693763930000581,
      "learning_rate": 3.102631578947369e-05,
      "loss": 0.3487,
      "step": 933
    },
    {
      "epoch": 1.3267045454545454,
      "grad_norm": 0.3486268281476277,
      "learning_rate": 3.1e-05,
      "loss": 0.385,
      "step": 934
    },
    {
      "epoch": 1.328125,
      "grad_norm": 0.2792257423082131,
      "learning_rate": 3.097368421052631e-05,
      "loss": 0.365,
      "step": 935
    },
    {
      "epoch": 1.3295454545454546,
      "grad_norm": 0.3016036759466146,
      "learning_rate": 3.094736842105263e-05,
      "loss": 0.367,
      "step": 936
    },
    {
      "epoch": 1.3309659090909092,
      "grad_norm": 0.28384154183145627,
      "learning_rate": 3.092105263157895e-05,
      "loss": 0.3524,
      "step": 937
    },
    {
      "epoch": 1.3323863636363638,
      "grad_norm": 0.2687005471728445,
      "learning_rate": 3.089473684210527e-05,
      "loss": 0.3675,
      "step": 938
    },
    {
      "epoch": 1.3338068181818181,
      "grad_norm": 0.2939343769053132,
      "learning_rate": 3.086842105263158e-05,
      "loss": 0.3597,
      "step": 939
    },
    {
      "epoch": 1.3352272727272727,
      "grad_norm": 0.314654767053052,
      "learning_rate": 3.084210526315789e-05,
      "loss": 0.3573,
      "step": 940
    },
    {
      "epoch": 1.3366477272727273,
      "grad_norm": 0.3039352744257816,
      "learning_rate": 3.0815789473684216e-05,
      "loss": 0.381,
      "step": 941
    },
    {
      "epoch": 1.3380681818181819,
      "grad_norm": 0.31613434515471395,
      "learning_rate": 3.078947368421053e-05,
      "loss": 0.367,
      "step": 942
    },
    {
      "epoch": 1.3394886363636362,
      "grad_norm": 0.32252463071145343,
      "learning_rate": 3.0763157894736846e-05,
      "loss": 0.3674,
      "step": 943
    },
    {
      "epoch": 1.3409090909090908,
      "grad_norm": 0.33554526589768086,
      "learning_rate": 3.073684210526316e-05,
      "loss": 0.3551,
      "step": 944
    },
    {
      "epoch": 1.3423295454545454,
      "grad_norm": 0.3371529083722537,
      "learning_rate": 3.0710526315789476e-05,
      "loss": 0.3696,
      "step": 945
    },
    {
      "epoch": 1.34375,
      "grad_norm": 0.31526648454893497,
      "learning_rate": 3.0684210526315795e-05,
      "loss": 0.3653,
      "step": 946
    },
    {
      "epoch": 1.3451704545454546,
      "grad_norm": 0.3828009717787276,
      "learning_rate": 3.0657894736842106e-05,
      "loss": 0.3706,
      "step": 947
    },
    {
      "epoch": 1.3465909090909092,
      "grad_norm": 0.41801947990786376,
      "learning_rate": 3.0631578947368425e-05,
      "loss": 0.3999,
      "step": 948
    },
    {
      "epoch": 1.3480113636363638,
      "grad_norm": 0.32251082460455643,
      "learning_rate": 3.0605263157894736e-05,
      "loss": 0.3606,
      "step": 949
    },
    {
      "epoch": 1.3494318181818181,
      "grad_norm": 0.35059412415659846,
      "learning_rate": 3.0578947368421054e-05,
      "loss": 0.3581,
      "step": 950
    },
    {
      "epoch": 1.3508522727272727,
      "grad_norm": 0.32067045112928544,
      "learning_rate": 3.055263157894737e-05,
      "loss": 0.3853,
      "step": 951
    },
    {
      "epoch": 1.3522727272727273,
      "grad_norm": 0.32853793874237647,
      "learning_rate": 3.0526315789473684e-05,
      "loss": 0.3544,
      "step": 952
    },
    {
      "epoch": 1.3536931818181819,
      "grad_norm": 0.31840885152692794,
      "learning_rate": 3.05e-05,
      "loss": 0.3587,
      "step": 953
    },
    {
      "epoch": 1.3551136363636362,
      "grad_norm": 0.3806758718946067,
      "learning_rate": 3.0473684210526314e-05,
      "loss": 0.3838,
      "step": 954
    },
    {
      "epoch": 1.3565340909090908,
      "grad_norm": 0.34638370957022485,
      "learning_rate": 3.0447368421052636e-05,
      "loss": 0.3827,
      "step": 955
    },
    {
      "epoch": 1.3579545454545454,
      "grad_norm": 0.3671992363001913,
      "learning_rate": 3.042105263157895e-05,
      "loss": 0.3605,
      "step": 956
    },
    {
      "epoch": 1.359375,
      "grad_norm": 0.3118206520920557,
      "learning_rate": 3.0394736842105266e-05,
      "loss": 0.3866,
      "step": 957
    },
    {
      "epoch": 1.3607954545454546,
      "grad_norm": 0.3230808742516086,
      "learning_rate": 3.036842105263158e-05,
      "loss": 0.3679,
      "step": 958
    },
    {
      "epoch": 1.3622159090909092,
      "grad_norm": 0.29733963362702864,
      "learning_rate": 3.0342105263157893e-05,
      "loss": 0.3694,
      "step": 959
    },
    {
      "epoch": 1.3636363636363638,
      "grad_norm": 0.2870607790960171,
      "learning_rate": 3.0315789473684214e-05,
      "loss": 0.3573,
      "step": 960
    },
    {
      "epoch": 1.3650568181818181,
      "grad_norm": 0.33845799164774765,
      "learning_rate": 3.028947368421053e-05,
      "loss": 0.3589,
      "step": 961
    },
    {
      "epoch": 1.3664772727272727,
      "grad_norm": 0.309227309700292,
      "learning_rate": 3.0263157894736844e-05,
      "loss": 0.3562,
      "step": 962
    },
    {
      "epoch": 1.3678977272727273,
      "grad_norm": 0.29604667846978844,
      "learning_rate": 3.023684210526316e-05,
      "loss": 0.3587,
      "step": 963
    },
    {
      "epoch": 1.3693181818181819,
      "grad_norm": 0.34237915545212716,
      "learning_rate": 3.021052631578947e-05,
      "loss": 0.3782,
      "step": 964
    },
    {
      "epoch": 1.3707386363636362,
      "grad_norm": 0.3097453173801252,
      "learning_rate": 3.0184210526315793e-05,
      "loss": 0.3689,
      "step": 965
    },
    {
      "epoch": 1.3721590909090908,
      "grad_norm": 0.36066055790859386,
      "learning_rate": 3.0157894736842108e-05,
      "loss": 0.3732,
      "step": 966
    },
    {
      "epoch": 1.3735795454545454,
      "grad_norm": 0.34892593202471583,
      "learning_rate": 3.0131578947368423e-05,
      "loss": 0.3756,
      "step": 967
    },
    {
      "epoch": 1.375,
      "grad_norm": 0.32282185369314387,
      "learning_rate": 3.0105263157894737e-05,
      "loss": 0.3817,
      "step": 968
    },
    {
      "epoch": 1.3764204545454546,
      "grad_norm": 0.2803395076794015,
      "learning_rate": 3.0078947368421056e-05,
      "loss": 0.3624,
      "step": 969
    },
    {
      "epoch": 1.3778409090909092,
      "grad_norm": 0.29667968151027646,
      "learning_rate": 3.005263157894737e-05,
      "loss": 0.3629,
      "step": 970
    },
    {
      "epoch": 1.3792613636363638,
      "grad_norm": 0.3120067866718459,
      "learning_rate": 3.0026315789473686e-05,
      "loss": 0.3692,
      "step": 971
    },
    {
      "epoch": 1.3806818181818181,
      "grad_norm": 0.3091673348581161,
      "learning_rate": 3e-05,
      "loss": 0.3751,
      "step": 972
    },
    {
      "epoch": 1.3821022727272727,
      "grad_norm": 0.32391140784785105,
      "learning_rate": 2.9973684210526316e-05,
      "loss": 0.3717,
      "step": 973
    },
    {
      "epoch": 1.3835227272727273,
      "grad_norm": 0.38076211046022457,
      "learning_rate": 2.9947368421052634e-05,
      "loss": 0.3758,
      "step": 974
    },
    {
      "epoch": 1.3849431818181819,
      "grad_norm": 0.26076243600805543,
      "learning_rate": 2.992105263157895e-05,
      "loss": 0.3606,
      "step": 975
    },
    {
      "epoch": 1.3863636363636362,
      "grad_norm": 0.34270509996879644,
      "learning_rate": 2.9894736842105264e-05,
      "loss": 0.3555,
      "step": 976
    },
    {
      "epoch": 1.3877840909090908,
      "grad_norm": 0.31978621701259996,
      "learning_rate": 2.986842105263158e-05,
      "loss": 0.3568,
      "step": 977
    },
    {
      "epoch": 1.3892045454545454,
      "grad_norm": 0.2972982183628563,
      "learning_rate": 2.9842105263157894e-05,
      "loss": 0.3643,
      "step": 978
    },
    {
      "epoch": 1.390625,
      "grad_norm": 0.32788291198148706,
      "learning_rate": 2.9815789473684212e-05,
      "loss": 0.3773,
      "step": 979
    },
    {
      "epoch": 1.3920454545454546,
      "grad_norm": 0.2808304714074782,
      "learning_rate": 2.9789473684210527e-05,
      "loss": 0.3731,
      "step": 980
    },
    {
      "epoch": 1.3934659090909092,
      "grad_norm": 0.3219366206423956,
      "learning_rate": 2.9763157894736842e-05,
      "loss": 0.3687,
      "step": 981
    },
    {
      "epoch": 1.3948863636363638,
      "grad_norm": 0.2919696481807413,
      "learning_rate": 2.9736842105263157e-05,
      "loss": 0.3524,
      "step": 982
    },
    {
      "epoch": 1.3963068181818181,
      "grad_norm": 0.3372235422308735,
      "learning_rate": 2.971052631578948e-05,
      "loss": 0.3485,
      "step": 983
    },
    {
      "epoch": 1.3977272727272727,
      "grad_norm": 0.3032056950876354,
      "learning_rate": 2.968421052631579e-05,
      "loss": 0.3449,
      "step": 984
    },
    {
      "epoch": 1.3991477272727273,
      "grad_norm": 0.3104484487689201,
      "learning_rate": 2.9657894736842106e-05,
      "loss": 0.3765,
      "step": 985
    },
    {
      "epoch": 1.4005681818181819,
      "grad_norm": 0.27062710421136316,
      "learning_rate": 2.963157894736842e-05,
      "loss": 0.363,
      "step": 986
    },
    {
      "epoch": 1.4019886363636362,
      "grad_norm": 0.3187203085131546,
      "learning_rate": 2.9605263157894735e-05,
      "loss": 0.3831,
      "step": 987
    },
    {
      "epoch": 1.4034090909090908,
      "grad_norm": 0.253922948393278,
      "learning_rate": 2.9578947368421057e-05,
      "loss": 0.3693,
      "step": 988
    },
    {
      "epoch": 1.4048295454545454,
      "grad_norm": 0.3067800466972296,
      "learning_rate": 2.9552631578947372e-05,
      "loss": 0.3631,
      "step": 989
    },
    {
      "epoch": 1.40625,
      "grad_norm": 0.30465125005657645,
      "learning_rate": 2.9526315789473684e-05,
      "loss": 0.3735,
      "step": 990
    },
    {
      "epoch": 1.4076704545454546,
      "grad_norm": 0.3010386797599766,
      "learning_rate": 2.95e-05,
      "loss": 0.3601,
      "step": 991
    },
    {
      "epoch": 1.4090909090909092,
      "grad_norm": 0.30890157126759,
      "learning_rate": 2.9473684210526314e-05,
      "loss": 0.3673,
      "step": 992
    },
    {
      "epoch": 1.4105113636363638,
      "grad_norm": 0.28964239160477323,
      "learning_rate": 2.9447368421052635e-05,
      "loss": 0.3496,
      "step": 993
    },
    {
      "epoch": 1.4119318181818181,
      "grad_norm": 0.3218882016182759,
      "learning_rate": 2.942105263157895e-05,
      "loss": 0.38,
      "step": 994
    },
    {
      "epoch": 1.4133522727272727,
      "grad_norm": 0.2885116066306885,
      "learning_rate": 2.9394736842105265e-05,
      "loss": 0.3833,
      "step": 995
    },
    {
      "epoch": 1.4147727272727273,
      "grad_norm": 0.28957700521037816,
      "learning_rate": 2.9368421052631577e-05,
      "loss": 0.3825,
      "step": 996
    },
    {
      "epoch": 1.4161931818181819,
      "grad_norm": 0.2716206802911739,
      "learning_rate": 2.93421052631579e-05,
      "loss": 0.3666,
      "step": 997
    },
    {
      "epoch": 1.4176136363636362,
      "grad_norm": 0.3075969663670002,
      "learning_rate": 2.9315789473684214e-05,
      "loss": 0.3633,
      "step": 998
    },
    {
      "epoch": 1.4190340909090908,
      "grad_norm": 0.2549655535753946,
      "learning_rate": 2.928947368421053e-05,
      "loss": 0.3517,
      "step": 999
    },
    {
      "epoch": 1.4204545454545454,
      "grad_norm": 0.29095427517739014,
      "learning_rate": 2.9263157894736844e-05,
      "loss": 0.3475,
      "step": 1000
    },
    {
      "epoch": 1.421875,
      "grad_norm": 0.2740882969198946,
      "learning_rate": 2.9236842105263155e-05,
      "loss": 0.3494,
      "step": 1001
    },
    {
      "epoch": 1.4232954545454546,
      "grad_norm": 0.25929487658254075,
      "learning_rate": 2.9210526315789477e-05,
      "loss": 0.3633,
      "step": 1002
    },
    {
      "epoch": 1.4247159090909092,
      "grad_norm": 0.30665446350813313,
      "learning_rate": 2.9184210526315792e-05,
      "loss": 0.3845,
      "step": 1003
    },
    {
      "epoch": 1.4261363636363638,
      "grad_norm": 0.2963836307116242,
      "learning_rate": 2.9157894736842107e-05,
      "loss": 0.3599,
      "step": 1004
    },
    {
      "epoch": 1.4275568181818181,
      "grad_norm": 0.32484449760954687,
      "learning_rate": 2.9131578947368422e-05,
      "loss": 0.363,
      "step": 1005
    },
    {
      "epoch": 1.4289772727272727,
      "grad_norm": 0.2609736427385125,
      "learning_rate": 2.910526315789474e-05,
      "loss": 0.3549,
      "step": 1006
    },
    {
      "epoch": 1.4303977272727273,
      "grad_norm": 0.3478071372268831,
      "learning_rate": 2.9078947368421055e-05,
      "loss": 0.379,
      "step": 1007
    },
    {
      "epoch": 1.4318181818181819,
      "grad_norm": 0.346695149962594,
      "learning_rate": 2.905263157894737e-05,
      "loss": 0.3682,
      "step": 1008
    },
    {
      "epoch": 1.4332386363636362,
      "grad_norm": 0.31122730704140183,
      "learning_rate": 2.9026315789473685e-05,
      "loss": 0.3649,
      "step": 1009
    },
    {
      "epoch": 1.4346590909090908,
      "grad_norm": 0.28791836698974005,
      "learning_rate": 2.9e-05,
      "loss": 0.3624,
      "step": 1010
    },
    {
      "epoch": 1.4360795454545454,
      "grad_norm": 0.46809996269048143,
      "learning_rate": 2.897368421052632e-05,
      "loss": 0.3946,
      "step": 1011
    },
    {
      "epoch": 1.4375,
      "grad_norm": 0.26014008777348024,
      "learning_rate": 2.8947368421052634e-05,
      "loss": 0.3617,
      "step": 1012
    },
    {
      "epoch": 1.4389204545454546,
      "grad_norm": 0.456748796297461,
      "learning_rate": 2.892105263157895e-05,
      "loss": 0.3666,
      "step": 1013
    },
    {
      "epoch": 1.4403409090909092,
      "grad_norm": 0.3018291206921549,
      "learning_rate": 2.8894736842105263e-05,
      "loss": 0.3825,
      "step": 1014
    },
    {
      "epoch": 1.4417613636363638,
      "grad_norm": 0.39213786570521836,
      "learning_rate": 2.886842105263158e-05,
      "loss": 0.3676,
      "step": 1015
    },
    {
      "epoch": 1.4431818181818181,
      "grad_norm": 0.3851611761162801,
      "learning_rate": 2.8842105263157897e-05,
      "loss": 0.381,
      "step": 1016
    },
    {
      "epoch": 1.4446022727272727,
      "grad_norm": 0.4066335188925192,
      "learning_rate": 2.8815789473684212e-05,
      "loss": 0.3633,
      "step": 1017
    },
    {
      "epoch": 1.4460227272727273,
      "grad_norm": 0.3605821967100767,
      "learning_rate": 2.8789473684210527e-05,
      "loss": 0.377,
      "step": 1018
    },
    {
      "epoch": 1.4474431818181819,
      "grad_norm": 0.39317301800557153,
      "learning_rate": 2.876315789473684e-05,
      "loss": 0.3696,
      "step": 1019
    },
    {
      "epoch": 1.4488636363636362,
      "grad_norm": 0.37721328397021753,
      "learning_rate": 2.8736842105263163e-05,
      "loss": 0.3761,
      "step": 1020
    },
    {
      "epoch": 1.4502840909090908,
      "grad_norm": 0.3332198849383193,
      "learning_rate": 2.8710526315789475e-05,
      "loss": 0.3601,
      "step": 1021
    },
    {
      "epoch": 1.4517045454545454,
      "grad_norm": 0.38738063234119946,
      "learning_rate": 2.868421052631579e-05,
      "loss": 0.3675,
      "step": 1022
    },
    {
      "epoch": 1.453125,
      "grad_norm": 0.32669413102513345,
      "learning_rate": 2.8657894736842105e-05,
      "loss": 0.361,
      "step": 1023
    },
    {
      "epoch": 1.4545454545454546,
      "grad_norm": 0.45904905544296665,
      "learning_rate": 2.863157894736842e-05,
      "loss": 0.3718,
      "step": 1024
    },
    {
      "epoch": 1.4559659090909092,
      "grad_norm": 0.3276733214404158,
      "learning_rate": 2.860526315789474e-05,
      "loss": 0.3639,
      "step": 1025
    },
    {
      "epoch": 1.4573863636363638,
      "grad_norm": 0.3517321446209726,
      "learning_rate": 2.8578947368421057e-05,
      "loss": 0.3335,
      "step": 1026
    },
    {
      "epoch": 1.4588068181818181,
      "grad_norm": 0.2934150828028833,
      "learning_rate": 2.8552631578947368e-05,
      "loss": 0.3464,
      "step": 1027
    },
    {
      "epoch": 1.4602272727272727,
      "grad_norm": 0.3631797329129774,
      "learning_rate": 2.8526315789473683e-05,
      "loss": 0.3699,
      "step": 1028
    },
    {
      "epoch": 1.4616477272727273,
      "grad_norm": 0.3120049412369448,
      "learning_rate": 2.8499999999999998e-05,
      "loss": 0.3557,
      "step": 1029
    },
    {
      "epoch": 1.4630681818181819,
      "grad_norm": 0.2675498837540677,
      "learning_rate": 2.847368421052632e-05,
      "loss": 0.3642,
      "step": 1030
    },
    {
      "epoch": 1.4644886363636362,
      "grad_norm": 0.3576195995588519,
      "learning_rate": 2.8447368421052635e-05,
      "loss": 0.365,
      "step": 1031
    },
    {
      "epoch": 1.4659090909090908,
      "grad_norm": 0.33494984908913084,
      "learning_rate": 2.842105263157895e-05,
      "loss": 0.3571,
      "step": 1032
    },
    {
      "epoch": 1.4673295454545454,
      "grad_norm": 0.3053248818182251,
      "learning_rate": 2.839473684210526e-05,
      "loss": 0.3518,
      "step": 1033
    },
    {
      "epoch": 1.46875,
      "grad_norm": 0.29045610627187984,
      "learning_rate": 2.8368421052631583e-05,
      "loss": 0.3632,
      "step": 1034
    },
    {
      "epoch": 1.4701704545454546,
      "grad_norm": 0.3435484664621267,
      "learning_rate": 2.8342105263157898e-05,
      "loss": 0.3698,
      "step": 1035
    },
    {
      "epoch": 1.4715909090909092,
      "grad_norm": 0.32943057118207353,
      "learning_rate": 2.8315789473684213e-05,
      "loss": 0.3606,
      "step": 1036
    },
    {
      "epoch": 1.4730113636363638,
      "grad_norm": 0.36062905299671,
      "learning_rate": 2.8289473684210528e-05,
      "loss": 0.3904,
      "step": 1037
    },
    {
      "epoch": 1.4744318181818181,
      "grad_norm": 0.31893891973369465,
      "learning_rate": 2.826315789473684e-05,
      "loss": 0.367,
      "step": 1038
    },
    {
      "epoch": 1.4758522727272727,
      "grad_norm": 0.3556192314105621,
      "learning_rate": 2.823684210526316e-05,
      "loss": 0.3385,
      "step": 1039
    },
    {
      "epoch": 1.4772727272727273,
      "grad_norm": 0.28370592053449023,
      "learning_rate": 2.8210526315789476e-05,
      "loss": 0.3554,
      "step": 1040
    },
    {
      "epoch": 1.4786931818181819,
      "grad_norm": 0.33100819637223555,
      "learning_rate": 2.818421052631579e-05,
      "loss": 0.3913,
      "step": 1041
    },
    {
      "epoch": 1.4801136363636362,
      "grad_norm": 0.2874490684272293,
      "learning_rate": 2.8157894736842106e-05,
      "loss": 0.3464,
      "step": 1042
    },
    {
      "epoch": 1.4815340909090908,
      "grad_norm": 0.3202209693604776,
      "learning_rate": 2.813157894736842e-05,
      "loss": 0.3819,
      "step": 1043
    },
    {
      "epoch": 1.4829545454545454,
      "grad_norm": 0.2957884398821572,
      "learning_rate": 2.810526315789474e-05,
      "loss": 0.3608,
      "step": 1044
    },
    {
      "epoch": 1.484375,
      "grad_norm": 0.3128669836001024,
      "learning_rate": 2.8078947368421055e-05,
      "loss": 0.3755,
      "step": 1045
    },
    {
      "epoch": 1.4857954545454546,
      "grad_norm": 0.3385227551291573,
      "learning_rate": 2.805263157894737e-05,
      "loss": 0.3668,
      "step": 1046
    },
    {
      "epoch": 1.4872159090909092,
      "grad_norm": 0.29625322418369315,
      "learning_rate": 2.8026315789473685e-05,
      "loss": 0.3849,
      "step": 1047
    },
    {
      "epoch": 1.4886363636363638,
      "grad_norm": 0.2798527008103768,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 0.3555,
      "step": 1048
    },
    {
      "epoch": 1.4900568181818181,
      "grad_norm": 0.3975037425656684,
      "learning_rate": 2.7973684210526318e-05,
      "loss": 0.3567,
      "step": 1049
    },
    {
      "epoch": 1.4914772727272727,
      "grad_norm": 0.2637279469459038,
      "learning_rate": 2.7947368421052633e-05,
      "loss": 0.3491,
      "step": 1050
    },
    {
      "epoch": 1.4928977272727273,
      "grad_norm": 0.35561723454027677,
      "learning_rate": 2.7921052631578948e-05,
      "loss": 0.3688,
      "step": 1051
    },
    {
      "epoch": 1.4943181818181819,
      "grad_norm": 0.31059370687464316,
      "learning_rate": 2.7894736842105263e-05,
      "loss": 0.3693,
      "step": 1052
    },
    {
      "epoch": 1.4957386363636362,
      "grad_norm": 0.27841716691569074,
      "learning_rate": 2.786842105263158e-05,
      "loss": 0.3671,
      "step": 1053
    },
    {
      "epoch": 1.4971590909090908,
      "grad_norm": 0.32356753015120615,
      "learning_rate": 2.7842105263157896e-05,
      "loss": 0.3561,
      "step": 1054
    },
    {
      "epoch": 1.4985795454545454,
      "grad_norm": 0.26856848803551975,
      "learning_rate": 2.781578947368421e-05,
      "loss": 0.3719,
      "step": 1055
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.27715382730176735,
      "learning_rate": 2.7789473684210526e-05,
      "loss": 0.3699,
      "step": 1056
    },
    {
      "epoch": 1.5014204545454546,
      "grad_norm": 0.3453078266197115,
      "learning_rate": 2.776315789473684e-05,
      "loss": 0.3757,
      "step": 1057
    },
    {
      "epoch": 1.5028409090909092,
      "grad_norm": 0.33592910785825597,
      "learning_rate": 2.773684210526316e-05,
      "loss": 0.3703,
      "step": 1058
    },
    {
      "epoch": 1.5042613636363638,
      "grad_norm": 0.33589360985832173,
      "learning_rate": 2.7710526315789474e-05,
      "loss": 0.3888,
      "step": 1059
    },
    {
      "epoch": 1.5056818181818183,
      "grad_norm": 0.33503231845772014,
      "learning_rate": 2.768421052631579e-05,
      "loss": 0.3763,
      "step": 1060
    },
    {
      "epoch": 1.5071022727272727,
      "grad_norm": 0.31487317914133384,
      "learning_rate": 2.7657894736842104e-05,
      "loss": 0.3585,
      "step": 1061
    },
    {
      "epoch": 1.5085227272727273,
      "grad_norm": 0.28652689921262403,
      "learning_rate": 2.7631578947368426e-05,
      "loss": 0.3727,
      "step": 1062
    },
    {
      "epoch": 1.5099431818181817,
      "grad_norm": 0.33212940948447345,
      "learning_rate": 2.760526315789474e-05,
      "loss": 0.3641,
      "step": 1063
    },
    {
      "epoch": 1.5113636363636362,
      "grad_norm": 0.3089341516964658,
      "learning_rate": 2.7578947368421053e-05,
      "loss": 0.3708,
      "step": 1064
    },
    {
      "epoch": 1.5127840909090908,
      "grad_norm": 0.3304741247684637,
      "learning_rate": 2.7552631578947368e-05,
      "loss": 0.3616,
      "step": 1065
    },
    {
      "epoch": 1.5142045454545454,
      "grad_norm": 0.27880399924755983,
      "learning_rate": 2.7526315789473683e-05,
      "loss": 0.3519,
      "step": 1066
    },
    {
      "epoch": 1.515625,
      "grad_norm": 0.36532498268676455,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 0.3786,
      "step": 1067
    },
    {
      "epoch": 1.5170454545454546,
      "grad_norm": 0.2932943444150953,
      "learning_rate": 2.747368421052632e-05,
      "loss": 0.3877,
      "step": 1068
    },
    {
      "epoch": 1.5184659090909092,
      "grad_norm": 0.26979805259315315,
      "learning_rate": 2.7447368421052634e-05,
      "loss": 0.3672,
      "step": 1069
    },
    {
      "epoch": 1.5198863636363638,
      "grad_norm": 0.2890020762141782,
      "learning_rate": 2.7421052631578946e-05,
      "loss": 0.3543,
      "step": 1070
    },
    {
      "epoch": 1.5213068181818183,
      "grad_norm": 0.32477893328936397,
      "learning_rate": 2.739473684210526e-05,
      "loss": 0.3832,
      "step": 1071
    },
    {
      "epoch": 1.5227272727272727,
      "grad_norm": 0.2710845973691223,
      "learning_rate": 2.7368421052631583e-05,
      "loss": 0.3795,
      "step": 1072
    },
    {
      "epoch": 1.5241477272727273,
      "grad_norm": 0.29748811706138534,
      "learning_rate": 2.7342105263157898e-05,
      "loss": 0.3662,
      "step": 1073
    },
    {
      "epoch": 1.5255681818181817,
      "grad_norm": 0.2972860593371942,
      "learning_rate": 2.7315789473684213e-05,
      "loss": 0.3669,
      "step": 1074
    },
    {
      "epoch": 1.5269886363636362,
      "grad_norm": 0.27259475297673985,
      "learning_rate": 2.7289473684210528e-05,
      "loss": 0.3758,
      "step": 1075
    },
    {
      "epoch": 1.5284090909090908,
      "grad_norm": 0.30688656471318054,
      "learning_rate": 2.7263157894736846e-05,
      "loss": 0.362,
      "step": 1076
    },
    {
      "epoch": 1.5298295454545454,
      "grad_norm": 0.2767354968476221,
      "learning_rate": 2.723684210526316e-05,
      "loss": 0.3583,
      "step": 1077
    },
    {
      "epoch": 1.53125,
      "grad_norm": 0.28316654611654657,
      "learning_rate": 2.7210526315789476e-05,
      "loss": 0.3703,
      "step": 1078
    },
    {
      "epoch": 1.5326704545454546,
      "grad_norm": 0.30196974881671673,
      "learning_rate": 2.718421052631579e-05,
      "loss": 0.3717,
      "step": 1079
    },
    {
      "epoch": 1.5340909090909092,
      "grad_norm": 0.32223442678655484,
      "learning_rate": 2.7157894736842106e-05,
      "loss": 0.3546,
      "step": 1080
    },
    {
      "epoch": 1.5355113636363638,
      "grad_norm": 0.33826494852690386,
      "learning_rate": 2.7131578947368424e-05,
      "loss": 0.3739,
      "step": 1081
    },
    {
      "epoch": 1.5369318181818183,
      "grad_norm": 0.29822107627466465,
      "learning_rate": 2.710526315789474e-05,
      "loss": 0.3765,
      "step": 1082
    },
    {
      "epoch": 1.5383522727272727,
      "grad_norm": 0.3034426733099742,
      "learning_rate": 2.7078947368421054e-05,
      "loss": 0.3491,
      "step": 1083
    },
    {
      "epoch": 1.5397727272727273,
      "grad_norm": 0.30834522208220416,
      "learning_rate": 2.705263157894737e-05,
      "loss": 0.3725,
      "step": 1084
    },
    {
      "epoch": 1.5411931818181817,
      "grad_norm": 0.31240048178195745,
      "learning_rate": 2.7026315789473684e-05,
      "loss": 0.3729,
      "step": 1085
    },
    {
      "epoch": 1.5426136363636362,
      "grad_norm": 0.26604991769016106,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 0.3615,
      "step": 1086
    },
    {
      "epoch": 1.5440340909090908,
      "grad_norm": 0.33059903926902195,
      "learning_rate": 2.6973684210526317e-05,
      "loss": 0.3661,
      "step": 1087
    },
    {
      "epoch": 1.5454545454545454,
      "grad_norm": 0.33350629574868723,
      "learning_rate": 2.6947368421052632e-05,
      "loss": 0.3685,
      "step": 1088
    },
    {
      "epoch": 1.546875,
      "grad_norm": 0.3184030622880853,
      "learning_rate": 2.6921052631578947e-05,
      "loss": 0.38,
      "step": 1089
    },
    {
      "epoch": 1.5482954545454546,
      "grad_norm": 0.2720689914034998,
      "learning_rate": 2.6894736842105266e-05,
      "loss": 0.3719,
      "step": 1090
    },
    {
      "epoch": 1.5497159090909092,
      "grad_norm": 0.30935286205914597,
      "learning_rate": 2.686842105263158e-05,
      "loss": 0.3563,
      "step": 1091
    },
    {
      "epoch": 1.5511363636363638,
      "grad_norm": 0.26810808866320085,
      "learning_rate": 2.6842105263157896e-05,
      "loss": 0.3655,
      "step": 1092
    },
    {
      "epoch": 1.5525568181818183,
      "grad_norm": 0.3219513391787826,
      "learning_rate": 2.681578947368421e-05,
      "loss": 0.3602,
      "step": 1093
    },
    {
      "epoch": 1.5539772727272727,
      "grad_norm": 0.2992037412806051,
      "learning_rate": 2.6789473684210526e-05,
      "loss": 0.3772,
      "step": 1094
    },
    {
      "epoch": 1.5553977272727273,
      "grad_norm": 0.3180290787234802,
      "learning_rate": 2.6763157894736844e-05,
      "loss": 0.3594,
      "step": 1095
    },
    {
      "epoch": 1.5568181818181817,
      "grad_norm": 0.33353457396747976,
      "learning_rate": 2.673684210526316e-05,
      "loss": 0.3686,
      "step": 1096
    },
    {
      "epoch": 1.5582386363636362,
      "grad_norm": 0.3657805527879292,
      "learning_rate": 2.6710526315789474e-05,
      "loss": 0.3768,
      "step": 1097
    },
    {
      "epoch": 1.5596590909090908,
      "grad_norm": 0.2859860473526388,
      "learning_rate": 2.668421052631579e-05,
      "loss": 0.359,
      "step": 1098
    },
    {
      "epoch": 1.5610795454545454,
      "grad_norm": 0.3080541661396133,
      "learning_rate": 2.6657894736842104e-05,
      "loss": 0.3737,
      "step": 1099
    },
    {
      "epoch": 1.5625,
      "grad_norm": 0.3217539282158535,
      "learning_rate": 2.6631578947368426e-05,
      "loss": 0.3736,
      "step": 1100
    },
    {
      "epoch": 1.5639204545454546,
      "grad_norm": 0.28979202991109215,
      "learning_rate": 2.6605263157894737e-05,
      "loss": 0.3614,
      "step": 1101
    },
    {
      "epoch": 1.5653409090909092,
      "grad_norm": 0.3047364236715056,
      "learning_rate": 2.6578947368421052e-05,
      "loss": 0.3643,
      "step": 1102
    },
    {
      "epoch": 1.5667613636363638,
      "grad_norm": 0.2873385594244748,
      "learning_rate": 2.6552631578947367e-05,
      "loss": 0.3669,
      "step": 1103
    },
    {
      "epoch": 1.5681818181818183,
      "grad_norm": 0.3426103775411652,
      "learning_rate": 2.652631578947369e-05,
      "loss": 0.3797,
      "step": 1104
    },
    {
      "epoch": 1.5696022727272727,
      "grad_norm": 0.3032721507418151,
      "learning_rate": 2.6500000000000004e-05,
      "loss": 0.3897,
      "step": 1105
    },
    {
      "epoch": 1.5710227272727273,
      "grad_norm": 0.26729676311631706,
      "learning_rate": 2.647368421052632e-05,
      "loss": 0.3653,
      "step": 1106
    },
    {
      "epoch": 1.5724431818181817,
      "grad_norm": 0.2957237192513544,
      "learning_rate": 2.644736842105263e-05,
      "loss": 0.3621,
      "step": 1107
    },
    {
      "epoch": 1.5738636363636362,
      "grad_norm": 0.28735938826154467,
      "learning_rate": 2.6421052631578945e-05,
      "loss": 0.3669,
      "step": 1108
    },
    {
      "epoch": 1.5752840909090908,
      "grad_norm": 0.2954444019960243,
      "learning_rate": 2.6394736842105267e-05,
      "loss": 0.3577,
      "step": 1109
    },
    {
      "epoch": 1.5767045454545454,
      "grad_norm": 0.29275141123237525,
      "learning_rate": 2.6368421052631582e-05,
      "loss": 0.3462,
      "step": 1110
    },
    {
      "epoch": 1.578125,
      "grad_norm": 0.29799546608923644,
      "learning_rate": 2.6342105263157897e-05,
      "loss": 0.3558,
      "step": 1111
    },
    {
      "epoch": 1.5795454545454546,
      "grad_norm": 0.338336997333166,
      "learning_rate": 2.6315789473684212e-05,
      "loss": 0.3634,
      "step": 1112
    },
    {
      "epoch": 1.5809659090909092,
      "grad_norm": 0.2871152055619976,
      "learning_rate": 2.6289473684210524e-05,
      "loss": 0.3522,
      "step": 1113
    },
    {
      "epoch": 1.5823863636363638,
      "grad_norm": 0.2838975348653714,
      "learning_rate": 2.6263157894736845e-05,
      "loss": 0.3507,
      "step": 1114
    },
    {
      "epoch": 1.5838068181818183,
      "grad_norm": 0.3179908565111353,
      "learning_rate": 2.623684210526316e-05,
      "loss": 0.3688,
      "step": 1115
    },
    {
      "epoch": 1.5852272727272727,
      "grad_norm": 0.32663373744992097,
      "learning_rate": 2.6210526315789475e-05,
      "loss": 0.3899,
      "step": 1116
    },
    {
      "epoch": 1.5866477272727273,
      "grad_norm": 0.3019678138003231,
      "learning_rate": 2.618421052631579e-05,
      "loss": 0.3732,
      "step": 1117
    },
    {
      "epoch": 1.5880681818181817,
      "grad_norm": 0.3362851517367805,
      "learning_rate": 2.615789473684211e-05,
      "loss": 0.389,
      "step": 1118
    },
    {
      "epoch": 1.5894886363636362,
      "grad_norm": 0.29715541062048323,
      "learning_rate": 2.6131578947368424e-05,
      "loss": 0.3699,
      "step": 1119
    },
    {
      "epoch": 1.5909090909090908,
      "grad_norm": 0.35846787989591145,
      "learning_rate": 2.610526315789474e-05,
      "loss": 0.3701,
      "step": 1120
    },
    {
      "epoch": 1.5923295454545454,
      "grad_norm": 0.2824198105072625,
      "learning_rate": 2.6078947368421053e-05,
      "loss": 0.349,
      "step": 1121
    },
    {
      "epoch": 1.59375,
      "grad_norm": 0.3222729270129295,
      "learning_rate": 2.605263157894737e-05,
      "loss": 0.3628,
      "step": 1122
    },
    {
      "epoch": 1.5951704545454546,
      "grad_norm": 0.3240691831671821,
      "learning_rate": 2.6026315789473687e-05,
      "loss": 0.3622,
      "step": 1123
    },
    {
      "epoch": 1.5965909090909092,
      "grad_norm": 0.250995744771598,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 0.3496,
      "step": 1124
    },
    {
      "epoch": 1.5980113636363638,
      "grad_norm": 0.29834357134369394,
      "learning_rate": 2.5973684210526317e-05,
      "loss": 0.3563,
      "step": 1125
    },
    {
      "epoch": 1.5994318181818183,
      "grad_norm": 0.2698383268129392,
      "learning_rate": 2.5947368421052632e-05,
      "loss": 0.342,
      "step": 1126
    },
    {
      "epoch": 1.6008522727272727,
      "grad_norm": 0.2647739150153137,
      "learning_rate": 2.5921052631578947e-05,
      "loss": 0.3748,
      "step": 1127
    },
    {
      "epoch": 1.6022727272727273,
      "grad_norm": 0.2732191853913898,
      "learning_rate": 2.5894736842105265e-05,
      "loss": 0.3668,
      "step": 1128
    },
    {
      "epoch": 1.6036931818181817,
      "grad_norm": 0.2524700462534969,
      "learning_rate": 2.586842105263158e-05,
      "loss": 0.357,
      "step": 1129
    },
    {
      "epoch": 1.6051136363636362,
      "grad_norm": 0.2846468262710605,
      "learning_rate": 2.5842105263157895e-05,
      "loss": 0.363,
      "step": 1130
    },
    {
      "epoch": 1.6065340909090908,
      "grad_norm": 0.31359651566421837,
      "learning_rate": 2.581578947368421e-05,
      "loss": 0.3618,
      "step": 1131
    },
    {
      "epoch": 1.6079545454545454,
      "grad_norm": 0.2499269768767022,
      "learning_rate": 2.578947368421053e-05,
      "loss": 0.3639,
      "step": 1132
    },
    {
      "epoch": 1.609375,
      "grad_norm": 0.30729915841260635,
      "learning_rate": 2.5763157894736843e-05,
      "loss": 0.3668,
      "step": 1133
    },
    {
      "epoch": 1.6107954545454546,
      "grad_norm": 0.3234238397391346,
      "learning_rate": 2.5736842105263158e-05,
      "loss": 0.3645,
      "step": 1134
    },
    {
      "epoch": 1.6122159090909092,
      "grad_norm": 0.29752755940099845,
      "learning_rate": 2.5710526315789473e-05,
      "loss": 0.3641,
      "step": 1135
    },
    {
      "epoch": 1.6136363636363638,
      "grad_norm": 0.30128429830711073,
      "learning_rate": 2.5684210526315788e-05,
      "loss": 0.3728,
      "step": 1136
    },
    {
      "epoch": 1.6150568181818183,
      "grad_norm": 0.3242647325710467,
      "learning_rate": 2.565789473684211e-05,
      "loss": 0.3782,
      "step": 1137
    },
    {
      "epoch": 1.6164772727272727,
      "grad_norm": 0.26228509285272783,
      "learning_rate": 2.563157894736842e-05,
      "loss": 0.358,
      "step": 1138
    },
    {
      "epoch": 1.6178977272727273,
      "grad_norm": 0.3488259801917119,
      "learning_rate": 2.5605263157894737e-05,
      "loss": 0.3664,
      "step": 1139
    },
    {
      "epoch": 1.6193181818181817,
      "grad_norm": 0.29184007966157893,
      "learning_rate": 2.557894736842105e-05,
      "loss": 0.3712,
      "step": 1140
    },
    {
      "epoch": 1.6207386363636362,
      "grad_norm": 0.3062834658811172,
      "learning_rate": 2.5552631578947366e-05,
      "loss": 0.3787,
      "step": 1141
    },
    {
      "epoch": 1.6221590909090908,
      "grad_norm": 0.29984267829584216,
      "learning_rate": 2.5526315789473688e-05,
      "loss": 0.3613,
      "step": 1142
    },
    {
      "epoch": 1.6235795454545454,
      "grad_norm": 0.31119297154746356,
      "learning_rate": 2.5500000000000003e-05,
      "loss": 0.361,
      "step": 1143
    },
    {
      "epoch": 1.625,
      "grad_norm": 0.30482212481242277,
      "learning_rate": 2.5473684210526315e-05,
      "loss": 0.3543,
      "step": 1144
    },
    {
      "epoch": 1.6264204545454546,
      "grad_norm": 0.26009308004686127,
      "learning_rate": 2.544736842105263e-05,
      "loss": 0.3596,
      "step": 1145
    },
    {
      "epoch": 1.6278409090909092,
      "grad_norm": 0.3401718716673445,
      "learning_rate": 2.542105263157895e-05,
      "loss": 0.3544,
      "step": 1146
    },
    {
      "epoch": 1.6292613636363638,
      "grad_norm": 0.2762930487547228,
      "learning_rate": 2.5394736842105266e-05,
      "loss": 0.3661,
      "step": 1147
    },
    {
      "epoch": 1.6306818181818183,
      "grad_norm": 0.2660080772071548,
      "learning_rate": 2.536842105263158e-05,
      "loss": 0.3726,
      "step": 1148
    },
    {
      "epoch": 1.6321022727272727,
      "grad_norm": 0.3003767155336572,
      "learning_rate": 2.5342105263157896e-05,
      "loss": 0.3666,
      "step": 1149
    },
    {
      "epoch": 1.6335227272727273,
      "grad_norm": 0.2889796212388802,
      "learning_rate": 2.5315789473684208e-05,
      "loss": 0.376,
      "step": 1150
    },
    {
      "epoch": 1.6349431818181817,
      "grad_norm": 0.28263750007668553,
      "learning_rate": 2.528947368421053e-05,
      "loss": 0.372,
      "step": 1151
    },
    {
      "epoch": 1.6363636363636362,
      "grad_norm": 0.28023840925933563,
      "learning_rate": 2.5263157894736845e-05,
      "loss": 0.3795,
      "step": 1152
    },
    {
      "epoch": 1.6377840909090908,
      "grad_norm": 0.33168794222222087,
      "learning_rate": 2.523684210526316e-05,
      "loss": 0.3567,
      "step": 1153
    },
    {
      "epoch": 1.6392045454545454,
      "grad_norm": 0.3099648298562745,
      "learning_rate": 2.5210526315789475e-05,
      "loss": 0.3787,
      "step": 1154
    },
    {
      "epoch": 1.640625,
      "grad_norm": 0.33603166775108967,
      "learning_rate": 2.5184210526315786e-05,
      "loss": 0.3357,
      "step": 1155
    },
    {
      "epoch": 1.6420454545454546,
      "grad_norm": 0.31061233957968626,
      "learning_rate": 2.5157894736842108e-05,
      "loss": 0.3847,
      "step": 1156
    },
    {
      "epoch": 1.6434659090909092,
      "grad_norm": 0.36293220907805174,
      "learning_rate": 2.5131578947368423e-05,
      "loss": 0.377,
      "step": 1157
    },
    {
      "epoch": 1.6448863636363638,
      "grad_norm": 0.3716261177119713,
      "learning_rate": 2.5105263157894738e-05,
      "loss": 0.3824,
      "step": 1158
    },
    {
      "epoch": 1.6463068181818183,
      "grad_norm": 0.27017749568305693,
      "learning_rate": 2.5078947368421053e-05,
      "loss": 0.3584,
      "step": 1159
    },
    {
      "epoch": 1.6477272727272727,
      "grad_norm": 0.4503414409176396,
      "learning_rate": 2.505263157894737e-05,
      "loss": 0.3668,
      "step": 1160
    },
    {
      "epoch": 1.6491477272727273,
      "grad_norm": 0.3073232851867166,
      "learning_rate": 2.5026315789473686e-05,
      "loss": 0.3881,
      "step": 1161
    },
    {
      "epoch": 1.6505681818181817,
      "grad_norm": 0.3383517803616872,
      "learning_rate": 2.5e-05,
      "loss": 0.371,
      "step": 1162
    },
    {
      "epoch": 1.6519886363636362,
      "grad_norm": 0.33178230366898986,
      "learning_rate": 2.4973684210526316e-05,
      "loss": 0.3643,
      "step": 1163
    },
    {
      "epoch": 1.6534090909090908,
      "grad_norm": 0.2797479444538623,
      "learning_rate": 2.4947368421052635e-05,
      "loss": 0.3737,
      "step": 1164
    },
    {
      "epoch": 1.6548295454545454,
      "grad_norm": 0.3150589794150268,
      "learning_rate": 2.4921052631578946e-05,
      "loss": 0.365,
      "step": 1165
    },
    {
      "epoch": 1.65625,
      "grad_norm": 0.2983295313561275,
      "learning_rate": 2.4894736842105264e-05,
      "loss": 0.3505,
      "step": 1166
    },
    {
      "epoch": 1.6576704545454546,
      "grad_norm": 0.27130442142361855,
      "learning_rate": 2.486842105263158e-05,
      "loss": 0.3499,
      "step": 1167
    },
    {
      "epoch": 1.6590909090909092,
      "grad_norm": 0.31450669526506625,
      "learning_rate": 2.4842105263157898e-05,
      "loss": 0.3623,
      "step": 1168
    },
    {
      "epoch": 1.6605113636363638,
      "grad_norm": 0.2891186036190422,
      "learning_rate": 2.4815789473684213e-05,
      "loss": 0.3682,
      "step": 1169
    },
    {
      "epoch": 1.6619318181818183,
      "grad_norm": 0.2775556265735893,
      "learning_rate": 2.4789473684210528e-05,
      "loss": 0.3621,
      "step": 1170
    },
    {
      "epoch": 1.6633522727272727,
      "grad_norm": 0.24973347302139062,
      "learning_rate": 2.4763157894736843e-05,
      "loss": 0.3654,
      "step": 1171
    },
    {
      "epoch": 1.6647727272727273,
      "grad_norm": 0.2843619436127991,
      "learning_rate": 2.4736842105263158e-05,
      "loss": 0.3751,
      "step": 1172
    },
    {
      "epoch": 1.6661931818181817,
      "grad_norm": 0.2884834007988963,
      "learning_rate": 2.4710526315789476e-05,
      "loss": 0.3768,
      "step": 1173
    },
    {
      "epoch": 1.6676136363636362,
      "grad_norm": 0.2766378765947513,
      "learning_rate": 2.468421052631579e-05,
      "loss": 0.3711,
      "step": 1174
    },
    {
      "epoch": 1.6690340909090908,
      "grad_norm": 0.28462215566609006,
      "learning_rate": 2.4657894736842106e-05,
      "loss": 0.3531,
      "step": 1175
    },
    {
      "epoch": 1.6704545454545454,
      "grad_norm": 0.2629498269696211,
      "learning_rate": 2.463157894736842e-05,
      "loss": 0.3539,
      "step": 1176
    },
    {
      "epoch": 1.671875,
      "grad_norm": 0.2867199196082538,
      "learning_rate": 2.4605263157894736e-05,
      "loss": 0.3578,
      "step": 1177
    },
    {
      "epoch": 1.6732954545454546,
      "grad_norm": 0.2714819339240332,
      "learning_rate": 2.4578947368421054e-05,
      "loss": 0.3443,
      "step": 1178
    },
    {
      "epoch": 1.6747159090909092,
      "grad_norm": 0.27970231595705886,
      "learning_rate": 2.455263157894737e-05,
      "loss": 0.3584,
      "step": 1179
    },
    {
      "epoch": 1.6761363636363638,
      "grad_norm": 0.25026166975803066,
      "learning_rate": 2.4526315789473688e-05,
      "loss": 0.3548,
      "step": 1180
    },
    {
      "epoch": 1.6775568181818183,
      "grad_norm": 0.27067723873906674,
      "learning_rate": 2.45e-05,
      "loss": 0.3675,
      "step": 1181
    },
    {
      "epoch": 1.6789772727272727,
      "grad_norm": 0.2628348922105677,
      "learning_rate": 2.4473684210526318e-05,
      "loss": 0.3525,
      "step": 1182
    },
    {
      "epoch": 1.6803977272727273,
      "grad_norm": 0.29575148534633416,
      "learning_rate": 2.4447368421052633e-05,
      "loss": 0.3585,
      "step": 1183
    },
    {
      "epoch": 1.6818181818181817,
      "grad_norm": 0.26774981364642053,
      "learning_rate": 2.4421052631578948e-05,
      "loss": 0.3663,
      "step": 1184
    },
    {
      "epoch": 1.6832386363636362,
      "grad_norm": 0.26198273163665037,
      "learning_rate": 2.4394736842105266e-05,
      "loss": 0.3715,
      "step": 1185
    },
    {
      "epoch": 1.6846590909090908,
      "grad_norm": 0.29520245045233556,
      "learning_rate": 2.436842105263158e-05,
      "loss": 0.3842,
      "step": 1186
    },
    {
      "epoch": 1.6860795454545454,
      "grad_norm": 0.32865246871271486,
      "learning_rate": 2.4342105263157896e-05,
      "loss": 0.3843,
      "step": 1187
    },
    {
      "epoch": 1.6875,
      "grad_norm": 0.32881333517293615,
      "learning_rate": 2.431578947368421e-05,
      "loss": 0.3578,
      "step": 1188
    },
    {
      "epoch": 1.6889204545454546,
      "grad_norm": 0.2656440898328279,
      "learning_rate": 2.428947368421053e-05,
      "loss": 0.3541,
      "step": 1189
    },
    {
      "epoch": 1.6903409090909092,
      "grad_norm": 0.3099341781199934,
      "learning_rate": 2.4263157894736844e-05,
      "loss": 0.3696,
      "step": 1190
    },
    {
      "epoch": 1.6917613636363638,
      "grad_norm": 0.31910724295457016,
      "learning_rate": 2.423684210526316e-05,
      "loss": 0.3771,
      "step": 1191
    },
    {
      "epoch": 1.6931818181818183,
      "grad_norm": 0.30094484439509517,
      "learning_rate": 2.4210526315789474e-05,
      "loss": 0.3815,
      "step": 1192
    },
    {
      "epoch": 1.6946022727272727,
      "grad_norm": 0.36728193450644686,
      "learning_rate": 2.418421052631579e-05,
      "loss": 0.3806,
      "step": 1193
    },
    {
      "epoch": 1.6960227272727273,
      "grad_norm": 0.3524848081790139,
      "learning_rate": 2.4157894736842107e-05,
      "loss": 0.3552,
      "step": 1194
    },
    {
      "epoch": 1.6974431818181817,
      "grad_norm": 0.3328191880487227,
      "learning_rate": 2.4131578947368422e-05,
      "loss": 0.3739,
      "step": 1195
    },
    {
      "epoch": 1.6988636363636362,
      "grad_norm": 0.3860111360131643,
      "learning_rate": 2.410526315789474e-05,
      "loss": 0.3672,
      "step": 1196
    },
    {
      "epoch": 1.7002840909090908,
      "grad_norm": 0.32295089412925015,
      "learning_rate": 2.4078947368421052e-05,
      "loss": 0.3585,
      "step": 1197
    },
    {
      "epoch": 1.7017045454545454,
      "grad_norm": 0.340233809396054,
      "learning_rate": 2.4052631578947367e-05,
      "loss": 0.3508,
      "step": 1198
    },
    {
      "epoch": 1.703125,
      "grad_norm": 0.4145293985804659,
      "learning_rate": 2.4026315789473686e-05,
      "loss": 0.3573,
      "step": 1199
    },
    {
      "epoch": 1.7045454545454546,
      "grad_norm": 0.27269730923069774,
      "learning_rate": 2.4e-05,
      "loss": 0.3619,
      "step": 1200
    },
    {
      "epoch": 1.7059659090909092,
      "grad_norm": 0.34989782281608117,
      "learning_rate": 2.397368421052632e-05,
      "loss": 0.3922,
      "step": 1201
    },
    {
      "epoch": 1.7073863636363638,
      "grad_norm": 0.3243258166551283,
      "learning_rate": 2.394736842105263e-05,
      "loss": 0.3764,
      "step": 1202
    },
    {
      "epoch": 1.7088068181818183,
      "grad_norm": 0.2914646084452754,
      "learning_rate": 2.392105263157895e-05,
      "loss": 0.347,
      "step": 1203
    },
    {
      "epoch": 1.7102272727272727,
      "grad_norm": 0.3354533410686916,
      "learning_rate": 2.3894736842105264e-05,
      "loss": 0.3709,
      "step": 1204
    },
    {
      "epoch": 1.7116477272727273,
      "grad_norm": 0.27453394813206233,
      "learning_rate": 2.386842105263158e-05,
      "loss": 0.3701,
      "step": 1205
    },
    {
      "epoch": 1.7130681818181817,
      "grad_norm": 0.3139246274541037,
      "learning_rate": 2.3842105263157897e-05,
      "loss": 0.3767,
      "step": 1206
    },
    {
      "epoch": 1.7144886363636362,
      "grad_norm": 0.323253912656703,
      "learning_rate": 2.3815789473684212e-05,
      "loss": 0.3662,
      "step": 1207
    },
    {
      "epoch": 1.7159090909090908,
      "grad_norm": 0.3093060267722578,
      "learning_rate": 2.3789473684210527e-05,
      "loss": 0.357,
      "step": 1208
    },
    {
      "epoch": 1.7173295454545454,
      "grad_norm": 0.3250206972317296,
      "learning_rate": 2.3763157894736842e-05,
      "loss": 0.3659,
      "step": 1209
    },
    {
      "epoch": 1.71875,
      "grad_norm": 0.3298294358445142,
      "learning_rate": 2.373684210526316e-05,
      "loss": 0.3482,
      "step": 1210
    },
    {
      "epoch": 1.7201704545454546,
      "grad_norm": 0.27400773591190564,
      "learning_rate": 2.3710526315789475e-05,
      "loss": 0.3592,
      "step": 1211
    },
    {
      "epoch": 1.7215909090909092,
      "grad_norm": 0.3104943061270921,
      "learning_rate": 2.368421052631579e-05,
      "loss": 0.3573,
      "step": 1212
    },
    {
      "epoch": 1.7230113636363638,
      "grad_norm": 0.2791982419875152,
      "learning_rate": 2.3657894736842105e-05,
      "loss": 0.3679,
      "step": 1213
    },
    {
      "epoch": 1.7244318181818183,
      "grad_norm": 0.2860407872701481,
      "learning_rate": 2.363157894736842e-05,
      "loss": 0.3731,
      "step": 1214
    },
    {
      "epoch": 1.7258522727272727,
      "grad_norm": 0.2897355344087316,
      "learning_rate": 2.360526315789474e-05,
      "loss": 0.364,
      "step": 1215
    },
    {
      "epoch": 1.7272727272727273,
      "grad_norm": 0.28124092772975023,
      "learning_rate": 2.3578947368421054e-05,
      "loss": 0.3623,
      "step": 1216
    },
    {
      "epoch": 1.7286931818181817,
      "grad_norm": 0.25992770448430824,
      "learning_rate": 2.3552631578947372e-05,
      "loss": 0.3664,
      "step": 1217
    },
    {
      "epoch": 1.7301136363636362,
      "grad_norm": 0.28793440784502966,
      "learning_rate": 2.3526315789473684e-05,
      "loss": 0.3794,
      "step": 1218
    },
    {
      "epoch": 1.7315340909090908,
      "grad_norm": 0.2912592094502039,
      "learning_rate": 2.35e-05,
      "loss": 0.3813,
      "step": 1219
    },
    {
      "epoch": 1.7329545454545454,
      "grad_norm": 0.29533248145303703,
      "learning_rate": 2.3473684210526317e-05,
      "loss": 0.3856,
      "step": 1220
    },
    {
      "epoch": 1.734375,
      "grad_norm": 0.32862921891888114,
      "learning_rate": 2.3447368421052632e-05,
      "loss": 0.3697,
      "step": 1221
    },
    {
      "epoch": 1.7357954545454546,
      "grad_norm": 0.2845393445275072,
      "learning_rate": 2.342105263157895e-05,
      "loss": 0.3619,
      "step": 1222
    },
    {
      "epoch": 1.7372159090909092,
      "grad_norm": 0.3004432773883623,
      "learning_rate": 2.3394736842105265e-05,
      "loss": 0.3752,
      "step": 1223
    },
    {
      "epoch": 1.7386363636363638,
      "grad_norm": 0.34503915904576266,
      "learning_rate": 2.336842105263158e-05,
      "loss": 0.3713,
      "step": 1224
    },
    {
      "epoch": 1.7400568181818183,
      "grad_norm": 0.2661595184351093,
      "learning_rate": 2.3342105263157895e-05,
      "loss": 0.3603,
      "step": 1225
    },
    {
      "epoch": 1.7414772727272727,
      "grad_norm": 0.31610422161215895,
      "learning_rate": 2.331578947368421e-05,
      "loss": 0.3693,
      "step": 1226
    },
    {
      "epoch": 1.7428977272727273,
      "grad_norm": 0.283407639815075,
      "learning_rate": 2.328947368421053e-05,
      "loss": 0.3544,
      "step": 1227
    },
    {
      "epoch": 1.7443181818181817,
      "grad_norm": 0.26040524021809053,
      "learning_rate": 2.3263157894736844e-05,
      "loss": 0.3923,
      "step": 1228
    },
    {
      "epoch": 1.7457386363636362,
      "grad_norm": 0.26083208536489616,
      "learning_rate": 2.323684210526316e-05,
      "loss": 0.3657,
      "step": 1229
    },
    {
      "epoch": 1.7471590909090908,
      "grad_norm": 0.2808935083488795,
      "learning_rate": 2.3210526315789473e-05,
      "loss": 0.3835,
      "step": 1230
    },
    {
      "epoch": 1.7485795454545454,
      "grad_norm": 0.2813917962810276,
      "learning_rate": 2.3184210526315792e-05,
      "loss": 0.3685,
      "step": 1231
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.2529789777763934,
      "learning_rate": 2.3157894736842107e-05,
      "loss": 0.3635,
      "step": 1232
    },
    {
      "epoch": 1.7514204545454546,
      "grad_norm": 0.2806109845079464,
      "learning_rate": 2.3131578947368422e-05,
      "loss": 0.3693,
      "step": 1233
    },
    {
      "epoch": 1.7528409090909092,
      "grad_norm": 0.29488713869707206,
      "learning_rate": 2.3105263157894737e-05,
      "loss": 0.3626,
      "step": 1234
    },
    {
      "epoch": 1.7542613636363638,
      "grad_norm": 0.2852248793296293,
      "learning_rate": 2.3078947368421052e-05,
      "loss": 0.3718,
      "step": 1235
    },
    {
      "epoch": 1.7556818181818183,
      "grad_norm": 0.31590433143833335,
      "learning_rate": 2.305263157894737e-05,
      "loss": 0.3851,
      "step": 1236
    },
    {
      "epoch": 1.7571022727272727,
      "grad_norm": 0.633620354550528,
      "learning_rate": 2.3026315789473685e-05,
      "loss": 0.3917,
      "step": 1237
    },
    {
      "epoch": 1.7585227272727273,
      "grad_norm": 0.3050278566079899,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 0.3665,
      "step": 1238
    },
    {
      "epoch": 1.7599431818181817,
      "grad_norm": 0.3157146961064933,
      "learning_rate": 2.297368421052632e-05,
      "loss": 0.3557,
      "step": 1239
    },
    {
      "epoch": 1.7613636363636362,
      "grad_norm": 0.2948289555098577,
      "learning_rate": 2.294736842105263e-05,
      "loss": 0.3577,
      "step": 1240
    },
    {
      "epoch": 1.7627840909090908,
      "grad_norm": 0.29593170189717355,
      "learning_rate": 2.292105263157895e-05,
      "loss": 0.3681,
      "step": 1241
    },
    {
      "epoch": 1.7642045454545454,
      "grad_norm": 0.3305824118534329,
      "learning_rate": 2.2894736842105263e-05,
      "loss": 0.3644,
      "step": 1242
    },
    {
      "epoch": 1.765625,
      "grad_norm": 0.3265831007472785,
      "learning_rate": 2.286842105263158e-05,
      "loss": 0.3873,
      "step": 1243
    },
    {
      "epoch": 1.7670454545454546,
      "grad_norm": 0.28188365405842186,
      "learning_rate": 2.2842105263157897e-05,
      "loss": 0.3595,
      "step": 1244
    },
    {
      "epoch": 1.7684659090909092,
      "grad_norm": 0.2975033290273846,
      "learning_rate": 2.281578947368421e-05,
      "loss": 0.3482,
      "step": 1245
    },
    {
      "epoch": 1.7698863636363638,
      "grad_norm": 0.2613814853127822,
      "learning_rate": 2.2789473684210527e-05,
      "loss": 0.3601,
      "step": 1246
    },
    {
      "epoch": 1.7713068181818183,
      "grad_norm": 0.30788833858770565,
      "learning_rate": 2.2763157894736845e-05,
      "loss": 0.3539,
      "step": 1247
    },
    {
      "epoch": 1.7727272727272727,
      "grad_norm": 0.281387494821403,
      "learning_rate": 2.273684210526316e-05,
      "loss": 0.3586,
      "step": 1248
    },
    {
      "epoch": 1.7741477272727273,
      "grad_norm": 0.28051495577157687,
      "learning_rate": 2.2710526315789475e-05,
      "loss": 0.3668,
      "step": 1249
    },
    {
      "epoch": 1.7755681818181817,
      "grad_norm": 0.336624230649722,
      "learning_rate": 2.268421052631579e-05,
      "loss": 0.3672,
      "step": 1250
    },
    {
      "epoch": 1.7769886363636362,
      "grad_norm": 0.27488740765271047,
      "learning_rate": 2.2657894736842105e-05,
      "loss": 0.3701,
      "step": 1251
    },
    {
      "epoch": 1.7784090909090908,
      "grad_norm": 0.28208013358932416,
      "learning_rate": 2.2631578947368423e-05,
      "loss": 0.3654,
      "step": 1252
    },
    {
      "epoch": 1.7798295454545454,
      "grad_norm": 0.3165903842761538,
      "learning_rate": 2.2605263157894738e-05,
      "loss": 0.378,
      "step": 1253
    },
    {
      "epoch": 1.78125,
      "grad_norm": 0.2744131149494829,
      "learning_rate": 2.2578947368421057e-05,
      "loss": 0.3596,
      "step": 1254
    },
    {
      "epoch": 1.7826704545454546,
      "grad_norm": 0.3418566686973068,
      "learning_rate": 2.2552631578947368e-05,
      "loss": 0.3815,
      "step": 1255
    },
    {
      "epoch": 1.7840909090909092,
      "grad_norm": 0.33110875120748867,
      "learning_rate": 2.2526315789473683e-05,
      "loss": 0.3492,
      "step": 1256
    },
    {
      "epoch": 1.7855113636363638,
      "grad_norm": 0.2851770853165084,
      "learning_rate": 2.25e-05,
      "loss": 0.3638,
      "step": 1257
    },
    {
      "epoch": 1.7869318181818183,
      "grad_norm": 0.3506222047980432,
      "learning_rate": 2.2473684210526316e-05,
      "loss": 0.3699,
      "step": 1258
    },
    {
      "epoch": 1.7883522727272727,
      "grad_norm": 0.36161354824845904,
      "learning_rate": 2.2447368421052635e-05,
      "loss": 0.3592,
      "step": 1259
    },
    {
      "epoch": 1.7897727272727273,
      "grad_norm": 0.2815179331550175,
      "learning_rate": 2.242105263157895e-05,
      "loss": 0.3439,
      "step": 1260
    },
    {
      "epoch": 1.7911931818181817,
      "grad_norm": 0.33974203693689314,
      "learning_rate": 2.2394736842105265e-05,
      "loss": 0.3702,
      "step": 1261
    },
    {
      "epoch": 1.7926136363636362,
      "grad_norm": 0.32330804290169596,
      "learning_rate": 2.236842105263158e-05,
      "loss": 0.3641,
      "step": 1262
    },
    {
      "epoch": 1.7940340909090908,
      "grad_norm": 0.3130999033951169,
      "learning_rate": 2.2342105263157895e-05,
      "loss": 0.3673,
      "step": 1263
    },
    {
      "epoch": 1.7954545454545454,
      "grad_norm": 0.337289673788045,
      "learning_rate": 2.2315789473684213e-05,
      "loss": 0.3465,
      "step": 1264
    },
    {
      "epoch": 1.796875,
      "grad_norm": 0.36775306065936286,
      "learning_rate": 2.2289473684210528e-05,
      "loss": 0.3568,
      "step": 1265
    },
    {
      "epoch": 1.7982954545454546,
      "grad_norm": 0.27012201843219225,
      "learning_rate": 2.2263157894736843e-05,
      "loss": 0.3749,
      "step": 1266
    },
    {
      "epoch": 1.7997159090909092,
      "grad_norm": 0.34229553303178883,
      "learning_rate": 2.2236842105263158e-05,
      "loss": 0.3514,
      "step": 1267
    },
    {
      "epoch": 1.8011363636363638,
      "grad_norm": 0.3602202611469066,
      "learning_rate": 2.2210526315789476e-05,
      "loss": 0.3721,
      "step": 1268
    },
    {
      "epoch": 1.8025568181818183,
      "grad_norm": 0.3728123972057873,
      "learning_rate": 2.218421052631579e-05,
      "loss": 0.3736,
      "step": 1269
    },
    {
      "epoch": 1.8039772727272727,
      "grad_norm": 0.34602491268425845,
      "learning_rate": 2.2157894736842106e-05,
      "loss": 0.3563,
      "step": 1270
    },
    {
      "epoch": 1.8053977272727273,
      "grad_norm": 0.3209219880077489,
      "learning_rate": 2.213157894736842e-05,
      "loss": 0.3882,
      "step": 1271
    },
    {
      "epoch": 1.8068181818181817,
      "grad_norm": 0.39919647679805637,
      "learning_rate": 2.2105263157894736e-05,
      "loss": 0.3679,
      "step": 1272
    },
    {
      "epoch": 1.8082386363636362,
      "grad_norm": 0.2845764487109758,
      "learning_rate": 2.2078947368421055e-05,
      "loss": 0.361,
      "step": 1273
    },
    {
      "epoch": 1.8096590909090908,
      "grad_norm": 0.3549542168058561,
      "learning_rate": 2.205263157894737e-05,
      "loss": 0.3642,
      "step": 1274
    },
    {
      "epoch": 1.8110795454545454,
      "grad_norm": 0.3583420303771315,
      "learning_rate": 2.2026315789473688e-05,
      "loss": 0.3555,
      "step": 1275
    },
    {
      "epoch": 1.8125,
      "grad_norm": 0.3215056472241503,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 0.3608,
      "step": 1276
    },
    {
      "epoch": 1.8139204545454546,
      "grad_norm": 0.32671020951351654,
      "learning_rate": 2.1973684210526314e-05,
      "loss": 0.3611,
      "step": 1277
    },
    {
      "epoch": 1.8153409090909092,
      "grad_norm": 0.392808466472641,
      "learning_rate": 2.1947368421052633e-05,
      "loss": 0.3764,
      "step": 1278
    },
    {
      "epoch": 1.8167613636363638,
      "grad_norm": 0.27987650019513655,
      "learning_rate": 2.1921052631578948e-05,
      "loss": 0.3501,
      "step": 1279
    },
    {
      "epoch": 1.8181818181818183,
      "grad_norm": 0.27405788681497945,
      "learning_rate": 2.1894736842105266e-05,
      "loss": 0.3621,
      "step": 1280
    },
    {
      "epoch": 1.8196022727272727,
      "grad_norm": 0.369998981853915,
      "learning_rate": 2.186842105263158e-05,
      "loss": 0.3727,
      "step": 1281
    },
    {
      "epoch": 1.8210227272727273,
      "grad_norm": 0.29075833151878766,
      "learning_rate": 2.1842105263157896e-05,
      "loss": 0.3689,
      "step": 1282
    },
    {
      "epoch": 1.8224431818181817,
      "grad_norm": 0.28928754413850855,
      "learning_rate": 2.181578947368421e-05,
      "loss": 0.3799,
      "step": 1283
    },
    {
      "epoch": 1.8238636363636362,
      "grad_norm": 0.38362833771552324,
      "learning_rate": 2.1789473684210526e-05,
      "loss": 0.3598,
      "step": 1284
    },
    {
      "epoch": 1.8252840909090908,
      "grad_norm": 0.28644601252019747,
      "learning_rate": 2.1763157894736844e-05,
      "loss": 0.3673,
      "step": 1285
    },
    {
      "epoch": 1.8267045454545454,
      "grad_norm": 0.2621711716910221,
      "learning_rate": 2.173684210526316e-05,
      "loss": 0.3718,
      "step": 1286
    },
    {
      "epoch": 1.828125,
      "grad_norm": 0.32316361242141284,
      "learning_rate": 2.1710526315789474e-05,
      "loss": 0.3773,
      "step": 1287
    },
    {
      "epoch": 1.8295454545454546,
      "grad_norm": 0.3624046593062028,
      "learning_rate": 2.168421052631579e-05,
      "loss": 0.3554,
      "step": 1288
    },
    {
      "epoch": 1.8309659090909092,
      "grad_norm": 0.28155707483129044,
      "learning_rate": 2.1657894736842108e-05,
      "loss": 0.3781,
      "step": 1289
    },
    {
      "epoch": 1.8323863636363638,
      "grad_norm": 0.3314762650491284,
      "learning_rate": 2.1631578947368423e-05,
      "loss": 0.3684,
      "step": 1290
    },
    {
      "epoch": 1.8338068181818183,
      "grad_norm": 0.3872677497526659,
      "learning_rate": 2.1605263157894738e-05,
      "loss": 0.3801,
      "step": 1291
    },
    {
      "epoch": 1.8352272727272727,
      "grad_norm": 0.27933839116131964,
      "learning_rate": 2.1578947368421053e-05,
      "loss": 0.3639,
      "step": 1292
    },
    {
      "epoch": 1.8366477272727273,
      "grad_norm": 0.3073311874053648,
      "learning_rate": 2.1552631578947367e-05,
      "loss": 0.3692,
      "step": 1293
    },
    {
      "epoch": 1.8380681818181817,
      "grad_norm": 0.30468671468126585,
      "learning_rate": 2.1526315789473686e-05,
      "loss": 0.3529,
      "step": 1294
    },
    {
      "epoch": 1.8394886363636362,
      "grad_norm": 0.27544036270757216,
      "learning_rate": 2.15e-05,
      "loss": 0.3536,
      "step": 1295
    },
    {
      "epoch": 1.8409090909090908,
      "grad_norm": 0.28148477983610287,
      "learning_rate": 2.147368421052632e-05,
      "loss": 0.3464,
      "step": 1296
    },
    {
      "epoch": 1.8423295454545454,
      "grad_norm": 0.3247935003179314,
      "learning_rate": 2.1447368421052634e-05,
      "loss": 0.3494,
      "step": 1297
    },
    {
      "epoch": 1.84375,
      "grad_norm": 0.2940001452034238,
      "learning_rate": 2.1421052631578946e-05,
      "loss": 0.3657,
      "step": 1298
    },
    {
      "epoch": 1.8451704545454546,
      "grad_norm": 0.32058511317042987,
      "learning_rate": 2.1394736842105264e-05,
      "loss": 0.3564,
      "step": 1299
    },
    {
      "epoch": 1.8465909090909092,
      "grad_norm": 0.31683698180178743,
      "learning_rate": 2.136842105263158e-05,
      "loss": 0.3412,
      "step": 1300
    },
    {
      "epoch": 1.8480113636363638,
      "grad_norm": 0.33425251431346686,
      "learning_rate": 2.1342105263157897e-05,
      "loss": 0.3733,
      "step": 1301
    },
    {
      "epoch": 1.8494318181818183,
      "grad_norm": 0.3827872990200614,
      "learning_rate": 2.1315789473684212e-05,
      "loss": 0.3563,
      "step": 1302
    },
    {
      "epoch": 1.8508522727272727,
      "grad_norm": 0.2868025940939306,
      "learning_rate": 2.1289473684210527e-05,
      "loss": 0.3667,
      "step": 1303
    },
    {
      "epoch": 1.8522727272727273,
      "grad_norm": 0.39204144263360435,
      "learning_rate": 2.1263157894736842e-05,
      "loss": 0.3687,
      "step": 1304
    },
    {
      "epoch": 1.8536931818181817,
      "grad_norm": 0.3244806911311525,
      "learning_rate": 2.1236842105263157e-05,
      "loss": 0.3424,
      "step": 1305
    },
    {
      "epoch": 1.8551136363636362,
      "grad_norm": 0.319866337250371,
      "learning_rate": 2.1210526315789476e-05,
      "loss": 0.3671,
      "step": 1306
    },
    {
      "epoch": 1.8565340909090908,
      "grad_norm": 0.3479734431410179,
      "learning_rate": 2.118421052631579e-05,
      "loss": 0.3613,
      "step": 1307
    },
    {
      "epoch": 1.8579545454545454,
      "grad_norm": 0.26579455388919154,
      "learning_rate": 2.1157894736842106e-05,
      "loss": 0.3615,
      "step": 1308
    },
    {
      "epoch": 1.859375,
      "grad_norm": 0.24940907458347136,
      "learning_rate": 2.113157894736842e-05,
      "loss": 0.3524,
      "step": 1309
    },
    {
      "epoch": 1.8607954545454546,
      "grad_norm": 0.3097472590674542,
      "learning_rate": 2.110526315789474e-05,
      "loss": 0.3671,
      "step": 1310
    },
    {
      "epoch": 1.8622159090909092,
      "grad_norm": 0.2723921397466185,
      "learning_rate": 2.1078947368421054e-05,
      "loss": 0.3738,
      "step": 1311
    },
    {
      "epoch": 1.8636363636363638,
      "grad_norm": 0.27040036820342367,
      "learning_rate": 2.105263157894737e-05,
      "loss": 0.3576,
      "step": 1312
    },
    {
      "epoch": 1.8650568181818183,
      "grad_norm": 0.30348043631577204,
      "learning_rate": 2.1026315789473687e-05,
      "loss": 0.3625,
      "step": 1313
    },
    {
      "epoch": 1.8664772727272727,
      "grad_norm": 0.26162539400504625,
      "learning_rate": 2.1e-05,
      "loss": 0.3794,
      "step": 1314
    },
    {
      "epoch": 1.8678977272727273,
      "grad_norm": 0.23631766805660925,
      "learning_rate": 2.0973684210526317e-05,
      "loss": 0.329,
      "step": 1315
    },
    {
      "epoch": 1.8693181818181817,
      "grad_norm": 0.304385349440308,
      "learning_rate": 2.0947368421052632e-05,
      "loss": 0.3592,
      "step": 1316
    },
    {
      "epoch": 1.8707386363636362,
      "grad_norm": 0.28736532243693785,
      "learning_rate": 2.092105263157895e-05,
      "loss": 0.3654,
      "step": 1317
    },
    {
      "epoch": 1.8721590909090908,
      "grad_norm": 0.25261785589323876,
      "learning_rate": 2.0894736842105266e-05,
      "loss": 0.3709,
      "step": 1318
    },
    {
      "epoch": 1.8735795454545454,
      "grad_norm": 0.3336803830554354,
      "learning_rate": 2.0868421052631577e-05,
      "loss": 0.3682,
      "step": 1319
    },
    {
      "epoch": 1.875,
      "grad_norm": 0.25637349210868726,
      "learning_rate": 2.0842105263157895e-05,
      "loss": 0.3553,
      "step": 1320
    },
    {
      "epoch": 1.8764204545454546,
      "grad_norm": 0.2755978345072242,
      "learning_rate": 2.081578947368421e-05,
      "loss": 0.346,
      "step": 1321
    },
    {
      "epoch": 1.8778409090909092,
      "grad_norm": 0.28722882871151606,
      "learning_rate": 2.078947368421053e-05,
      "loss": 0.3607,
      "step": 1322
    },
    {
      "epoch": 1.8792613636363638,
      "grad_norm": 0.28277454461602275,
      "learning_rate": 2.0763157894736844e-05,
      "loss": 0.3647,
      "step": 1323
    },
    {
      "epoch": 1.8806818181818183,
      "grad_norm": 0.24443556516308831,
      "learning_rate": 2.073684210526316e-05,
      "loss": 0.3737,
      "step": 1324
    },
    {
      "epoch": 1.8821022727272727,
      "grad_norm": 0.36521744438662984,
      "learning_rate": 2.0710526315789474e-05,
      "loss": 0.3716,
      "step": 1325
    },
    {
      "epoch": 1.8835227272727273,
      "grad_norm": 0.2574638194764725,
      "learning_rate": 2.068421052631579e-05,
      "loss": 0.3727,
      "step": 1326
    },
    {
      "epoch": 1.8849431818181817,
      "grad_norm": 0.27869997928932905,
      "learning_rate": 2.0657894736842107e-05,
      "loss": 0.3493,
      "step": 1327
    },
    {
      "epoch": 1.8863636363636362,
      "grad_norm": 0.30473391518542503,
      "learning_rate": 2.0631578947368422e-05,
      "loss": 0.387,
      "step": 1328
    },
    {
      "epoch": 1.8877840909090908,
      "grad_norm": 0.2958203005480052,
      "learning_rate": 2.0605263157894737e-05,
      "loss": 0.3566,
      "step": 1329
    },
    {
      "epoch": 1.8892045454545454,
      "grad_norm": 0.30429732570356294,
      "learning_rate": 2.0578947368421052e-05,
      "loss": 0.3627,
      "step": 1330
    },
    {
      "epoch": 1.890625,
      "grad_norm": 0.265242098947472,
      "learning_rate": 2.055263157894737e-05,
      "loss": 0.3596,
      "step": 1331
    },
    {
      "epoch": 1.8920454545454546,
      "grad_norm": 0.3052183300720802,
      "learning_rate": 2.0526315789473685e-05,
      "loss": 0.3568,
      "step": 1332
    },
    {
      "epoch": 1.8934659090909092,
      "grad_norm": 0.3322802598193392,
      "learning_rate": 2.05e-05,
      "loss": 0.3613,
      "step": 1333
    },
    {
      "epoch": 1.8948863636363638,
      "grad_norm": 0.29963346583300426,
      "learning_rate": 2.047368421052632e-05,
      "loss": 0.3604,
      "step": 1334
    },
    {
      "epoch": 1.8963068181818183,
      "grad_norm": 0.29759148475446207,
      "learning_rate": 2.044736842105263e-05,
      "loss": 0.3692,
      "step": 1335
    },
    {
      "epoch": 1.8977272727272727,
      "grad_norm": 0.26763950963423877,
      "learning_rate": 2.042105263157895e-05,
      "loss": 0.3465,
      "step": 1336
    },
    {
      "epoch": 1.8991477272727273,
      "grad_norm": 0.2545902527757771,
      "learning_rate": 2.0394736842105264e-05,
      "loss": 0.3463,
      "step": 1337
    },
    {
      "epoch": 1.9005681818181817,
      "grad_norm": 0.2718227776715872,
      "learning_rate": 2.0368421052631582e-05,
      "loss": 0.3724,
      "step": 1338
    },
    {
      "epoch": 1.9019886363636362,
      "grad_norm": 0.2974973595011264,
      "learning_rate": 2.0342105263157897e-05,
      "loss": 0.369,
      "step": 1339
    },
    {
      "epoch": 1.9034090909090908,
      "grad_norm": 0.28746637764665695,
      "learning_rate": 2.0315789473684212e-05,
      "loss": 0.3542,
      "step": 1340
    },
    {
      "epoch": 1.9048295454545454,
      "grad_norm": 0.2733679489052221,
      "learning_rate": 2.0289473684210527e-05,
      "loss": 0.3673,
      "step": 1341
    },
    {
      "epoch": 1.90625,
      "grad_norm": 0.29666872623286217,
      "learning_rate": 2.0263157894736842e-05,
      "loss": 0.3682,
      "step": 1342
    },
    {
      "epoch": 1.9076704545454546,
      "grad_norm": 0.310214674547992,
      "learning_rate": 2.023684210526316e-05,
      "loss": 0.3766,
      "step": 1343
    },
    {
      "epoch": 1.9090909090909092,
      "grad_norm": 0.26344447631762813,
      "learning_rate": 2.0210526315789475e-05,
      "loss": 0.3652,
      "step": 1344
    },
    {
      "epoch": 1.9105113636363638,
      "grad_norm": 0.2912263222187671,
      "learning_rate": 2.018421052631579e-05,
      "loss": 0.3572,
      "step": 1345
    },
    {
      "epoch": 1.9119318181818183,
      "grad_norm": 0.32190678451932214,
      "learning_rate": 2.0157894736842105e-05,
      "loss": 0.3655,
      "step": 1346
    },
    {
      "epoch": 1.9133522727272727,
      "grad_norm": 0.32387790334768535,
      "learning_rate": 2.013157894736842e-05,
      "loss": 0.3513,
      "step": 1347
    },
    {
      "epoch": 1.9147727272727273,
      "grad_norm": 0.2983080839091067,
      "learning_rate": 2.010526315789474e-05,
      "loss": 0.3668,
      "step": 1348
    },
    {
      "epoch": 1.9161931818181817,
      "grad_norm": 0.27646639809707946,
      "learning_rate": 2.0078947368421053e-05,
      "loss": 0.3393,
      "step": 1349
    },
    {
      "epoch": 1.9176136363636362,
      "grad_norm": 0.3138496536128878,
      "learning_rate": 2.005263157894737e-05,
      "loss": 0.3399,
      "step": 1350
    },
    {
      "epoch": 1.9190340909090908,
      "grad_norm": 0.2982776830413004,
      "learning_rate": 2.0026315789473683e-05,
      "loss": 0.3397,
      "step": 1351
    },
    {
      "epoch": 1.9204545454545454,
      "grad_norm": 0.31017323454387924,
      "learning_rate": 2e-05,
      "loss": 0.336,
      "step": 1352
    },
    {
      "epoch": 1.921875,
      "grad_norm": 0.29537751806182855,
      "learning_rate": 1.9973684210526317e-05,
      "loss": 0.3567,
      "step": 1353
    },
    {
      "epoch": 1.9232954545454546,
      "grad_norm": 0.330226615521189,
      "learning_rate": 1.994736842105263e-05,
      "loss": 0.3748,
      "step": 1354
    },
    {
      "epoch": 1.9247159090909092,
      "grad_norm": 0.30458030977390477,
      "learning_rate": 1.992105263157895e-05,
      "loss": 0.3706,
      "step": 1355
    },
    {
      "epoch": 1.9261363636363638,
      "grad_norm": 0.27613019632581465,
      "learning_rate": 1.9894736842105265e-05,
      "loss": 0.3334,
      "step": 1356
    },
    {
      "epoch": 1.9275568181818183,
      "grad_norm": 0.30404851948697514,
      "learning_rate": 1.986842105263158e-05,
      "loss": 0.3755,
      "step": 1357
    },
    {
      "epoch": 1.9289772727272727,
      "grad_norm": 0.29044456318936895,
      "learning_rate": 1.9842105263157895e-05,
      "loss": 0.3732,
      "step": 1358
    },
    {
      "epoch": 1.9303977272727273,
      "grad_norm": 0.2569170082793818,
      "learning_rate": 1.9815789473684213e-05,
      "loss": 0.3596,
      "step": 1359
    },
    {
      "epoch": 1.9318181818181817,
      "grad_norm": 0.2580534281648247,
      "learning_rate": 1.9789473684210528e-05,
      "loss": 0.3733,
      "step": 1360
    },
    {
      "epoch": 1.9332386363636362,
      "grad_norm": 0.29218282330541706,
      "learning_rate": 1.9763157894736843e-05,
      "loss": 0.3495,
      "step": 1361
    },
    {
      "epoch": 1.9346590909090908,
      "grad_norm": 0.30055328501880374,
      "learning_rate": 1.9736842105263158e-05,
      "loss": 0.369,
      "step": 1362
    },
    {
      "epoch": 1.9360795454545454,
      "grad_norm": 0.2765214011648755,
      "learning_rate": 1.9710526315789473e-05,
      "loss": 0.3754,
      "step": 1363
    },
    {
      "epoch": 1.9375,
      "grad_norm": 0.29131301527464376,
      "learning_rate": 1.968421052631579e-05,
      "loss": 0.3584,
      "step": 1364
    },
    {
      "epoch": 1.9389204545454546,
      "grad_norm": 0.322143992025955,
      "learning_rate": 1.9657894736842106e-05,
      "loss": 0.3467,
      "step": 1365
    },
    {
      "epoch": 1.9403409090909092,
      "grad_norm": 0.27003890942123016,
      "learning_rate": 1.963157894736842e-05,
      "loss": 0.3769,
      "step": 1366
    },
    {
      "epoch": 1.9417613636363638,
      "grad_norm": 0.24664378355426983,
      "learning_rate": 1.9605263157894736e-05,
      "loss": 0.3633,
      "step": 1367
    },
    {
      "epoch": 1.9431818181818183,
      "grad_norm": 0.315539804601932,
      "learning_rate": 1.957894736842105e-05,
      "loss": 0.3669,
      "step": 1368
    },
    {
      "epoch": 1.9446022727272727,
      "grad_norm": 0.2917968110033054,
      "learning_rate": 1.955263157894737e-05,
      "loss": 0.3402,
      "step": 1369
    },
    {
      "epoch": 1.9460227272727273,
      "grad_norm": 0.2668816450648217,
      "learning_rate": 1.9526315789473685e-05,
      "loss": 0.3723,
      "step": 1370
    },
    {
      "epoch": 1.9474431818181817,
      "grad_norm": 0.3067650967669374,
      "learning_rate": 1.9500000000000003e-05,
      "loss": 0.3725,
      "step": 1371
    },
    {
      "epoch": 1.9488636363636362,
      "grad_norm": 0.3390334356081662,
      "learning_rate": 1.9473684210526315e-05,
      "loss": 0.3896,
      "step": 1372
    },
    {
      "epoch": 1.9502840909090908,
      "grad_norm": 0.270850952205965,
      "learning_rate": 1.9447368421052633e-05,
      "loss": 0.3489,
      "step": 1373
    },
    {
      "epoch": 1.9517045454545454,
      "grad_norm": 0.31206741215932726,
      "learning_rate": 1.9421052631578948e-05,
      "loss": 0.3594,
      "step": 1374
    },
    {
      "epoch": 1.953125,
      "grad_norm": 0.33401943524433414,
      "learning_rate": 1.9394736842105263e-05,
      "loss": 0.3779,
      "step": 1375
    },
    {
      "epoch": 1.9545454545454546,
      "grad_norm": 0.27682684295411325,
      "learning_rate": 1.936842105263158e-05,
      "loss": 0.3607,
      "step": 1376
    },
    {
      "epoch": 1.9559659090909092,
      "grad_norm": 0.27052761539150033,
      "learning_rate": 1.9342105263157896e-05,
      "loss": 0.3532,
      "step": 1377
    },
    {
      "epoch": 1.9573863636363638,
      "grad_norm": 0.3128821574169643,
      "learning_rate": 1.931578947368421e-05,
      "loss": 0.384,
      "step": 1378
    },
    {
      "epoch": 1.9588068181818183,
      "grad_norm": 0.27460369421024633,
      "learning_rate": 1.9289473684210526e-05,
      "loss": 0.3652,
      "step": 1379
    },
    {
      "epoch": 1.9602272727272727,
      "grad_norm": 0.2907011187939592,
      "learning_rate": 1.9263157894736845e-05,
      "loss": 0.3639,
      "step": 1380
    },
    {
      "epoch": 1.9616477272727273,
      "grad_norm": 0.26857996335362233,
      "learning_rate": 1.923684210526316e-05,
      "loss": 0.3793,
      "step": 1381
    },
    {
      "epoch": 1.9630681818181817,
      "grad_norm": 0.278552698652288,
      "learning_rate": 1.9210526315789474e-05,
      "loss": 0.37,
      "step": 1382
    },
    {
      "epoch": 1.9644886363636362,
      "grad_norm": 0.28436539378876097,
      "learning_rate": 1.918421052631579e-05,
      "loss": 0.351,
      "step": 1383
    },
    {
      "epoch": 1.9659090909090908,
      "grad_norm": 0.27218530747702524,
      "learning_rate": 1.9157894736842104e-05,
      "loss": 0.3658,
      "step": 1384
    },
    {
      "epoch": 1.9673295454545454,
      "grad_norm": 0.253228649247091,
      "learning_rate": 1.9131578947368423e-05,
      "loss": 0.3648,
      "step": 1385
    },
    {
      "epoch": 1.96875,
      "grad_norm": 0.26397788445138226,
      "learning_rate": 1.9105263157894738e-05,
      "loss": 0.3532,
      "step": 1386
    },
    {
      "epoch": 1.9701704545454546,
      "grad_norm": 0.2593035968365077,
      "learning_rate": 1.9078947368421056e-05,
      "loss": 0.3637,
      "step": 1387
    },
    {
      "epoch": 1.9715909090909092,
      "grad_norm": 0.2895846466619537,
      "learning_rate": 1.9052631578947368e-05,
      "loss": 0.3591,
      "step": 1388
    },
    {
      "epoch": 1.9730113636363638,
      "grad_norm": 0.2910661801861165,
      "learning_rate": 1.9026315789473683e-05,
      "loss": 0.3762,
      "step": 1389
    },
    {
      "epoch": 1.9744318181818183,
      "grad_norm": 0.2810442845225306,
      "learning_rate": 1.9e-05,
      "loss": 0.3671,
      "step": 1390
    },
    {
      "epoch": 1.9758522727272727,
      "grad_norm": 0.3046894417000149,
      "learning_rate": 1.8973684210526316e-05,
      "loss": 0.3492,
      "step": 1391
    },
    {
      "epoch": 1.9772727272727273,
      "grad_norm": 0.28061065669745083,
      "learning_rate": 1.8947368421052634e-05,
      "loss": 0.361,
      "step": 1392
    },
    {
      "epoch": 1.9786931818181817,
      "grad_norm": 0.2825768419854393,
      "learning_rate": 1.892105263157895e-05,
      "loss": 0.3672,
      "step": 1393
    },
    {
      "epoch": 1.9801136363636362,
      "grad_norm": 0.2841301540829614,
      "learning_rate": 1.8894736842105264e-05,
      "loss": 0.3671,
      "step": 1394
    },
    {
      "epoch": 1.9815340909090908,
      "grad_norm": 0.2720671382463663,
      "learning_rate": 1.886842105263158e-05,
      "loss": 0.3636,
      "step": 1395
    },
    {
      "epoch": 1.9829545454545454,
      "grad_norm": 0.3553566370613781,
      "learning_rate": 1.8842105263157894e-05,
      "loss": 0.3814,
      "step": 1396
    },
    {
      "epoch": 1.984375,
      "grad_norm": 0.24837846271344466,
      "learning_rate": 1.8815789473684213e-05,
      "loss": 0.3582,
      "step": 1397
    },
    {
      "epoch": 1.9857954545454546,
      "grad_norm": 0.26866776777843054,
      "learning_rate": 1.8789473684210528e-05,
      "loss": 0.3396,
      "step": 1398
    },
    {
      "epoch": 1.9872159090909092,
      "grad_norm": 0.2976367475947587,
      "learning_rate": 1.8763157894736843e-05,
      "loss": 0.3681,
      "step": 1399
    },
    {
      "epoch": 1.9886363636363638,
      "grad_norm": 0.2816633250295196,
      "learning_rate": 1.8736842105263158e-05,
      "loss": 0.3553,
      "step": 1400
    },
    {
      "epoch": 1.9900568181818183,
      "grad_norm": 0.2686753481202176,
      "learning_rate": 1.8710526315789476e-05,
      "loss": 0.3555,
      "step": 1401
    },
    {
      "epoch": 1.9914772727272727,
      "grad_norm": 0.27069130112423123,
      "learning_rate": 1.868421052631579e-05,
      "loss": 0.3597,
      "step": 1402
    },
    {
      "epoch": 1.9928977272727273,
      "grad_norm": 0.28972880470412266,
      "learning_rate": 1.865789473684211e-05,
      "loss": 0.3745,
      "step": 1403
    },
    {
      "epoch": 1.9943181818181817,
      "grad_norm": 0.3111044339546024,
      "learning_rate": 1.863157894736842e-05,
      "loss": 0.3716,
      "step": 1404
    },
    {
      "epoch": 1.9957386363636362,
      "grad_norm": 0.25567395053267405,
      "learning_rate": 1.8605263157894736e-05,
      "loss": 0.3656,
      "step": 1405
    },
    {
      "epoch": 1.9971590909090908,
      "grad_norm": 0.29627877733137137,
      "learning_rate": 1.8578947368421054e-05,
      "loss": 0.3526,
      "step": 1406
    },
    {
      "epoch": 1.9985795454545454,
      "grad_norm": 0.28223620595968213,
      "learning_rate": 1.855263157894737e-05,
      "loss": 0.3578,
      "step": 1407
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.232659328435687,
      "learning_rate": 1.8526315789473687e-05,
      "loss": 0.3257,
      "step": 1408
    },
    {
      "epoch": 2.0014204545454546,
      "grad_norm": 0.3120443755182193,
      "learning_rate": 1.85e-05,
      "loss": 0.2873,
      "step": 1409
    },
    {
      "epoch": 2.002840909090909,
      "grad_norm": 0.28501391374946594,
      "learning_rate": 1.8473684210526317e-05,
      "loss": 0.2948,
      "step": 1410
    },
    {
      "epoch": 2.0042613636363638,
      "grad_norm": 0.29780063015308383,
      "learning_rate": 1.8447368421052632e-05,
      "loss": 0.3057,
      "step": 1411
    },
    {
      "epoch": 2.0056818181818183,
      "grad_norm": 0.31961312673298764,
      "learning_rate": 1.8421052631578947e-05,
      "loss": 0.2951,
      "step": 1412
    },
    {
      "epoch": 2.007102272727273,
      "grad_norm": 0.2882420797738513,
      "learning_rate": 1.8394736842105266e-05,
      "loss": 0.2966,
      "step": 1413
    },
    {
      "epoch": 2.008522727272727,
      "grad_norm": 0.3505915486321542,
      "learning_rate": 1.836842105263158e-05,
      "loss": 0.284,
      "step": 1414
    },
    {
      "epoch": 2.0099431818181817,
      "grad_norm": 0.30307478064006055,
      "learning_rate": 1.8342105263157896e-05,
      "loss": 0.2789,
      "step": 1415
    },
    {
      "epoch": 2.0113636363636362,
      "grad_norm": 0.2623281529070418,
      "learning_rate": 1.831578947368421e-05,
      "loss": 0.2953,
      "step": 1416
    },
    {
      "epoch": 2.012784090909091,
      "grad_norm": 0.36596352446114366,
      "learning_rate": 1.828947368421053e-05,
      "loss": 0.2878,
      "step": 1417
    },
    {
      "epoch": 2.0142045454545454,
      "grad_norm": 0.3170240033311091,
      "learning_rate": 1.8263157894736844e-05,
      "loss": 0.2839,
      "step": 1418
    },
    {
      "epoch": 2.015625,
      "grad_norm": 0.2819379719495478,
      "learning_rate": 1.823684210526316e-05,
      "loss": 0.2879,
      "step": 1419
    },
    {
      "epoch": 2.0170454545454546,
      "grad_norm": 0.313435044077569,
      "learning_rate": 1.8210526315789474e-05,
      "loss": 0.3048,
      "step": 1420
    },
    {
      "epoch": 2.018465909090909,
      "grad_norm": 0.2758497138356,
      "learning_rate": 1.818421052631579e-05,
      "loss": 0.2877,
      "step": 1421
    },
    {
      "epoch": 2.0198863636363638,
      "grad_norm": 0.26236621082245026,
      "learning_rate": 1.8157894736842107e-05,
      "loss": 0.282,
      "step": 1422
    },
    {
      "epoch": 2.0213068181818183,
      "grad_norm": 0.28431462256558365,
      "learning_rate": 1.8131578947368422e-05,
      "loss": 0.2834,
      "step": 1423
    },
    {
      "epoch": 2.022727272727273,
      "grad_norm": 0.24973597088162158,
      "learning_rate": 1.810526315789474e-05,
      "loss": 0.2818,
      "step": 1424
    },
    {
      "epoch": 2.024147727272727,
      "grad_norm": 0.27717028595730236,
      "learning_rate": 1.8078947368421052e-05,
      "loss": 0.2853,
      "step": 1425
    },
    {
      "epoch": 2.0255681818181817,
      "grad_norm": 0.26364002552256055,
      "learning_rate": 1.8052631578947367e-05,
      "loss": 0.2812,
      "step": 1426
    },
    {
      "epoch": 2.0269886363636362,
      "grad_norm": 0.28499311370403463,
      "learning_rate": 1.8026315789473685e-05,
      "loss": 0.2927,
      "step": 1427
    },
    {
      "epoch": 2.028409090909091,
      "grad_norm": 0.24397427688872453,
      "learning_rate": 1.8e-05,
      "loss": 0.2827,
      "step": 1428
    },
    {
      "epoch": 2.0298295454545454,
      "grad_norm": 0.22644643731406855,
      "learning_rate": 1.797368421052632e-05,
      "loss": 0.271,
      "step": 1429
    },
    {
      "epoch": 2.03125,
      "grad_norm": 0.27021931802222415,
      "learning_rate": 1.7947368421052634e-05,
      "loss": 0.2872,
      "step": 1430
    },
    {
      "epoch": 2.0326704545454546,
      "grad_norm": 0.2576765545769877,
      "learning_rate": 1.792105263157895e-05,
      "loss": 0.2715,
      "step": 1431
    },
    {
      "epoch": 2.034090909090909,
      "grad_norm": 0.23088811057520783,
      "learning_rate": 1.7894736842105264e-05,
      "loss": 0.2791,
      "step": 1432
    },
    {
      "epoch": 2.0355113636363638,
      "grad_norm": 0.26505417488865496,
      "learning_rate": 1.786842105263158e-05,
      "loss": 0.2856,
      "step": 1433
    },
    {
      "epoch": 2.0369318181818183,
      "grad_norm": 0.26288365589716894,
      "learning_rate": 1.7842105263157897e-05,
      "loss": 0.2948,
      "step": 1434
    },
    {
      "epoch": 2.038352272727273,
      "grad_norm": 0.22430368603436548,
      "learning_rate": 1.7815789473684212e-05,
      "loss": 0.2708,
      "step": 1435
    },
    {
      "epoch": 2.039772727272727,
      "grad_norm": 0.2549499347377479,
      "learning_rate": 1.7789473684210527e-05,
      "loss": 0.2996,
      "step": 1436
    },
    {
      "epoch": 2.0411931818181817,
      "grad_norm": 0.23563593907613434,
      "learning_rate": 1.7763157894736842e-05,
      "loss": 0.2797,
      "step": 1437
    },
    {
      "epoch": 2.0426136363636362,
      "grad_norm": 0.26486203074961956,
      "learning_rate": 1.773684210526316e-05,
      "loss": 0.306,
      "step": 1438
    },
    {
      "epoch": 2.044034090909091,
      "grad_norm": 0.2361334542262243,
      "learning_rate": 1.7710526315789475e-05,
      "loss": 0.2818,
      "step": 1439
    },
    {
      "epoch": 2.0454545454545454,
      "grad_norm": 0.23851612378404186,
      "learning_rate": 1.768421052631579e-05,
      "loss": 0.2849,
      "step": 1440
    },
    {
      "epoch": 2.046875,
      "grad_norm": 0.22032919717793104,
      "learning_rate": 1.7657894736842105e-05,
      "loss": 0.2691,
      "step": 1441
    },
    {
      "epoch": 2.0482954545454546,
      "grad_norm": 0.2515207529439441,
      "learning_rate": 1.763157894736842e-05,
      "loss": 0.2826,
      "step": 1442
    },
    {
      "epoch": 2.049715909090909,
      "grad_norm": 0.23792271891595493,
      "learning_rate": 1.760526315789474e-05,
      "loss": 0.2888,
      "step": 1443
    },
    {
      "epoch": 2.0511363636363638,
      "grad_norm": 0.24592523569234054,
      "learning_rate": 1.7578947368421054e-05,
      "loss": 0.287,
      "step": 1444
    },
    {
      "epoch": 2.0525568181818183,
      "grad_norm": 0.2476476482850894,
      "learning_rate": 1.7552631578947372e-05,
      "loss": 0.2851,
      "step": 1445
    },
    {
      "epoch": 2.053977272727273,
      "grad_norm": 0.22954625096640308,
      "learning_rate": 1.7526315789473683e-05,
      "loss": 0.2873,
      "step": 1446
    },
    {
      "epoch": 2.055397727272727,
      "grad_norm": 0.2379466090845736,
      "learning_rate": 1.75e-05,
      "loss": 0.2985,
      "step": 1447
    },
    {
      "epoch": 2.0568181818181817,
      "grad_norm": 0.24694201011613198,
      "learning_rate": 1.7473684210526317e-05,
      "loss": 0.2853,
      "step": 1448
    },
    {
      "epoch": 2.0582386363636362,
      "grad_norm": 0.25775853291503065,
      "learning_rate": 1.7447368421052632e-05,
      "loss": 0.2853,
      "step": 1449
    },
    {
      "epoch": 2.059659090909091,
      "grad_norm": 0.25925290380325205,
      "learning_rate": 1.742105263157895e-05,
      "loss": 0.2889,
      "step": 1450
    },
    {
      "epoch": 2.0610795454545454,
      "grad_norm": 0.22487005103595803,
      "learning_rate": 1.7394736842105265e-05,
      "loss": 0.2803,
      "step": 1451
    },
    {
      "epoch": 2.0625,
      "grad_norm": 0.2613467706610791,
      "learning_rate": 1.736842105263158e-05,
      "loss": 0.2872,
      "step": 1452
    },
    {
      "epoch": 2.0639204545454546,
      "grad_norm": 0.2774351371587688,
      "learning_rate": 1.7342105263157895e-05,
      "loss": 0.2851,
      "step": 1453
    },
    {
      "epoch": 2.065340909090909,
      "grad_norm": 0.2687384177923622,
      "learning_rate": 1.731578947368421e-05,
      "loss": 0.2865,
      "step": 1454
    },
    {
      "epoch": 2.0667613636363638,
      "grad_norm": 0.24319514398334532,
      "learning_rate": 1.728947368421053e-05,
      "loss": 0.2809,
      "step": 1455
    },
    {
      "epoch": 2.0681818181818183,
      "grad_norm": 0.2887930352303555,
      "learning_rate": 1.7263157894736843e-05,
      "loss": 0.2894,
      "step": 1456
    },
    {
      "epoch": 2.069602272727273,
      "grad_norm": 0.26180284294745215,
      "learning_rate": 1.723684210526316e-05,
      "loss": 0.2864,
      "step": 1457
    },
    {
      "epoch": 2.071022727272727,
      "grad_norm": 0.23385579363078374,
      "learning_rate": 1.7210526315789473e-05,
      "loss": 0.2974,
      "step": 1458
    },
    {
      "epoch": 2.0724431818181817,
      "grad_norm": 0.24439802604942118,
      "learning_rate": 1.718421052631579e-05,
      "loss": 0.2745,
      "step": 1459
    },
    {
      "epoch": 2.0738636363636362,
      "grad_norm": 0.23692238676680666,
      "learning_rate": 1.7157894736842107e-05,
      "loss": 0.2872,
      "step": 1460
    },
    {
      "epoch": 2.075284090909091,
      "grad_norm": 0.23944151172707007,
      "learning_rate": 1.713157894736842e-05,
      "loss": 0.3009,
      "step": 1461
    },
    {
      "epoch": 2.0767045454545454,
      "grad_norm": 0.22256504077205944,
      "learning_rate": 1.7105263157894737e-05,
      "loss": 0.2952,
      "step": 1462
    },
    {
      "epoch": 2.078125,
      "grad_norm": 0.2385368149833647,
      "learning_rate": 1.707894736842105e-05,
      "loss": 0.2827,
      "step": 1463
    },
    {
      "epoch": 2.0795454545454546,
      "grad_norm": 0.2443257858999333,
      "learning_rate": 1.705263157894737e-05,
      "loss": 0.2892,
      "step": 1464
    },
    {
      "epoch": 2.080965909090909,
      "grad_norm": 0.23011318395317193,
      "learning_rate": 1.7026315789473685e-05,
      "loss": 0.2991,
      "step": 1465
    },
    {
      "epoch": 2.0823863636363638,
      "grad_norm": 0.23813946068338798,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 0.2884,
      "step": 1466
    },
    {
      "epoch": 2.0838068181818183,
      "grad_norm": 0.252201777214687,
      "learning_rate": 1.6973684210526318e-05,
      "loss": 0.2857,
      "step": 1467
    },
    {
      "epoch": 2.085227272727273,
      "grad_norm": 0.26097230333516325,
      "learning_rate": 1.694736842105263e-05,
      "loss": 0.2891,
      "step": 1468
    },
    {
      "epoch": 2.086647727272727,
      "grad_norm": 0.2524925390895555,
      "learning_rate": 1.6921052631578948e-05,
      "loss": 0.285,
      "step": 1469
    },
    {
      "epoch": 2.0880681818181817,
      "grad_norm": 0.23916686086485814,
      "learning_rate": 1.6894736842105263e-05,
      "loss": 0.2833,
      "step": 1470
    },
    {
      "epoch": 2.0894886363636362,
      "grad_norm": 0.2421747496250856,
      "learning_rate": 1.686842105263158e-05,
      "loss": 0.2975,
      "step": 1471
    },
    {
      "epoch": 2.090909090909091,
      "grad_norm": 0.27878266729348067,
      "learning_rate": 1.6842105263157896e-05,
      "loss": 0.3032,
      "step": 1472
    },
    {
      "epoch": 2.0923295454545454,
      "grad_norm": 0.2502056213855697,
      "learning_rate": 1.681578947368421e-05,
      "loss": 0.2869,
      "step": 1473
    },
    {
      "epoch": 2.09375,
      "grad_norm": 0.25040900946034195,
      "learning_rate": 1.6789473684210526e-05,
      "loss": 0.2917,
      "step": 1474
    },
    {
      "epoch": 2.0951704545454546,
      "grad_norm": 0.22100103310150218,
      "learning_rate": 1.676315789473684e-05,
      "loss": 0.283,
      "step": 1475
    },
    {
      "epoch": 2.096590909090909,
      "grad_norm": 0.3202756297814597,
      "learning_rate": 1.673684210526316e-05,
      "loss": 0.2894,
      "step": 1476
    },
    {
      "epoch": 2.0980113636363638,
      "grad_norm": 0.2613837810391721,
      "learning_rate": 1.6710526315789475e-05,
      "loss": 0.2809,
      "step": 1477
    },
    {
      "epoch": 2.0994318181818183,
      "grad_norm": 0.24201228249252657,
      "learning_rate": 1.668421052631579e-05,
      "loss": 0.3065,
      "step": 1478
    },
    {
      "epoch": 2.100852272727273,
      "grad_norm": 0.2820364574073259,
      "learning_rate": 1.6657894736842105e-05,
      "loss": 0.2818,
      "step": 1479
    },
    {
      "epoch": 2.102272727272727,
      "grad_norm": 0.261772650566443,
      "learning_rate": 1.6631578947368423e-05,
      "loss": 0.2724,
      "step": 1480
    },
    {
      "epoch": 2.1036931818181817,
      "grad_norm": 0.23526551520458738,
      "learning_rate": 1.6605263157894738e-05,
      "loss": 0.283,
      "step": 1481
    },
    {
      "epoch": 2.1051136363636362,
      "grad_norm": 0.23546372956729408,
      "learning_rate": 1.6578947368421053e-05,
      "loss": 0.2747,
      "step": 1482
    },
    {
      "epoch": 2.106534090909091,
      "grad_norm": 0.23479414384946068,
      "learning_rate": 1.6552631578947368e-05,
      "loss": 0.2767,
      "step": 1483
    },
    {
      "epoch": 2.1079545454545454,
      "grad_norm": 0.20656697351784273,
      "learning_rate": 1.6526315789473683e-05,
      "loss": 0.266,
      "step": 1484
    },
    {
      "epoch": 2.109375,
      "grad_norm": 0.21343309915793196,
      "learning_rate": 1.65e-05,
      "loss": 0.2832,
      "step": 1485
    },
    {
      "epoch": 2.1107954545454546,
      "grad_norm": 0.2269799753068697,
      "learning_rate": 1.6473684210526316e-05,
      "loss": 0.2945,
      "step": 1486
    },
    {
      "epoch": 2.112215909090909,
      "grad_norm": 0.22215049710985746,
      "learning_rate": 1.6447368421052635e-05,
      "loss": 0.2755,
      "step": 1487
    },
    {
      "epoch": 2.1136363636363638,
      "grad_norm": 0.22792780776742064,
      "learning_rate": 1.642105263157895e-05,
      "loss": 0.3038,
      "step": 1488
    },
    {
      "epoch": 2.1150568181818183,
      "grad_norm": 0.2020947819389658,
      "learning_rate": 1.639473684210526e-05,
      "loss": 0.2848,
      "step": 1489
    },
    {
      "epoch": 2.116477272727273,
      "grad_norm": 0.22376902452913108,
      "learning_rate": 1.636842105263158e-05,
      "loss": 0.2741,
      "step": 1490
    },
    {
      "epoch": 2.117897727272727,
      "grad_norm": 0.2251860456631497,
      "learning_rate": 1.6342105263157894e-05,
      "loss": 0.2937,
      "step": 1491
    },
    {
      "epoch": 2.1193181818181817,
      "grad_norm": 0.2269267380501234,
      "learning_rate": 1.6315789473684213e-05,
      "loss": 0.3016,
      "step": 1492
    },
    {
      "epoch": 2.1207386363636362,
      "grad_norm": 0.24574209523547919,
      "learning_rate": 1.6289473684210528e-05,
      "loss": 0.2892,
      "step": 1493
    },
    {
      "epoch": 2.122159090909091,
      "grad_norm": 0.22573990444219433,
      "learning_rate": 1.6263157894736843e-05,
      "loss": 0.2776,
      "step": 1494
    },
    {
      "epoch": 2.1235795454545454,
      "grad_norm": 0.22577810852720004,
      "learning_rate": 1.6236842105263158e-05,
      "loss": 0.2813,
      "step": 1495
    },
    {
      "epoch": 2.125,
      "grad_norm": 0.22918873199346568,
      "learning_rate": 1.6210526315789473e-05,
      "loss": 0.288,
      "step": 1496
    },
    {
      "epoch": 2.1264204545454546,
      "grad_norm": 0.26556878688116736,
      "learning_rate": 1.618421052631579e-05,
      "loss": 0.2948,
      "step": 1497
    },
    {
      "epoch": 2.127840909090909,
      "grad_norm": 0.22715677950285726,
      "learning_rate": 1.6157894736842106e-05,
      "loss": 0.2818,
      "step": 1498
    },
    {
      "epoch": 2.1292613636363638,
      "grad_norm": 0.22860438084264695,
      "learning_rate": 1.613157894736842e-05,
      "loss": 0.2911,
      "step": 1499
    },
    {
      "epoch": 2.1306818181818183,
      "grad_norm": 0.23571289949773766,
      "learning_rate": 1.6105263157894736e-05,
      "loss": 0.2925,
      "step": 1500
    },
    {
      "epoch": 2.132102272727273,
      "grad_norm": 0.24294369973261246,
      "learning_rate": 1.6078947368421054e-05,
      "loss": 0.3064,
      "step": 1501
    },
    {
      "epoch": 2.133522727272727,
      "grad_norm": 0.23901533469441244,
      "learning_rate": 1.605263157894737e-05,
      "loss": 0.2976,
      "step": 1502
    },
    {
      "epoch": 2.1349431818181817,
      "grad_norm": 0.2439662631029793,
      "learning_rate": 1.6026315789473684e-05,
      "loss": 0.2899,
      "step": 1503
    },
    {
      "epoch": 2.1363636363636362,
      "grad_norm": 0.21199748259466272,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.2839,
      "step": 1504
    },
    {
      "epoch": 2.137784090909091,
      "grad_norm": 0.2432918669719161,
      "learning_rate": 1.5973684210526314e-05,
      "loss": 0.3005,
      "step": 1505
    },
    {
      "epoch": 2.1392045454545454,
      "grad_norm": 0.23288870460407154,
      "learning_rate": 1.5947368421052633e-05,
      "loss": 0.278,
      "step": 1506
    },
    {
      "epoch": 2.140625,
      "grad_norm": 0.23982546811246674,
      "learning_rate": 1.5921052631578948e-05,
      "loss": 0.2885,
      "step": 1507
    },
    {
      "epoch": 2.1420454545454546,
      "grad_norm": 0.22446505832714106,
      "learning_rate": 1.5894736842105266e-05,
      "loss": 0.2924,
      "step": 1508
    },
    {
      "epoch": 2.143465909090909,
      "grad_norm": 0.2508715411182908,
      "learning_rate": 1.586842105263158e-05,
      "loss": 0.2938,
      "step": 1509
    },
    {
      "epoch": 2.1448863636363638,
      "grad_norm": 0.23430094034147878,
      "learning_rate": 1.5842105263157896e-05,
      "loss": 0.2826,
      "step": 1510
    },
    {
      "epoch": 2.1463068181818183,
      "grad_norm": 0.23286881527131448,
      "learning_rate": 1.581578947368421e-05,
      "loss": 0.2719,
      "step": 1511
    },
    {
      "epoch": 2.147727272727273,
      "grad_norm": 0.2590615532699332,
      "learning_rate": 1.5789473684210526e-05,
      "loss": 0.2853,
      "step": 1512
    },
    {
      "epoch": 2.149147727272727,
      "grad_norm": 0.25817252586976813,
      "learning_rate": 1.5763157894736844e-05,
      "loss": 0.2809,
      "step": 1513
    },
    {
      "epoch": 2.1505681818181817,
      "grad_norm": 0.2602157822049796,
      "learning_rate": 1.573684210526316e-05,
      "loss": 0.2944,
      "step": 1514
    },
    {
      "epoch": 2.1519886363636362,
      "grad_norm": 0.2381691324672256,
      "learning_rate": 1.5710526315789474e-05,
      "loss": 0.2908,
      "step": 1515
    },
    {
      "epoch": 2.153409090909091,
      "grad_norm": 0.2559737463641799,
      "learning_rate": 1.568421052631579e-05,
      "loss": 0.2841,
      "step": 1516
    },
    {
      "epoch": 2.1548295454545454,
      "grad_norm": 0.25355455932978727,
      "learning_rate": 1.5657894736842104e-05,
      "loss": 0.2823,
      "step": 1517
    },
    {
      "epoch": 2.15625,
      "grad_norm": 0.28293512799452464,
      "learning_rate": 1.5631578947368422e-05,
      "loss": 0.2985,
      "step": 1518
    },
    {
      "epoch": 2.1576704545454546,
      "grad_norm": 0.2501093746494721,
      "learning_rate": 1.5605263157894737e-05,
      "loss": 0.2879,
      "step": 1519
    },
    {
      "epoch": 2.159090909090909,
      "grad_norm": 0.23172286860835467,
      "learning_rate": 1.5578947368421056e-05,
      "loss": 0.277,
      "step": 1520
    },
    {
      "epoch": 2.1605113636363638,
      "grad_norm": 0.23522204518317769,
      "learning_rate": 1.5552631578947367e-05,
      "loss": 0.2787,
      "step": 1521
    },
    {
      "epoch": 2.1619318181818183,
      "grad_norm": 0.27166729450397503,
      "learning_rate": 1.5526315789473686e-05,
      "loss": 0.2975,
      "step": 1522
    },
    {
      "epoch": 2.163352272727273,
      "grad_norm": 0.31573885829872306,
      "learning_rate": 1.55e-05,
      "loss": 0.2953,
      "step": 1523
    },
    {
      "epoch": 2.164772727272727,
      "grad_norm": 0.2470600304983806,
      "learning_rate": 1.5473684210526316e-05,
      "loss": 0.2849,
      "step": 1524
    },
    {
      "epoch": 2.1661931818181817,
      "grad_norm": 0.23368224400902032,
      "learning_rate": 1.5447368421052634e-05,
      "loss": 0.2837,
      "step": 1525
    },
    {
      "epoch": 2.1676136363636362,
      "grad_norm": 0.30582140386415846,
      "learning_rate": 1.5421052631578946e-05,
      "loss": 0.2911,
      "step": 1526
    },
    {
      "epoch": 2.169034090909091,
      "grad_norm": 0.253082577710225,
      "learning_rate": 1.5394736842105264e-05,
      "loss": 0.2928,
      "step": 1527
    },
    {
      "epoch": 2.1704545454545454,
      "grad_norm": 0.2644666308298717,
      "learning_rate": 1.536842105263158e-05,
      "loss": 0.2835,
      "step": 1528
    },
    {
      "epoch": 2.171875,
      "grad_norm": 0.2581511397154595,
      "learning_rate": 1.5342105263157897e-05,
      "loss": 0.2866,
      "step": 1529
    },
    {
      "epoch": 2.1732954545454546,
      "grad_norm": 0.2711451636668079,
      "learning_rate": 1.5315789473684212e-05,
      "loss": 0.2897,
      "step": 1530
    },
    {
      "epoch": 2.174715909090909,
      "grad_norm": 0.2845489741357301,
      "learning_rate": 1.5289473684210527e-05,
      "loss": 0.2769,
      "step": 1531
    },
    {
      "epoch": 2.1761363636363638,
      "grad_norm": 0.23203023727583022,
      "learning_rate": 1.5263157894736842e-05,
      "loss": 0.2756,
      "step": 1532
    },
    {
      "epoch": 2.1775568181818183,
      "grad_norm": 0.2443672979085951,
      "learning_rate": 1.5236842105263157e-05,
      "loss": 0.2751,
      "step": 1533
    },
    {
      "epoch": 2.178977272727273,
      "grad_norm": 0.2210142070981885,
      "learning_rate": 1.5210526315789476e-05,
      "loss": 0.261,
      "step": 1534
    },
    {
      "epoch": 2.180397727272727,
      "grad_norm": 0.23386957588112053,
      "learning_rate": 1.518421052631579e-05,
      "loss": 0.2944,
      "step": 1535
    },
    {
      "epoch": 2.1818181818181817,
      "grad_norm": 0.23064613707928588,
      "learning_rate": 1.5157894736842107e-05,
      "loss": 0.2891,
      "step": 1536
    },
    {
      "epoch": 2.1832386363636362,
      "grad_norm": 0.22136725526698667,
      "learning_rate": 1.5131578947368422e-05,
      "loss": 0.2835,
      "step": 1537
    },
    {
      "epoch": 2.184659090909091,
      "grad_norm": 0.23168580265397518,
      "learning_rate": 1.5105263157894735e-05,
      "loss": 0.3096,
      "step": 1538
    },
    {
      "epoch": 2.1860795454545454,
      "grad_norm": 0.2483288810233192,
      "learning_rate": 1.5078947368421054e-05,
      "loss": 0.2797,
      "step": 1539
    },
    {
      "epoch": 2.1875,
      "grad_norm": 0.22784140440919892,
      "learning_rate": 1.5052631578947369e-05,
      "loss": 0.2857,
      "step": 1540
    },
    {
      "epoch": 2.1889204545454546,
      "grad_norm": 0.2451988527085582,
      "learning_rate": 1.5026315789473685e-05,
      "loss": 0.2885,
      "step": 1541
    },
    {
      "epoch": 2.190340909090909,
      "grad_norm": 0.22289717182008714,
      "learning_rate": 1.5e-05,
      "loss": 0.2868,
      "step": 1542
    },
    {
      "epoch": 2.1917613636363638,
      "grad_norm": 0.22641039240086985,
      "learning_rate": 1.4973684210526317e-05,
      "loss": 0.2807,
      "step": 1543
    },
    {
      "epoch": 2.1931818181818183,
      "grad_norm": 0.23738839874618722,
      "learning_rate": 1.4947368421052632e-05,
      "loss": 0.2823,
      "step": 1544
    },
    {
      "epoch": 2.194602272727273,
      "grad_norm": 0.2334387191416705,
      "learning_rate": 1.4921052631578947e-05,
      "loss": 0.2888,
      "step": 1545
    },
    {
      "epoch": 2.196022727272727,
      "grad_norm": 0.2452174842706825,
      "learning_rate": 1.4894736842105264e-05,
      "loss": 0.2867,
      "step": 1546
    },
    {
      "epoch": 2.1974431818181817,
      "grad_norm": 0.2560231750808039,
      "learning_rate": 1.4868421052631579e-05,
      "loss": 0.2979,
      "step": 1547
    },
    {
      "epoch": 2.1988636363636362,
      "grad_norm": 0.24731451316935132,
      "learning_rate": 1.4842105263157895e-05,
      "loss": 0.2783,
      "step": 1548
    },
    {
      "epoch": 2.200284090909091,
      "grad_norm": 0.2338856039331362,
      "learning_rate": 1.481578947368421e-05,
      "loss": 0.2815,
      "step": 1549
    },
    {
      "epoch": 2.2017045454545454,
      "grad_norm": 0.2748466316030393,
      "learning_rate": 1.4789473684210529e-05,
      "loss": 0.2861,
      "step": 1550
    },
    {
      "epoch": 2.203125,
      "grad_norm": 0.2267511916727485,
      "learning_rate": 1.4763157894736842e-05,
      "loss": 0.2916,
      "step": 1551
    },
    {
      "epoch": 2.2045454545454546,
      "grad_norm": 0.2453670916795553,
      "learning_rate": 1.4736842105263157e-05,
      "loss": 0.2893,
      "step": 1552
    },
    {
      "epoch": 2.205965909090909,
      "grad_norm": 0.2684447834411973,
      "learning_rate": 1.4710526315789475e-05,
      "loss": 0.2946,
      "step": 1553
    },
    {
      "epoch": 2.2073863636363638,
      "grad_norm": 0.2339994731401368,
      "learning_rate": 1.4684210526315789e-05,
      "loss": 0.2917,
      "step": 1554
    },
    {
      "epoch": 2.2088068181818183,
      "grad_norm": 0.24182519477311792,
      "learning_rate": 1.4657894736842107e-05,
      "loss": 0.295,
      "step": 1555
    },
    {
      "epoch": 2.210227272727273,
      "grad_norm": 0.23634355140877386,
      "learning_rate": 1.4631578947368422e-05,
      "loss": 0.3037,
      "step": 1556
    },
    {
      "epoch": 2.211647727272727,
      "grad_norm": 0.2376441033752254,
      "learning_rate": 1.4605263157894739e-05,
      "loss": 0.2994,
      "step": 1557
    },
    {
      "epoch": 2.2130681818181817,
      "grad_norm": 0.23592963679134032,
      "learning_rate": 1.4578947368421053e-05,
      "loss": 0.2855,
      "step": 1558
    },
    {
      "epoch": 2.2144886363636362,
      "grad_norm": 0.2389724296104485,
      "learning_rate": 1.455263157894737e-05,
      "loss": 0.2843,
      "step": 1559
    },
    {
      "epoch": 2.215909090909091,
      "grad_norm": 0.27273981594770386,
      "learning_rate": 1.4526315789473685e-05,
      "loss": 0.2861,
      "step": 1560
    },
    {
      "epoch": 2.2173295454545454,
      "grad_norm": 0.22140654674895618,
      "learning_rate": 1.45e-05,
      "loss": 0.283,
      "step": 1561
    },
    {
      "epoch": 2.21875,
      "grad_norm": 0.2198519538722085,
      "learning_rate": 1.4473684210526317e-05,
      "loss": 0.2777,
      "step": 1562
    },
    {
      "epoch": 2.2201704545454546,
      "grad_norm": 0.22875483718666167,
      "learning_rate": 1.4447368421052632e-05,
      "loss": 0.2796,
      "step": 1563
    },
    {
      "epoch": 2.221590909090909,
      "grad_norm": 0.22598234238739778,
      "learning_rate": 1.4421052631578948e-05,
      "loss": 0.2841,
      "step": 1564
    },
    {
      "epoch": 2.2230113636363638,
      "grad_norm": 0.24108112531392603,
      "learning_rate": 1.4394736842105263e-05,
      "loss": 0.2692,
      "step": 1565
    },
    {
      "epoch": 2.2244318181818183,
      "grad_norm": 0.2575862651190822,
      "learning_rate": 1.4368421052631582e-05,
      "loss": 0.2883,
      "step": 1566
    },
    {
      "epoch": 2.225852272727273,
      "grad_norm": 0.22093437898612514,
      "learning_rate": 1.4342105263157895e-05,
      "loss": 0.2775,
      "step": 1567
    },
    {
      "epoch": 2.227272727272727,
      "grad_norm": 0.22313682616965844,
      "learning_rate": 1.431578947368421e-05,
      "loss": 0.2922,
      "step": 1568
    },
    {
      "epoch": 2.2286931818181817,
      "grad_norm": 0.2288516580032956,
      "learning_rate": 1.4289473684210528e-05,
      "loss": 0.2811,
      "step": 1569
    },
    {
      "epoch": 2.2301136363636362,
      "grad_norm": 0.28738237161590646,
      "learning_rate": 1.4263157894736842e-05,
      "loss": 0.2853,
      "step": 1570
    },
    {
      "epoch": 2.231534090909091,
      "grad_norm": 0.22401733700534535,
      "learning_rate": 1.423684210526316e-05,
      "loss": 0.2849,
      "step": 1571
    },
    {
      "epoch": 2.2329545454545454,
      "grad_norm": 0.22434337476064423,
      "learning_rate": 1.4210526315789475e-05,
      "loss": 0.2903,
      "step": 1572
    },
    {
      "epoch": 2.234375,
      "grad_norm": 0.26555314438354044,
      "learning_rate": 1.4184210526315792e-05,
      "loss": 0.285,
      "step": 1573
    },
    {
      "epoch": 2.2357954545454546,
      "grad_norm": 0.2369278315997259,
      "learning_rate": 1.4157894736842107e-05,
      "loss": 0.2778,
      "step": 1574
    },
    {
      "epoch": 2.237215909090909,
      "grad_norm": 0.2238885475545879,
      "learning_rate": 1.413157894736842e-05,
      "loss": 0.2896,
      "step": 1575
    },
    {
      "epoch": 2.2386363636363638,
      "grad_norm": 0.213942081614159,
      "learning_rate": 1.4105263157894738e-05,
      "loss": 0.2786,
      "step": 1576
    },
    {
      "epoch": 2.2400568181818183,
      "grad_norm": 0.28446277815023774,
      "learning_rate": 1.4078947368421053e-05,
      "loss": 0.3083,
      "step": 1577
    },
    {
      "epoch": 2.241477272727273,
      "grad_norm": 0.22044849490310245,
      "learning_rate": 1.405263157894737e-05,
      "loss": 0.2775,
      "step": 1578
    },
    {
      "epoch": 2.242897727272727,
      "grad_norm": 0.22818124687624167,
      "learning_rate": 1.4026315789473685e-05,
      "loss": 0.2912,
      "step": 1579
    },
    {
      "epoch": 2.2443181818181817,
      "grad_norm": 0.20685825799610558,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 0.2728,
      "step": 1580
    },
    {
      "epoch": 2.2457386363636362,
      "grad_norm": 0.22661531654487038,
      "learning_rate": 1.3973684210526316e-05,
      "loss": 0.2809,
      "step": 1581
    },
    {
      "epoch": 2.247159090909091,
      "grad_norm": 0.23546429525222623,
      "learning_rate": 1.3947368421052631e-05,
      "loss": 0.2896,
      "step": 1582
    },
    {
      "epoch": 2.2485795454545454,
      "grad_norm": 0.2694861575150227,
      "learning_rate": 1.3921052631578948e-05,
      "loss": 0.2977,
      "step": 1583
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.22770118026087577,
      "learning_rate": 1.3894736842105263e-05,
      "loss": 0.2905,
      "step": 1584
    },
    {
      "epoch": 2.2514204545454546,
      "grad_norm": 0.2413666368950816,
      "learning_rate": 1.386842105263158e-05,
      "loss": 0.2899,
      "step": 1585
    },
    {
      "epoch": 2.252840909090909,
      "grad_norm": 0.2604298001736635,
      "learning_rate": 1.3842105263157895e-05,
      "loss": 0.2942,
      "step": 1586
    },
    {
      "epoch": 2.2542613636363638,
      "grad_norm": 0.22141055765318865,
      "learning_rate": 1.3815789473684213e-05,
      "loss": 0.2857,
      "step": 1587
    },
    {
      "epoch": 2.2556818181818183,
      "grad_norm": 0.23218029978137064,
      "learning_rate": 1.3789473684210526e-05,
      "loss": 0.3052,
      "step": 1588
    },
    {
      "epoch": 2.2571022727272725,
      "grad_norm": 0.24027426902189764,
      "learning_rate": 1.3763157894736841e-05,
      "loss": 0.2791,
      "step": 1589
    },
    {
      "epoch": 2.2585227272727275,
      "grad_norm": 0.23150094844624486,
      "learning_rate": 1.373684210526316e-05,
      "loss": 0.2993,
      "step": 1590
    },
    {
      "epoch": 2.2599431818181817,
      "grad_norm": 0.22682638728567178,
      "learning_rate": 1.3710526315789473e-05,
      "loss": 0.2764,
      "step": 1591
    },
    {
      "epoch": 2.2613636363636362,
      "grad_norm": 0.21283772119102354,
      "learning_rate": 1.3684210526315791e-05,
      "loss": 0.3034,
      "step": 1592
    },
    {
      "epoch": 2.262784090909091,
      "grad_norm": 0.24637506860442143,
      "learning_rate": 1.3657894736842106e-05,
      "loss": 0.2851,
      "step": 1593
    },
    {
      "epoch": 2.2642045454545454,
      "grad_norm": 0.22090740451833357,
      "learning_rate": 1.3631578947368423e-05,
      "loss": 0.2676,
      "step": 1594
    },
    {
      "epoch": 2.265625,
      "grad_norm": 0.21997821145409766,
      "learning_rate": 1.3605263157894738e-05,
      "loss": 0.2879,
      "step": 1595
    },
    {
      "epoch": 2.2670454545454546,
      "grad_norm": 0.22531915841804187,
      "learning_rate": 1.3578947368421053e-05,
      "loss": 0.2848,
      "step": 1596
    },
    {
      "epoch": 2.268465909090909,
      "grad_norm": 0.2137337264310719,
      "learning_rate": 1.355263157894737e-05,
      "loss": 0.2736,
      "step": 1597
    },
    {
      "epoch": 2.2698863636363638,
      "grad_norm": 0.213039370823097,
      "learning_rate": 1.3526315789473685e-05,
      "loss": 0.2869,
      "step": 1598
    },
    {
      "epoch": 2.2713068181818183,
      "grad_norm": 0.237367839334122,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 0.2954,
      "step": 1599
    },
    {
      "epoch": 2.2727272727272725,
      "grad_norm": 0.23341200266675044,
      "learning_rate": 1.3473684210526316e-05,
      "loss": 0.2821,
      "step": 1600
    },
    {
      "epoch": 2.2741477272727275,
      "grad_norm": 0.2029190785390699,
      "learning_rate": 1.3447368421052633e-05,
      "loss": 0.2948,
      "step": 1601
    },
    {
      "epoch": 2.2755681818181817,
      "grad_norm": 0.20760619372254446,
      "learning_rate": 1.3421052631578948e-05,
      "loss": 0.2673,
      "step": 1602
    },
    {
      "epoch": 2.2769886363636362,
      "grad_norm": 0.2300779538967087,
      "learning_rate": 1.3394736842105263e-05,
      "loss": 0.299,
      "step": 1603
    },
    {
      "epoch": 2.278409090909091,
      "grad_norm": 0.21822264791486812,
      "learning_rate": 1.336842105263158e-05,
      "loss": 0.2769,
      "step": 1604
    },
    {
      "epoch": 2.2798295454545454,
      "grad_norm": 0.22096390594212273,
      "learning_rate": 1.3342105263157894e-05,
      "loss": 0.2949,
      "step": 1605
    },
    {
      "epoch": 2.28125,
      "grad_norm": 0.26018905595403813,
      "learning_rate": 1.3315789473684213e-05,
      "loss": 0.2947,
      "step": 1606
    },
    {
      "epoch": 2.2826704545454546,
      "grad_norm": 0.23872146662899724,
      "learning_rate": 1.3289473684210526e-05,
      "loss": 0.2943,
      "step": 1607
    },
    {
      "epoch": 2.284090909090909,
      "grad_norm": 0.23628674962492488,
      "learning_rate": 1.3263157894736844e-05,
      "loss": 0.2931,
      "step": 1608
    },
    {
      "epoch": 2.2855113636363638,
      "grad_norm": 0.2356853138745016,
      "learning_rate": 1.323684210526316e-05,
      "loss": 0.2871,
      "step": 1609
    },
    {
      "epoch": 2.2869318181818183,
      "grad_norm": 0.23301524013415034,
      "learning_rate": 1.3210526315789473e-05,
      "loss": 0.2786,
      "step": 1610
    },
    {
      "epoch": 2.2883522727272725,
      "grad_norm": 0.2226143240536239,
      "learning_rate": 1.3184210526315791e-05,
      "loss": 0.3059,
      "step": 1611
    },
    {
      "epoch": 2.2897727272727275,
      "grad_norm": 0.2518870070302093,
      "learning_rate": 1.3157894736842106e-05,
      "loss": 0.3045,
      "step": 1612
    },
    {
      "epoch": 2.2911931818181817,
      "grad_norm": 0.2383871763340484,
      "learning_rate": 1.3131578947368423e-05,
      "loss": 0.2886,
      "step": 1613
    },
    {
      "epoch": 2.2926136363636362,
      "grad_norm": 0.23604585828322636,
      "learning_rate": 1.3105263157894738e-05,
      "loss": 0.2939,
      "step": 1614
    },
    {
      "epoch": 2.294034090909091,
      "grad_norm": 0.263403276516726,
      "learning_rate": 1.3078947368421054e-05,
      "loss": 0.2869,
      "step": 1615
    },
    {
      "epoch": 2.2954545454545454,
      "grad_norm": 0.23986627365613367,
      "learning_rate": 1.305263157894737e-05,
      "loss": 0.2914,
      "step": 1616
    },
    {
      "epoch": 2.296875,
      "grad_norm": 0.24876820200026617,
      "learning_rate": 1.3026315789473684e-05,
      "loss": 0.296,
      "step": 1617
    },
    {
      "epoch": 2.2982954545454546,
      "grad_norm": 0.2278230952226003,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.279,
      "step": 1618
    },
    {
      "epoch": 2.299715909090909,
      "grad_norm": 0.21475720512060384,
      "learning_rate": 1.2973684210526316e-05,
      "loss": 0.308,
      "step": 1619
    },
    {
      "epoch": 2.3011363636363638,
      "grad_norm": 0.2311227890691602,
      "learning_rate": 1.2947368421052633e-05,
      "loss": 0.3027,
      "step": 1620
    },
    {
      "epoch": 2.3025568181818183,
      "grad_norm": 0.21764275214799442,
      "learning_rate": 1.2921052631578948e-05,
      "loss": 0.2792,
      "step": 1621
    },
    {
      "epoch": 2.3039772727272725,
      "grad_norm": 0.2355648931256157,
      "learning_rate": 1.2894736842105264e-05,
      "loss": 0.283,
      "step": 1622
    },
    {
      "epoch": 2.3053977272727275,
      "grad_norm": 0.2351880585642382,
      "learning_rate": 1.2868421052631579e-05,
      "loss": 0.2835,
      "step": 1623
    },
    {
      "epoch": 2.3068181818181817,
      "grad_norm": 0.2326034898688473,
      "learning_rate": 1.2842105263157894e-05,
      "loss": 0.2871,
      "step": 1624
    },
    {
      "epoch": 2.3082386363636362,
      "grad_norm": 0.22461933039147175,
      "learning_rate": 1.281578947368421e-05,
      "loss": 0.2911,
      "step": 1625
    },
    {
      "epoch": 2.309659090909091,
      "grad_norm": 0.21324590454936107,
      "learning_rate": 1.2789473684210526e-05,
      "loss": 0.2732,
      "step": 1626
    },
    {
      "epoch": 2.3110795454545454,
      "grad_norm": 0.24548702608294015,
      "learning_rate": 1.2763157894736844e-05,
      "loss": 0.2988,
      "step": 1627
    },
    {
      "epoch": 2.3125,
      "grad_norm": 0.2192635219712895,
      "learning_rate": 1.2736842105263157e-05,
      "loss": 0.2819,
      "step": 1628
    },
    {
      "epoch": 2.3139204545454546,
      "grad_norm": 0.2564694601793526,
      "learning_rate": 1.2710526315789476e-05,
      "loss": 0.2992,
      "step": 1629
    },
    {
      "epoch": 2.315340909090909,
      "grad_norm": 0.22722038629394534,
      "learning_rate": 1.268421052631579e-05,
      "loss": 0.2759,
      "step": 1630
    },
    {
      "epoch": 2.3167613636363638,
      "grad_norm": 0.21178518771968194,
      "learning_rate": 1.2657894736842104e-05,
      "loss": 0.2852,
      "step": 1631
    },
    {
      "epoch": 2.3181818181818183,
      "grad_norm": 0.2336697946412485,
      "learning_rate": 1.2631578947368422e-05,
      "loss": 0.2874,
      "step": 1632
    },
    {
      "epoch": 2.3196022727272725,
      "grad_norm": 0.24814901762513306,
      "learning_rate": 1.2605263157894737e-05,
      "loss": 0.2881,
      "step": 1633
    },
    {
      "epoch": 2.3210227272727275,
      "grad_norm": 0.24278420734025166,
      "learning_rate": 1.2578947368421054e-05,
      "loss": 0.2872,
      "step": 1634
    },
    {
      "epoch": 2.3224431818181817,
      "grad_norm": 0.2190665634531621,
      "learning_rate": 1.2552631578947369e-05,
      "loss": 0.2841,
      "step": 1635
    },
    {
      "epoch": 2.3238636363636362,
      "grad_norm": 0.25618294249185386,
      "learning_rate": 1.2526315789473686e-05,
      "loss": 0.3039,
      "step": 1636
    },
    {
      "epoch": 2.325284090909091,
      "grad_norm": 0.24367318996796583,
      "learning_rate": 1.25e-05,
      "loss": 0.2951,
      "step": 1637
    },
    {
      "epoch": 2.3267045454545454,
      "grad_norm": 0.2408770619579771,
      "learning_rate": 1.2473684210526317e-05,
      "loss": 0.2942,
      "step": 1638
    },
    {
      "epoch": 2.328125,
      "grad_norm": 0.22116109540640183,
      "learning_rate": 1.2447368421052632e-05,
      "loss": 0.2753,
      "step": 1639
    },
    {
      "epoch": 2.3295454545454546,
      "grad_norm": 0.23096700974404852,
      "learning_rate": 1.2421052631578949e-05,
      "loss": 0.2716,
      "step": 1640
    },
    {
      "epoch": 2.330965909090909,
      "grad_norm": 0.240013318601306,
      "learning_rate": 1.2394736842105264e-05,
      "loss": 0.309,
      "step": 1641
    },
    {
      "epoch": 2.3323863636363638,
      "grad_norm": 0.21995605825605116,
      "learning_rate": 1.2368421052631579e-05,
      "loss": 0.2816,
      "step": 1642
    },
    {
      "epoch": 2.3338068181818183,
      "grad_norm": 0.22641724879189148,
      "learning_rate": 1.2342105263157896e-05,
      "loss": 0.2893,
      "step": 1643
    },
    {
      "epoch": 2.3352272727272725,
      "grad_norm": 0.2678591132211738,
      "learning_rate": 1.231578947368421e-05,
      "loss": 0.3121,
      "step": 1644
    },
    {
      "epoch": 2.3366477272727275,
      "grad_norm": 0.22364637817544958,
      "learning_rate": 1.2289473684210527e-05,
      "loss": 0.2916,
      "step": 1645
    },
    {
      "epoch": 2.3380681818181817,
      "grad_norm": 0.21294285738572732,
      "learning_rate": 1.2263157894736844e-05,
      "loss": 0.2859,
      "step": 1646
    },
    {
      "epoch": 2.3394886363636362,
      "grad_norm": 0.24317490380092593,
      "learning_rate": 1.2236842105263159e-05,
      "loss": 0.2654,
      "step": 1647
    },
    {
      "epoch": 2.340909090909091,
      "grad_norm": 0.24245693164186477,
      "learning_rate": 1.2210526315789474e-05,
      "loss": 0.2935,
      "step": 1648
    },
    {
      "epoch": 2.3423295454545454,
      "grad_norm": 0.21962445089567,
      "learning_rate": 1.218421052631579e-05,
      "loss": 0.273,
      "step": 1649
    },
    {
      "epoch": 2.34375,
      "grad_norm": 0.2500843365743065,
      "learning_rate": 1.2157894736842105e-05,
      "loss": 0.2921,
      "step": 1650
    },
    {
      "epoch": 2.3451704545454546,
      "grad_norm": 0.25961531470998334,
      "learning_rate": 1.2131578947368422e-05,
      "loss": 0.285,
      "step": 1651
    },
    {
      "epoch": 2.346590909090909,
      "grad_norm": 0.24654061965664528,
      "learning_rate": 1.2105263157894737e-05,
      "loss": 0.2935,
      "step": 1652
    },
    {
      "epoch": 2.3480113636363638,
      "grad_norm": 0.21051926532350038,
      "learning_rate": 1.2078947368421054e-05,
      "loss": 0.2769,
      "step": 1653
    },
    {
      "epoch": 2.3494318181818183,
      "grad_norm": 0.25964967562469465,
      "learning_rate": 1.205263157894737e-05,
      "loss": 0.2904,
      "step": 1654
    },
    {
      "epoch": 2.3508522727272725,
      "grad_norm": 0.22566126821950916,
      "learning_rate": 1.2026315789473684e-05,
      "loss": 0.2689,
      "step": 1655
    },
    {
      "epoch": 2.3522727272727275,
      "grad_norm": 0.2343615239222533,
      "learning_rate": 1.2e-05,
      "loss": 0.2856,
      "step": 1656
    },
    {
      "epoch": 2.3536931818181817,
      "grad_norm": 0.200374962785224,
      "learning_rate": 1.1973684210526315e-05,
      "loss": 0.2855,
      "step": 1657
    },
    {
      "epoch": 2.3551136363636362,
      "grad_norm": 0.22454079664582777,
      "learning_rate": 1.1947368421052632e-05,
      "loss": 0.2766,
      "step": 1658
    },
    {
      "epoch": 2.356534090909091,
      "grad_norm": 0.23845913924113207,
      "learning_rate": 1.1921052631578949e-05,
      "loss": 0.2952,
      "step": 1659
    },
    {
      "epoch": 2.3579545454545454,
      "grad_norm": 0.21602208250168292,
      "learning_rate": 1.1894736842105264e-05,
      "loss": 0.2989,
      "step": 1660
    },
    {
      "epoch": 2.359375,
      "grad_norm": 0.2569043051226051,
      "learning_rate": 1.186842105263158e-05,
      "loss": 0.2938,
      "step": 1661
    },
    {
      "epoch": 2.3607954545454546,
      "grad_norm": 0.23898479471245254,
      "learning_rate": 1.1842105263157895e-05,
      "loss": 0.2919,
      "step": 1662
    },
    {
      "epoch": 2.362215909090909,
      "grad_norm": 0.22549806416610846,
      "learning_rate": 1.181578947368421e-05,
      "loss": 0.3025,
      "step": 1663
    },
    {
      "epoch": 2.3636363636363638,
      "grad_norm": 0.2238471800695212,
      "learning_rate": 1.1789473684210527e-05,
      "loss": 0.2867,
      "step": 1664
    },
    {
      "epoch": 2.3650568181818183,
      "grad_norm": 0.23475058985535627,
      "learning_rate": 1.1763157894736842e-05,
      "loss": 0.2885,
      "step": 1665
    },
    {
      "epoch": 2.3664772727272725,
      "grad_norm": 0.22331377420359028,
      "learning_rate": 1.1736842105263158e-05,
      "loss": 0.2889,
      "step": 1666
    },
    {
      "epoch": 2.3678977272727275,
      "grad_norm": 0.22352000845937428,
      "learning_rate": 1.1710526315789475e-05,
      "loss": 0.2837,
      "step": 1667
    },
    {
      "epoch": 2.3693181818181817,
      "grad_norm": 0.23248855594404055,
      "learning_rate": 1.168421052631579e-05,
      "loss": 0.2752,
      "step": 1668
    },
    {
      "epoch": 2.3707386363636362,
      "grad_norm": 0.19911822541694918,
      "learning_rate": 1.1657894736842105e-05,
      "loss": 0.2723,
      "step": 1669
    },
    {
      "epoch": 2.372159090909091,
      "grad_norm": 0.22956984252821874,
      "learning_rate": 1.1631578947368422e-05,
      "loss": 0.276,
      "step": 1670
    },
    {
      "epoch": 2.3735795454545454,
      "grad_norm": 0.21900874635396944,
      "learning_rate": 1.1605263157894737e-05,
      "loss": 0.2892,
      "step": 1671
    },
    {
      "epoch": 2.375,
      "grad_norm": 0.20834610543846369,
      "learning_rate": 1.1578947368421053e-05,
      "loss": 0.2777,
      "step": 1672
    },
    {
      "epoch": 2.3764204545454546,
      "grad_norm": 0.23652069710009802,
      "learning_rate": 1.1552631578947368e-05,
      "loss": 0.2882,
      "step": 1673
    },
    {
      "epoch": 2.377840909090909,
      "grad_norm": 0.22054657380680143,
      "learning_rate": 1.1526315789473685e-05,
      "loss": 0.2845,
      "step": 1674
    },
    {
      "epoch": 2.3792613636363638,
      "grad_norm": 0.23131436152850182,
      "learning_rate": 1.1500000000000002e-05,
      "loss": 0.2812,
      "step": 1675
    },
    {
      "epoch": 2.3806818181818183,
      "grad_norm": 0.22544422588107654,
      "learning_rate": 1.1473684210526315e-05,
      "loss": 0.2788,
      "step": 1676
    },
    {
      "epoch": 2.3821022727272725,
      "grad_norm": 0.22363324202343013,
      "learning_rate": 1.1447368421052632e-05,
      "loss": 0.2875,
      "step": 1677
    },
    {
      "epoch": 2.3835227272727275,
      "grad_norm": 0.236818103850594,
      "learning_rate": 1.1421052631578948e-05,
      "loss": 0.2921,
      "step": 1678
    },
    {
      "epoch": 2.3849431818181817,
      "grad_norm": 0.23146461037164456,
      "learning_rate": 1.1394736842105263e-05,
      "loss": 0.2869,
      "step": 1679
    },
    {
      "epoch": 2.3863636363636362,
      "grad_norm": 0.21132148875878773,
      "learning_rate": 1.136842105263158e-05,
      "loss": 0.2762,
      "step": 1680
    },
    {
      "epoch": 2.387784090909091,
      "grad_norm": 0.2003795643651152,
      "learning_rate": 1.1342105263157895e-05,
      "loss": 0.2693,
      "step": 1681
    },
    {
      "epoch": 2.3892045454545454,
      "grad_norm": 0.21788042192366125,
      "learning_rate": 1.1315789473684212e-05,
      "loss": 0.2813,
      "step": 1682
    },
    {
      "epoch": 2.390625,
      "grad_norm": 0.25015585172004584,
      "learning_rate": 1.1289473684210528e-05,
      "loss": 0.3127,
      "step": 1683
    },
    {
      "epoch": 2.3920454545454546,
      "grad_norm": 0.21492530444922664,
      "learning_rate": 1.1263157894736842e-05,
      "loss": 0.2804,
      "step": 1684
    },
    {
      "epoch": 2.393465909090909,
      "grad_norm": 0.242167393278772,
      "learning_rate": 1.1236842105263158e-05,
      "loss": 0.3052,
      "step": 1685
    },
    {
      "epoch": 2.3948863636363638,
      "grad_norm": 0.23293851525270345,
      "learning_rate": 1.1210526315789475e-05,
      "loss": 0.2876,
      "step": 1686
    },
    {
      "epoch": 2.3963068181818183,
      "grad_norm": 0.2410431788820137,
      "learning_rate": 1.118421052631579e-05,
      "loss": 0.2897,
      "step": 1687
    },
    {
      "epoch": 2.3977272727272725,
      "grad_norm": 0.2563018205720928,
      "learning_rate": 1.1157894736842106e-05,
      "loss": 0.2954,
      "step": 1688
    },
    {
      "epoch": 2.3991477272727275,
      "grad_norm": 0.24067588956953204,
      "learning_rate": 1.1131578947368421e-05,
      "loss": 0.2865,
      "step": 1689
    },
    {
      "epoch": 2.4005681818181817,
      "grad_norm": 0.21815624127371197,
      "learning_rate": 1.1105263157894738e-05,
      "loss": 0.2778,
      "step": 1690
    },
    {
      "epoch": 2.4019886363636362,
      "grad_norm": 0.2127382125193727,
      "learning_rate": 1.1078947368421053e-05,
      "loss": 0.2961,
      "step": 1691
    },
    {
      "epoch": 2.403409090909091,
      "grad_norm": 0.2240673373299753,
      "learning_rate": 1.1052631578947368e-05,
      "loss": 0.2836,
      "step": 1692
    },
    {
      "epoch": 2.4048295454545454,
      "grad_norm": 0.22500462074489835,
      "learning_rate": 1.1026315789473685e-05,
      "loss": 0.2875,
      "step": 1693
    },
    {
      "epoch": 2.40625,
      "grad_norm": 0.21188238772760984,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.2776,
      "step": 1694
    },
    {
      "epoch": 2.4076704545454546,
      "grad_norm": 0.23966210273125899,
      "learning_rate": 1.0973684210526316e-05,
      "loss": 0.2848,
      "step": 1695
    },
    {
      "epoch": 2.409090909090909,
      "grad_norm": 0.250917902025772,
      "learning_rate": 1.0947368421052633e-05,
      "loss": 0.2924,
      "step": 1696
    },
    {
      "epoch": 2.4105113636363638,
      "grad_norm": 0.20291018064465874,
      "learning_rate": 1.0921052631578948e-05,
      "loss": 0.2712,
      "step": 1697
    },
    {
      "epoch": 2.4119318181818183,
      "grad_norm": 0.22504357333317082,
      "learning_rate": 1.0894736842105263e-05,
      "loss": 0.2894,
      "step": 1698
    },
    {
      "epoch": 2.4133522727272725,
      "grad_norm": 0.2435524913589297,
      "learning_rate": 1.086842105263158e-05,
      "loss": 0.287,
      "step": 1699
    },
    {
      "epoch": 2.4147727272727275,
      "grad_norm": 0.2513543041470002,
      "learning_rate": 1.0842105263157895e-05,
      "loss": 0.2758,
      "step": 1700
    },
    {
      "epoch": 2.4161931818181817,
      "grad_norm": 0.2195085131078361,
      "learning_rate": 1.0815789473684211e-05,
      "loss": 0.3036,
      "step": 1701
    },
    {
      "epoch": 2.4176136363636362,
      "grad_norm": 0.22363327983290718,
      "learning_rate": 1.0789473684210526e-05,
      "loss": 0.2862,
      "step": 1702
    },
    {
      "epoch": 2.419034090909091,
      "grad_norm": 0.24099049947822496,
      "learning_rate": 1.0763157894736843e-05,
      "loss": 0.2888,
      "step": 1703
    },
    {
      "epoch": 2.4204545454545454,
      "grad_norm": 0.2490426437028897,
      "learning_rate": 1.073684210526316e-05,
      "loss": 0.2864,
      "step": 1704
    },
    {
      "epoch": 2.421875,
      "grad_norm": 0.22929647725352478,
      "learning_rate": 1.0710526315789473e-05,
      "loss": 0.2989,
      "step": 1705
    },
    {
      "epoch": 2.4232954545454546,
      "grad_norm": 0.22268746623204183,
      "learning_rate": 1.068421052631579e-05,
      "loss": 0.2877,
      "step": 1706
    },
    {
      "epoch": 2.424715909090909,
      "grad_norm": 0.22968762121434558,
      "learning_rate": 1.0657894736842106e-05,
      "loss": 0.2831,
      "step": 1707
    },
    {
      "epoch": 2.4261363636363638,
      "grad_norm": 0.23326292072509452,
      "learning_rate": 1.0631578947368421e-05,
      "loss": 0.2813,
      "step": 1708
    },
    {
      "epoch": 2.4275568181818183,
      "grad_norm": 0.24710361734787079,
      "learning_rate": 1.0605263157894738e-05,
      "loss": 0.3032,
      "step": 1709
    },
    {
      "epoch": 2.4289772727272725,
      "grad_norm": 0.20341202538952033,
      "learning_rate": 1.0578947368421053e-05,
      "loss": 0.2679,
      "step": 1710
    },
    {
      "epoch": 2.4303977272727275,
      "grad_norm": 0.2324220755179762,
      "learning_rate": 1.055263157894737e-05,
      "loss": 0.2849,
      "step": 1711
    },
    {
      "epoch": 2.4318181818181817,
      "grad_norm": 0.22020699079795655,
      "learning_rate": 1.0526315789473684e-05,
      "loss": 0.2837,
      "step": 1712
    },
    {
      "epoch": 2.4332386363636362,
      "grad_norm": 0.21076195865253372,
      "learning_rate": 1.05e-05,
      "loss": 0.2798,
      "step": 1713
    },
    {
      "epoch": 2.434659090909091,
      "grad_norm": 0.22014636813364535,
      "learning_rate": 1.0473684210526316e-05,
      "loss": 0.2746,
      "step": 1714
    },
    {
      "epoch": 2.4360795454545454,
      "grad_norm": 0.2119634632720555,
      "learning_rate": 1.0447368421052633e-05,
      "loss": 0.2811,
      "step": 1715
    },
    {
      "epoch": 2.4375,
      "grad_norm": 0.1947662279986864,
      "learning_rate": 1.0421052631578948e-05,
      "loss": 0.2776,
      "step": 1716
    },
    {
      "epoch": 2.4389204545454546,
      "grad_norm": 0.21825441026367717,
      "learning_rate": 1.0394736842105264e-05,
      "loss": 0.285,
      "step": 1717
    },
    {
      "epoch": 2.440340909090909,
      "grad_norm": 0.2537325320429484,
      "learning_rate": 1.036842105263158e-05,
      "loss": 0.2955,
      "step": 1718
    },
    {
      "epoch": 2.4417613636363638,
      "grad_norm": 0.2252290883253739,
      "learning_rate": 1.0342105263157894e-05,
      "loss": 0.2875,
      "step": 1719
    },
    {
      "epoch": 2.4431818181818183,
      "grad_norm": 0.20448733162943905,
      "learning_rate": 1.0315789473684211e-05,
      "loss": 0.2693,
      "step": 1720
    },
    {
      "epoch": 2.4446022727272725,
      "grad_norm": 0.24205892764650208,
      "learning_rate": 1.0289473684210526e-05,
      "loss": 0.2847,
      "step": 1721
    },
    {
      "epoch": 2.4460227272727275,
      "grad_norm": 0.2482314009814398,
      "learning_rate": 1.0263157894736843e-05,
      "loss": 0.2872,
      "step": 1722
    },
    {
      "epoch": 2.4474431818181817,
      "grad_norm": 0.23394450760010146,
      "learning_rate": 1.023684210526316e-05,
      "loss": 0.2891,
      "step": 1723
    },
    {
      "epoch": 2.4488636363636362,
      "grad_norm": 0.21855519373233262,
      "learning_rate": 1.0210526315789474e-05,
      "loss": 0.2755,
      "step": 1724
    },
    {
      "epoch": 2.450284090909091,
      "grad_norm": 0.2370563131441957,
      "learning_rate": 1.0184210526315791e-05,
      "loss": 0.2741,
      "step": 1725
    },
    {
      "epoch": 2.4517045454545454,
      "grad_norm": 0.24439743758487237,
      "learning_rate": 1.0157894736842106e-05,
      "loss": 0.2976,
      "step": 1726
    },
    {
      "epoch": 2.453125,
      "grad_norm": 0.20756914182967273,
      "learning_rate": 1.0131578947368421e-05,
      "loss": 0.2827,
      "step": 1727
    },
    {
      "epoch": 2.4545454545454546,
      "grad_norm": 0.22519637571346535,
      "learning_rate": 1.0105263157894738e-05,
      "loss": 0.2925,
      "step": 1728
    },
    {
      "epoch": 2.455965909090909,
      "grad_norm": 0.2289775372444571,
      "learning_rate": 1.0078947368421053e-05,
      "loss": 0.2927,
      "step": 1729
    },
    {
      "epoch": 2.4573863636363638,
      "grad_norm": 0.23601534990300138,
      "learning_rate": 1.005263157894737e-05,
      "loss": 0.2885,
      "step": 1730
    },
    {
      "epoch": 2.4588068181818183,
      "grad_norm": 0.26168288828543085,
      "learning_rate": 1.0026315789473686e-05,
      "loss": 0.2746,
      "step": 1731
    },
    {
      "epoch": 2.4602272727272725,
      "grad_norm": 0.26140765399487664,
      "learning_rate": 1e-05,
      "loss": 0.2784,
      "step": 1732
    },
    {
      "epoch": 2.4616477272727275,
      "grad_norm": 0.23101243607452002,
      "learning_rate": 9.973684210526316e-06,
      "loss": 0.2978,
      "step": 1733
    },
    {
      "epoch": 2.4630681818181817,
      "grad_norm": 0.2290501203304175,
      "learning_rate": 9.947368421052632e-06,
      "loss": 0.2762,
      "step": 1734
    },
    {
      "epoch": 2.4644886363636362,
      "grad_norm": 0.23163474399340475,
      "learning_rate": 9.921052631578947e-06,
      "loss": 0.283,
      "step": 1735
    },
    {
      "epoch": 2.465909090909091,
      "grad_norm": 0.24054648231644324,
      "learning_rate": 9.894736842105264e-06,
      "loss": 0.2877,
      "step": 1736
    },
    {
      "epoch": 2.4673295454545454,
      "grad_norm": 0.23236652170591837,
      "learning_rate": 9.868421052631579e-06,
      "loss": 0.2952,
      "step": 1737
    },
    {
      "epoch": 2.46875,
      "grad_norm": 0.2129967643456758,
      "learning_rate": 9.842105263157896e-06,
      "loss": 0.2832,
      "step": 1738
    },
    {
      "epoch": 2.4701704545454546,
      "grad_norm": 0.21889927631070955,
      "learning_rate": 9.81578947368421e-06,
      "loss": 0.2868,
      "step": 1739
    },
    {
      "epoch": 2.471590909090909,
      "grad_norm": 0.23457729338888467,
      "learning_rate": 9.789473684210526e-06,
      "loss": 0.2923,
      "step": 1740
    },
    {
      "epoch": 2.4730113636363638,
      "grad_norm": 0.23448252516099405,
      "learning_rate": 9.763157894736842e-06,
      "loss": 0.2804,
      "step": 1741
    },
    {
      "epoch": 2.4744318181818183,
      "grad_norm": 0.2142592256409191,
      "learning_rate": 9.736842105263157e-06,
      "loss": 0.2848,
      "step": 1742
    },
    {
      "epoch": 2.4758522727272725,
      "grad_norm": 0.22239383860123182,
      "learning_rate": 9.710526315789474e-06,
      "loss": 0.2835,
      "step": 1743
    },
    {
      "epoch": 2.4772727272727275,
      "grad_norm": 0.2707459043711267,
      "learning_rate": 9.68421052631579e-06,
      "loss": 0.3032,
      "step": 1744
    },
    {
      "epoch": 2.4786931818181817,
      "grad_norm": 0.21380575302888633,
      "learning_rate": 9.657894736842106e-06,
      "loss": 0.2855,
      "step": 1745
    },
    {
      "epoch": 2.4801136363636362,
      "grad_norm": 0.23586169271689464,
      "learning_rate": 9.631578947368422e-06,
      "loss": 0.2932,
      "step": 1746
    },
    {
      "epoch": 2.481534090909091,
      "grad_norm": 0.21880602656699705,
      "learning_rate": 9.605263157894737e-06,
      "loss": 0.2851,
      "step": 1747
    },
    {
      "epoch": 2.4829545454545454,
      "grad_norm": 0.22267636273225497,
      "learning_rate": 9.578947368421052e-06,
      "loss": 0.2686,
      "step": 1748
    },
    {
      "epoch": 2.484375,
      "grad_norm": 0.23546362378071942,
      "learning_rate": 9.552631578947369e-06,
      "loss": 0.2918,
      "step": 1749
    },
    {
      "epoch": 2.4857954545454546,
      "grad_norm": 0.22261396215667612,
      "learning_rate": 9.526315789473684e-06,
      "loss": 0.2853,
      "step": 1750
    },
    {
      "epoch": 2.487215909090909,
      "grad_norm": 0.22934261411498205,
      "learning_rate": 9.5e-06,
      "loss": 0.2747,
      "step": 1751
    },
    {
      "epoch": 2.4886363636363638,
      "grad_norm": 0.21228395722270238,
      "learning_rate": 9.473684210526317e-06,
      "loss": 0.2816,
      "step": 1752
    },
    {
      "epoch": 2.4900568181818183,
      "grad_norm": 0.20982058080472596,
      "learning_rate": 9.447368421052632e-06,
      "loss": 0.2664,
      "step": 1753
    },
    {
      "epoch": 2.4914772727272725,
      "grad_norm": 0.2010397353625089,
      "learning_rate": 9.421052631578947e-06,
      "loss": 0.2779,
      "step": 1754
    },
    {
      "epoch": 2.4928977272727275,
      "grad_norm": 0.23054754433871852,
      "learning_rate": 9.394736842105264e-06,
      "loss": 0.2855,
      "step": 1755
    },
    {
      "epoch": 2.4943181818181817,
      "grad_norm": 0.24562161068436925,
      "learning_rate": 9.368421052631579e-06,
      "loss": 0.2894,
      "step": 1756
    },
    {
      "epoch": 2.4957386363636362,
      "grad_norm": 0.21772661626974726,
      "learning_rate": 9.342105263157895e-06,
      "loss": 0.2785,
      "step": 1757
    },
    {
      "epoch": 2.497159090909091,
      "grad_norm": 0.22217218380075907,
      "learning_rate": 9.31578947368421e-06,
      "loss": 0.2833,
      "step": 1758
    },
    {
      "epoch": 2.4985795454545454,
      "grad_norm": 0.2253429982518151,
      "learning_rate": 9.289473684210527e-06,
      "loss": 0.269,
      "step": 1759
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.21722298474087903,
      "learning_rate": 9.263157894736844e-06,
      "loss": 0.2886,
      "step": 1760
    },
    {
      "epoch": 2.5014204545454546,
      "grad_norm": 0.2029945526756793,
      "learning_rate": 9.236842105263159e-06,
      "loss": 0.2904,
      "step": 1761
    },
    {
      "epoch": 2.502840909090909,
      "grad_norm": 0.23866951258303623,
      "learning_rate": 9.210526315789474e-06,
      "loss": 0.2939,
      "step": 1762
    },
    {
      "epoch": 2.5042613636363638,
      "grad_norm": 0.2641897499754154,
      "learning_rate": 9.18421052631579e-06,
      "loss": 0.2924,
      "step": 1763
    },
    {
      "epoch": 2.5056818181818183,
      "grad_norm": 0.23090784625575791,
      "learning_rate": 9.157894736842105e-06,
      "loss": 0.305,
      "step": 1764
    },
    {
      "epoch": 2.5071022727272725,
      "grad_norm": 0.22565635534045567,
      "learning_rate": 9.131578947368422e-06,
      "loss": 0.2647,
      "step": 1765
    },
    {
      "epoch": 2.5085227272727275,
      "grad_norm": 0.25306765326729896,
      "learning_rate": 9.105263157894737e-06,
      "loss": 0.3054,
      "step": 1766
    },
    {
      "epoch": 2.5099431818181817,
      "grad_norm": 0.20885506120593714,
      "learning_rate": 9.078947368421054e-06,
      "loss": 0.294,
      "step": 1767
    },
    {
      "epoch": 2.5113636363636362,
      "grad_norm": 0.2269599035054617,
      "learning_rate": 9.05263157894737e-06,
      "loss": 0.2796,
      "step": 1768
    },
    {
      "epoch": 2.512784090909091,
      "grad_norm": 0.2372324300100332,
      "learning_rate": 9.026315789473684e-06,
      "loss": 0.3012,
      "step": 1769
    },
    {
      "epoch": 2.5142045454545454,
      "grad_norm": 0.21986578674405477,
      "learning_rate": 9e-06,
      "loss": 0.2855,
      "step": 1770
    },
    {
      "epoch": 2.515625,
      "grad_norm": 0.23137697664971443,
      "learning_rate": 8.973684210526317e-06,
      "loss": 0.2912,
      "step": 1771
    },
    {
      "epoch": 2.5170454545454546,
      "grad_norm": 0.2274049567820983,
      "learning_rate": 8.947368421052632e-06,
      "loss": 0.2893,
      "step": 1772
    },
    {
      "epoch": 2.518465909090909,
      "grad_norm": 0.2282093439509061,
      "learning_rate": 8.921052631578949e-06,
      "loss": 0.2872,
      "step": 1773
    },
    {
      "epoch": 2.5198863636363638,
      "grad_norm": 0.224712261880999,
      "learning_rate": 8.894736842105264e-06,
      "loss": 0.2941,
      "step": 1774
    },
    {
      "epoch": 2.5213068181818183,
      "grad_norm": 0.2204345677921128,
      "learning_rate": 8.86842105263158e-06,
      "loss": 0.272,
      "step": 1775
    },
    {
      "epoch": 2.5227272727272725,
      "grad_norm": 0.2089569325547565,
      "learning_rate": 8.842105263157895e-06,
      "loss": 0.2709,
      "step": 1776
    },
    {
      "epoch": 2.5241477272727275,
      "grad_norm": 0.2225255128790774,
      "learning_rate": 8.81578947368421e-06,
      "loss": 0.276,
      "step": 1777
    },
    {
      "epoch": 2.5255681818181817,
      "grad_norm": 0.23074270277199935,
      "learning_rate": 8.789473684210527e-06,
      "loss": 0.302,
      "step": 1778
    },
    {
      "epoch": 2.5269886363636362,
      "grad_norm": 0.22245902800027023,
      "learning_rate": 8.763157894736842e-06,
      "loss": 0.2901,
      "step": 1779
    },
    {
      "epoch": 2.528409090909091,
      "grad_norm": 0.22123836037196207,
      "learning_rate": 8.736842105263158e-06,
      "loss": 0.2876,
      "step": 1780
    },
    {
      "epoch": 2.5298295454545454,
      "grad_norm": 0.22581830349339765,
      "learning_rate": 8.710526315789475e-06,
      "loss": 0.2989,
      "step": 1781
    },
    {
      "epoch": 2.53125,
      "grad_norm": 0.24358056695218563,
      "learning_rate": 8.68421052631579e-06,
      "loss": 0.291,
      "step": 1782
    },
    {
      "epoch": 2.5326704545454546,
      "grad_norm": 0.24537935532279617,
      "learning_rate": 8.657894736842105e-06,
      "loss": 0.2844,
      "step": 1783
    },
    {
      "epoch": 2.534090909090909,
      "grad_norm": 0.22489728916995883,
      "learning_rate": 8.631578947368422e-06,
      "loss": 0.2897,
      "step": 1784
    },
    {
      "epoch": 2.5355113636363638,
      "grad_norm": 0.22602388539915103,
      "learning_rate": 8.605263157894737e-06,
      "loss": 0.2905,
      "step": 1785
    },
    {
      "epoch": 2.5369318181818183,
      "grad_norm": 1.0855290519512812,
      "learning_rate": 8.578947368421053e-06,
      "loss": 0.2856,
      "step": 1786
    },
    {
      "epoch": 2.5383522727272725,
      "grad_norm": 0.20585190570150794,
      "learning_rate": 8.552631578947368e-06,
      "loss": 0.2847,
      "step": 1787
    },
    {
      "epoch": 2.5397727272727275,
      "grad_norm": 0.21675316588561305,
      "learning_rate": 8.526315789473685e-06,
      "loss": 0.2943,
      "step": 1788
    },
    {
      "epoch": 2.5411931818181817,
      "grad_norm": 0.21542247258270508,
      "learning_rate": 8.500000000000002e-06,
      "loss": 0.2805,
      "step": 1789
    },
    {
      "epoch": 2.5426136363636362,
      "grad_norm": 0.23718411565970546,
      "learning_rate": 8.473684210526315e-06,
      "loss": 0.29,
      "step": 1790
    },
    {
      "epoch": 2.544034090909091,
      "grad_norm": 0.23643942206526378,
      "learning_rate": 8.447368421052632e-06,
      "loss": 0.2903,
      "step": 1791
    },
    {
      "epoch": 2.5454545454545454,
      "grad_norm": 0.20834786561825075,
      "learning_rate": 8.421052631578948e-06,
      "loss": 0.2854,
      "step": 1792
    },
    {
      "epoch": 2.546875,
      "grad_norm": 0.21941017080157318,
      "learning_rate": 8.394736842105263e-06,
      "loss": 0.306,
      "step": 1793
    },
    {
      "epoch": 2.5482954545454546,
      "grad_norm": 0.20820078026087704,
      "learning_rate": 8.36842105263158e-06,
      "loss": 0.2899,
      "step": 1794
    },
    {
      "epoch": 2.549715909090909,
      "grad_norm": 0.21961398794466958,
      "learning_rate": 8.342105263157895e-06,
      "loss": 0.289,
      "step": 1795
    },
    {
      "epoch": 2.5511363636363638,
      "grad_norm": 0.24408735734921383,
      "learning_rate": 8.315789473684212e-06,
      "loss": 0.2873,
      "step": 1796
    },
    {
      "epoch": 2.5525568181818183,
      "grad_norm": 0.2030218436240362,
      "learning_rate": 8.289473684210526e-06,
      "loss": 0.2716,
      "step": 1797
    },
    {
      "epoch": 2.5539772727272725,
      "grad_norm": 0.2060398848151977,
      "learning_rate": 8.263157894736841e-06,
      "loss": 0.2849,
      "step": 1798
    },
    {
      "epoch": 2.5553977272727275,
      "grad_norm": 0.1972328849444099,
      "learning_rate": 8.236842105263158e-06,
      "loss": 0.2973,
      "step": 1799
    },
    {
      "epoch": 2.5568181818181817,
      "grad_norm": 0.2176659259015495,
      "learning_rate": 8.210526315789475e-06,
      "loss": 0.2925,
      "step": 1800
    },
    {
      "epoch": 2.5582386363636362,
      "grad_norm": 0.2312439322650524,
      "learning_rate": 8.18421052631579e-06,
      "loss": 0.2921,
      "step": 1801
    },
    {
      "epoch": 2.559659090909091,
      "grad_norm": 0.2074724114503062,
      "learning_rate": 8.157894736842106e-06,
      "loss": 0.2768,
      "step": 1802
    },
    {
      "epoch": 2.5610795454545454,
      "grad_norm": 0.2149375080865575,
      "learning_rate": 8.131578947368421e-06,
      "loss": 0.2906,
      "step": 1803
    },
    {
      "epoch": 2.5625,
      "grad_norm": 0.22243875618527434,
      "learning_rate": 8.105263157894736e-06,
      "loss": 0.2815,
      "step": 1804
    },
    {
      "epoch": 2.5639204545454546,
      "grad_norm": 0.21243813398015043,
      "learning_rate": 8.078947368421053e-06,
      "loss": 0.2937,
      "step": 1805
    },
    {
      "epoch": 2.565340909090909,
      "grad_norm": 0.22948710311206397,
      "learning_rate": 8.052631578947368e-06,
      "loss": 0.2821,
      "step": 1806
    },
    {
      "epoch": 2.5667613636363638,
      "grad_norm": 0.22284620428431354,
      "learning_rate": 8.026315789473685e-06,
      "loss": 0.2819,
      "step": 1807
    },
    {
      "epoch": 2.5681818181818183,
      "grad_norm": 0.21216486118504874,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.2884,
      "step": 1808
    },
    {
      "epoch": 2.5696022727272725,
      "grad_norm": 0.20568264366931457,
      "learning_rate": 7.973684210526316e-06,
      "loss": 0.2837,
      "step": 1809
    },
    {
      "epoch": 2.5710227272727275,
      "grad_norm": 0.2432415750990266,
      "learning_rate": 7.947368421052633e-06,
      "loss": 0.2937,
      "step": 1810
    },
    {
      "epoch": 2.5724431818181817,
      "grad_norm": 0.2081595430168041,
      "learning_rate": 7.921052631578948e-06,
      "loss": 0.2848,
      "step": 1811
    },
    {
      "epoch": 2.5738636363636362,
      "grad_norm": 0.20741351863299193,
      "learning_rate": 7.894736842105263e-06,
      "loss": 0.2698,
      "step": 1812
    },
    {
      "epoch": 2.575284090909091,
      "grad_norm": 0.2211916033521008,
      "learning_rate": 7.86842105263158e-06,
      "loss": 0.2868,
      "step": 1813
    },
    {
      "epoch": 2.5767045454545454,
      "grad_norm": 0.21358730315186195,
      "learning_rate": 7.842105263157895e-06,
      "loss": 0.2829,
      "step": 1814
    },
    {
      "epoch": 2.578125,
      "grad_norm": 0.19110874563801886,
      "learning_rate": 7.815789473684211e-06,
      "loss": 0.2758,
      "step": 1815
    },
    {
      "epoch": 2.5795454545454546,
      "grad_norm": 0.22474797053580253,
      "learning_rate": 7.789473684210528e-06,
      "loss": 0.2953,
      "step": 1816
    },
    {
      "epoch": 2.580965909090909,
      "grad_norm": 0.204342235826731,
      "learning_rate": 7.763157894736843e-06,
      "loss": 0.2703,
      "step": 1817
    },
    {
      "epoch": 2.5823863636363638,
      "grad_norm": 0.2171545342012138,
      "learning_rate": 7.736842105263158e-06,
      "loss": 0.2913,
      "step": 1818
    },
    {
      "epoch": 2.5838068181818183,
      "grad_norm": 0.20488176375702105,
      "learning_rate": 7.710526315789473e-06,
      "loss": 0.2758,
      "step": 1819
    },
    {
      "epoch": 2.5852272727272725,
      "grad_norm": 0.21246825686802773,
      "learning_rate": 7.68421052631579e-06,
      "loss": 0.2755,
      "step": 1820
    },
    {
      "epoch": 2.5866477272727275,
      "grad_norm": 0.22541243936945535,
      "learning_rate": 7.657894736842106e-06,
      "loss": 0.3112,
      "step": 1821
    },
    {
      "epoch": 2.5880681818181817,
      "grad_norm": 0.20287454307565753,
      "learning_rate": 7.631578947368421e-06,
      "loss": 0.2881,
      "step": 1822
    },
    {
      "epoch": 2.5894886363636362,
      "grad_norm": 0.2238630973317767,
      "learning_rate": 7.605263157894738e-06,
      "loss": 0.2781,
      "step": 1823
    },
    {
      "epoch": 2.590909090909091,
      "grad_norm": 0.2236601305389228,
      "learning_rate": 7.578947368421054e-06,
      "loss": 0.2848,
      "step": 1824
    },
    {
      "epoch": 2.5923295454545454,
      "grad_norm": 0.2149632350785527,
      "learning_rate": 7.552631578947368e-06,
      "loss": 0.3032,
      "step": 1825
    },
    {
      "epoch": 2.59375,
      "grad_norm": 0.22330270369033453,
      "learning_rate": 7.526315789473684e-06,
      "loss": 0.2849,
      "step": 1826
    },
    {
      "epoch": 2.5951704545454546,
      "grad_norm": 0.20796925272878322,
      "learning_rate": 7.5e-06,
      "loss": 0.286,
      "step": 1827
    },
    {
      "epoch": 2.596590909090909,
      "grad_norm": 0.2124094241743557,
      "learning_rate": 7.473684210526316e-06,
      "loss": 0.2685,
      "step": 1828
    },
    {
      "epoch": 2.5980113636363638,
      "grad_norm": 0.22521896883460488,
      "learning_rate": 7.447368421052632e-06,
      "loss": 0.2775,
      "step": 1829
    },
    {
      "epoch": 2.5994318181818183,
      "grad_norm": 0.23641992839526124,
      "learning_rate": 7.421052631578948e-06,
      "loss": 0.289,
      "step": 1830
    },
    {
      "epoch": 2.6008522727272725,
      "grad_norm": 0.2113538792238693,
      "learning_rate": 7.394736842105264e-06,
      "loss": 0.2841,
      "step": 1831
    },
    {
      "epoch": 2.6022727272727275,
      "grad_norm": 0.22617600404172417,
      "learning_rate": 7.3684210526315784e-06,
      "loss": 0.2713,
      "step": 1832
    },
    {
      "epoch": 2.6036931818181817,
      "grad_norm": 0.22872938281750183,
      "learning_rate": 7.342105263157894e-06,
      "loss": 0.304,
      "step": 1833
    },
    {
      "epoch": 2.6051136363636362,
      "grad_norm": 0.2319570892075966,
      "learning_rate": 7.315789473684211e-06,
      "loss": 0.2672,
      "step": 1834
    },
    {
      "epoch": 2.606534090909091,
      "grad_norm": 0.2180786881727629,
      "learning_rate": 7.289473684210527e-06,
      "loss": 0.2974,
      "step": 1835
    },
    {
      "epoch": 2.6079545454545454,
      "grad_norm": 0.20847613939115434,
      "learning_rate": 7.2631578947368426e-06,
      "loss": 0.2615,
      "step": 1836
    },
    {
      "epoch": 2.609375,
      "grad_norm": 0.22056565617974155,
      "learning_rate": 7.236842105263158e-06,
      "loss": 0.2823,
      "step": 1837
    },
    {
      "epoch": 2.6107954545454546,
      "grad_norm": 0.2549457709596277,
      "learning_rate": 7.210526315789474e-06,
      "loss": 0.2941,
      "step": 1838
    },
    {
      "epoch": 2.612215909090909,
      "grad_norm": 0.22408852519128003,
      "learning_rate": 7.184210526315791e-06,
      "loss": 0.2894,
      "step": 1839
    },
    {
      "epoch": 2.6136363636363638,
      "grad_norm": 0.2019426063480025,
      "learning_rate": 7.157894736842105e-06,
      "loss": 0.2914,
      "step": 1840
    },
    {
      "epoch": 2.6150568181818183,
      "grad_norm": 0.2723150643157409,
      "learning_rate": 7.131578947368421e-06,
      "loss": 0.3076,
      "step": 1841
    },
    {
      "epoch": 2.6164772727272725,
      "grad_norm": 0.20630348774662452,
      "learning_rate": 7.1052631578947375e-06,
      "loss": 0.2772,
      "step": 1842
    },
    {
      "epoch": 2.6178977272727275,
      "grad_norm": 0.21615128658799165,
      "learning_rate": 7.078947368421053e-06,
      "loss": 0.2983,
      "step": 1843
    },
    {
      "epoch": 2.6193181818181817,
      "grad_norm": 0.20664937761033472,
      "learning_rate": 7.052631578947369e-06,
      "loss": 0.2761,
      "step": 1844
    },
    {
      "epoch": 2.6207386363636362,
      "grad_norm": 0.19903063112936217,
      "learning_rate": 7.026315789473685e-06,
      "loss": 0.2827,
      "step": 1845
    },
    {
      "epoch": 2.622159090909091,
      "grad_norm": 0.2185115100378836,
      "learning_rate": 7.000000000000001e-06,
      "loss": 0.2897,
      "step": 1846
    },
    {
      "epoch": 2.6235795454545454,
      "grad_norm": 0.21402268120245202,
      "learning_rate": 6.973684210526316e-06,
      "loss": 0.2867,
      "step": 1847
    },
    {
      "epoch": 2.625,
      "grad_norm": 0.20689728777400007,
      "learning_rate": 6.9473684210526315e-06,
      "loss": 0.2797,
      "step": 1848
    },
    {
      "epoch": 2.6264204545454546,
      "grad_norm": 0.20566712893647376,
      "learning_rate": 6.921052631578947e-06,
      "loss": 0.2747,
      "step": 1849
    },
    {
      "epoch": 2.627840909090909,
      "grad_norm": 0.20427080091317196,
      "learning_rate": 6.894736842105263e-06,
      "loss": 0.2816,
      "step": 1850
    },
    {
      "epoch": 2.6292613636363638,
      "grad_norm": 0.2013894217940976,
      "learning_rate": 6.86842105263158e-06,
      "loss": 0.274,
      "step": 1851
    },
    {
      "epoch": 2.6306818181818183,
      "grad_norm": 0.21898844034398804,
      "learning_rate": 6.842105263157896e-06,
      "loss": 0.2804,
      "step": 1852
    },
    {
      "epoch": 2.6321022727272725,
      "grad_norm": 0.21601841565195448,
      "learning_rate": 6.8157894736842115e-06,
      "loss": 0.2814,
      "step": 1853
    },
    {
      "epoch": 2.6335227272727275,
      "grad_norm": 0.24468300520314526,
      "learning_rate": 6.7894736842105264e-06,
      "loss": 0.2855,
      "step": 1854
    },
    {
      "epoch": 2.6349431818181817,
      "grad_norm": 0.2310526050154595,
      "learning_rate": 6.763157894736842e-06,
      "loss": 0.3042,
      "step": 1855
    },
    {
      "epoch": 2.6363636363636362,
      "grad_norm": 0.2404968998901195,
      "learning_rate": 6.736842105263158e-06,
      "loss": 0.2883,
      "step": 1856
    },
    {
      "epoch": 2.637784090909091,
      "grad_norm": 0.2174757309284916,
      "learning_rate": 6.710526315789474e-06,
      "loss": 0.2879,
      "step": 1857
    },
    {
      "epoch": 2.6392045454545454,
      "grad_norm": 0.23015404968746142,
      "learning_rate": 6.68421052631579e-06,
      "loss": 0.2897,
      "step": 1858
    },
    {
      "epoch": 2.640625,
      "grad_norm": 0.19685818672677668,
      "learning_rate": 6.657894736842106e-06,
      "loss": 0.2844,
      "step": 1859
    },
    {
      "epoch": 2.6420454545454546,
      "grad_norm": 0.22060797569313723,
      "learning_rate": 6.631578947368422e-06,
      "loss": 0.283,
      "step": 1860
    },
    {
      "epoch": 2.643465909090909,
      "grad_norm": 0.21836512359183707,
      "learning_rate": 6.605263157894736e-06,
      "loss": 0.2991,
      "step": 1861
    },
    {
      "epoch": 2.6448863636363638,
      "grad_norm": 0.22564103042724223,
      "learning_rate": 6.578947368421053e-06,
      "loss": 0.2818,
      "step": 1862
    },
    {
      "epoch": 2.6463068181818183,
      "grad_norm": 0.2171605864703467,
      "learning_rate": 6.552631578947369e-06,
      "loss": 0.3104,
      "step": 1863
    },
    {
      "epoch": 2.6477272727272725,
      "grad_norm": 0.21519559235522598,
      "learning_rate": 6.526315789473685e-06,
      "loss": 0.2854,
      "step": 1864
    },
    {
      "epoch": 2.6491477272727275,
      "grad_norm": 0.20833173031427094,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 0.27,
      "step": 1865
    },
    {
      "epoch": 2.6505681818181817,
      "grad_norm": 0.1979785588860183,
      "learning_rate": 6.473684210526316e-06,
      "loss": 0.2776,
      "step": 1866
    },
    {
      "epoch": 2.6519886363636362,
      "grad_norm": 0.2305158960522121,
      "learning_rate": 6.447368421052632e-06,
      "loss": 0.2663,
      "step": 1867
    },
    {
      "epoch": 2.653409090909091,
      "grad_norm": 0.2111565742802499,
      "learning_rate": 6.421052631578947e-06,
      "loss": 0.2816,
      "step": 1868
    },
    {
      "epoch": 2.6548295454545454,
      "grad_norm": 0.22791663865928724,
      "learning_rate": 6.394736842105263e-06,
      "loss": 0.2891,
      "step": 1869
    },
    {
      "epoch": 2.65625,
      "grad_norm": 0.22582285781193676,
      "learning_rate": 6.368421052631579e-06,
      "loss": 0.2988,
      "step": 1870
    },
    {
      "epoch": 2.6576704545454546,
      "grad_norm": 0.23343776576637543,
      "learning_rate": 6.342105263157895e-06,
      "loss": 0.2806,
      "step": 1871
    },
    {
      "epoch": 2.659090909090909,
      "grad_norm": 0.2082927488731646,
      "learning_rate": 6.315789473684211e-06,
      "loss": 0.2858,
      "step": 1872
    },
    {
      "epoch": 2.6605113636363638,
      "grad_norm": 0.19776949104922925,
      "learning_rate": 6.289473684210527e-06,
      "loss": 0.2915,
      "step": 1873
    },
    {
      "epoch": 2.6619318181818183,
      "grad_norm": 0.2381528660930004,
      "learning_rate": 6.263157894736843e-06,
      "loss": 0.2915,
      "step": 1874
    },
    {
      "epoch": 2.6633522727272725,
      "grad_norm": 0.20494478820295017,
      "learning_rate": 6.236842105263159e-06,
      "loss": 0.2874,
      "step": 1875
    },
    {
      "epoch": 2.6647727272727275,
      "grad_norm": 0.19173920707075495,
      "learning_rate": 6.2105263157894745e-06,
      "loss": 0.275,
      "step": 1876
    },
    {
      "epoch": 2.6661931818181817,
      "grad_norm": 0.22434062136395733,
      "learning_rate": 6.184210526315789e-06,
      "loss": 0.2894,
      "step": 1877
    },
    {
      "epoch": 2.6676136363636362,
      "grad_norm": 0.20231183596269633,
      "learning_rate": 6.157894736842105e-06,
      "loss": 0.283,
      "step": 1878
    },
    {
      "epoch": 2.669034090909091,
      "grad_norm": 0.2007199792096695,
      "learning_rate": 6.131578947368422e-06,
      "loss": 0.278,
      "step": 1879
    },
    {
      "epoch": 2.6704545454545454,
      "grad_norm": 0.20729185879604586,
      "learning_rate": 6.105263157894737e-06,
      "loss": 0.2894,
      "step": 1880
    },
    {
      "epoch": 2.671875,
      "grad_norm": 0.20770077145456178,
      "learning_rate": 6.078947368421053e-06,
      "loss": 0.2703,
      "step": 1881
    },
    {
      "epoch": 2.6732954545454546,
      "grad_norm": 0.23126236618219884,
      "learning_rate": 6.0526315789473685e-06,
      "loss": 0.2869,
      "step": 1882
    },
    {
      "epoch": 2.674715909090909,
      "grad_norm": 0.20939280520386463,
      "learning_rate": 6.026315789473685e-06,
      "loss": 0.2783,
      "step": 1883
    },
    {
      "epoch": 2.6761363636363638,
      "grad_norm": 0.20133907602310033,
      "learning_rate": 6e-06,
      "loss": 0.2873,
      "step": 1884
    },
    {
      "epoch": 2.6775568181818183,
      "grad_norm": 0.20504969912115575,
      "learning_rate": 5.973684210526316e-06,
      "loss": 0.2734,
      "step": 1885
    },
    {
      "epoch": 2.6789772727272725,
      "grad_norm": 0.20299868950371697,
      "learning_rate": 5.947368421052632e-06,
      "loss": 0.2804,
      "step": 1886
    },
    {
      "epoch": 2.6803977272727275,
      "grad_norm": 0.21123726120003905,
      "learning_rate": 5.921052631578948e-06,
      "loss": 0.2818,
      "step": 1887
    },
    {
      "epoch": 2.6818181818181817,
      "grad_norm": 0.20787309255684605,
      "learning_rate": 5.8947368421052634e-06,
      "loss": 0.2994,
      "step": 1888
    },
    {
      "epoch": 2.6832386363636362,
      "grad_norm": 0.19244698435488866,
      "learning_rate": 5.868421052631579e-06,
      "loss": 0.2706,
      "step": 1889
    },
    {
      "epoch": 2.684659090909091,
      "grad_norm": 0.20263744666784628,
      "learning_rate": 5.842105263157895e-06,
      "loss": 0.2881,
      "step": 1890
    },
    {
      "epoch": 2.6860795454545454,
      "grad_norm": 0.1950326179456346,
      "learning_rate": 5.815789473684211e-06,
      "loss": 0.2659,
      "step": 1891
    },
    {
      "epoch": 2.6875,
      "grad_norm": 0.22102947065446757,
      "learning_rate": 5.789473684210527e-06,
      "loss": 0.2763,
      "step": 1892
    },
    {
      "epoch": 2.6889204545454546,
      "grad_norm": 0.21875723152480803,
      "learning_rate": 5.7631578947368425e-06,
      "loss": 0.2812,
      "step": 1893
    },
    {
      "epoch": 2.690340909090909,
      "grad_norm": 0.19888565247689213,
      "learning_rate": 5.7368421052631575e-06,
      "loss": 0.2723,
      "step": 1894
    },
    {
      "epoch": 2.6917613636363638,
      "grad_norm": 0.22430899520552855,
      "learning_rate": 5.710526315789474e-06,
      "loss": 0.2968,
      "step": 1895
    },
    {
      "epoch": 2.6931818181818183,
      "grad_norm": 0.2092035739214887,
      "learning_rate": 5.68421052631579e-06,
      "loss": 0.2756,
      "step": 1896
    },
    {
      "epoch": 2.6946022727272725,
      "grad_norm": 0.20624397034906836,
      "learning_rate": 5.657894736842106e-06,
      "loss": 0.2807,
      "step": 1897
    },
    {
      "epoch": 2.6960227272727275,
      "grad_norm": 0.20282037368920053,
      "learning_rate": 5.631578947368421e-06,
      "loss": 0.2847,
      "step": 1898
    },
    {
      "epoch": 2.6974431818181817,
      "grad_norm": 0.3937809092710545,
      "learning_rate": 5.6052631578947374e-06,
      "loss": 0.2811,
      "step": 1899
    },
    {
      "epoch": 2.6988636363636362,
      "grad_norm": 0.20489279768807478,
      "learning_rate": 5.578947368421053e-06,
      "loss": 0.2772,
      "step": 1900
    },
    {
      "epoch": 2.700284090909091,
      "grad_norm": 0.20454845563919774,
      "learning_rate": 5.552631578947369e-06,
      "loss": 0.2817,
      "step": 1901
    },
    {
      "epoch": 2.7017045454545454,
      "grad_norm": 0.19787265352259983,
      "learning_rate": 5.526315789473684e-06,
      "loss": 0.2728,
      "step": 1902
    },
    {
      "epoch": 2.703125,
      "grad_norm": 0.2096782280921765,
      "learning_rate": 5.500000000000001e-06,
      "loss": 0.2893,
      "step": 1903
    },
    {
      "epoch": 2.7045454545454546,
      "grad_norm": 0.20095155197725387,
      "learning_rate": 5.4736842105263165e-06,
      "loss": 0.2828,
      "step": 1904
    },
    {
      "epoch": 2.705965909090909,
      "grad_norm": 0.21662198607919067,
      "learning_rate": 5.4473684210526315e-06,
      "loss": 0.2851,
      "step": 1905
    },
    {
      "epoch": 2.7073863636363638,
      "grad_norm": 0.20397341352956308,
      "learning_rate": 5.421052631578947e-06,
      "loss": 0.2787,
      "step": 1906
    },
    {
      "epoch": 2.7088068181818183,
      "grad_norm": 0.19753564981819774,
      "learning_rate": 5.394736842105263e-06,
      "loss": 0.2757,
      "step": 1907
    },
    {
      "epoch": 2.7102272727272725,
      "grad_norm": 0.20574124825848208,
      "learning_rate": 5.36842105263158e-06,
      "loss": 0.2839,
      "step": 1908
    },
    {
      "epoch": 2.7116477272727275,
      "grad_norm": 0.19635220203014092,
      "learning_rate": 5.342105263157895e-06,
      "loss": 0.2796,
      "step": 1909
    },
    {
      "epoch": 2.7130681818181817,
      "grad_norm": 0.21305864109638156,
      "learning_rate": 5.315789473684211e-06,
      "loss": 0.2838,
      "step": 1910
    },
    {
      "epoch": 2.7144886363636362,
      "grad_norm": 0.21160888405019926,
      "learning_rate": 5.289473684210526e-06,
      "loss": 0.2946,
      "step": 1911
    },
    {
      "epoch": 2.715909090909091,
      "grad_norm": 0.21098384948399923,
      "learning_rate": 5.263157894736842e-06,
      "loss": 0.2887,
      "step": 1912
    },
    {
      "epoch": 2.7173295454545454,
      "grad_norm": 0.20323798381348757,
      "learning_rate": 5.236842105263158e-06,
      "loss": 0.2817,
      "step": 1913
    },
    {
      "epoch": 2.71875,
      "grad_norm": 0.19827068984872645,
      "learning_rate": 5.210526315789474e-06,
      "loss": 0.2736,
      "step": 1914
    },
    {
      "epoch": 2.7201704545454546,
      "grad_norm": 0.2376517973070773,
      "learning_rate": 5.18421052631579e-06,
      "loss": 0.2948,
      "step": 1915
    },
    {
      "epoch": 2.721590909090909,
      "grad_norm": 0.22173768187907517,
      "learning_rate": 5.1578947368421055e-06,
      "loss": 0.2832,
      "step": 1916
    },
    {
      "epoch": 2.7230113636363638,
      "grad_norm": 0.21264453671922745,
      "learning_rate": 5.131578947368421e-06,
      "loss": 0.2864,
      "step": 1917
    },
    {
      "epoch": 2.7244318181818183,
      "grad_norm": 0.24318128829402905,
      "learning_rate": 5.105263157894737e-06,
      "loss": 0.3011,
      "step": 1918
    },
    {
      "epoch": 2.7258522727272725,
      "grad_norm": 0.21050218587491706,
      "learning_rate": 5.078947368421053e-06,
      "loss": 0.2908,
      "step": 1919
    },
    {
      "epoch": 2.7272727272727275,
      "grad_norm": 0.21190735578923311,
      "learning_rate": 5.052631578947369e-06,
      "loss": 0.2777,
      "step": 1920
    },
    {
      "epoch": 2.7286931818181817,
      "grad_norm": 0.23414278304454672,
      "learning_rate": 5.026315789473685e-06,
      "loss": 0.2917,
      "step": 1921
    },
    {
      "epoch": 2.7301136363636362,
      "grad_norm": 0.2525697196217279,
      "learning_rate": 5e-06,
      "loss": 0.2986,
      "step": 1922
    },
    {
      "epoch": 2.731534090909091,
      "grad_norm": 0.19934983543245474,
      "learning_rate": 4.973684210526316e-06,
      "loss": 0.2776,
      "step": 1923
    },
    {
      "epoch": 2.7329545454545454,
      "grad_norm": 0.21297182748896717,
      "learning_rate": 4.947368421052632e-06,
      "loss": 0.2742,
      "step": 1924
    },
    {
      "epoch": 2.734375,
      "grad_norm": 0.2293761031486547,
      "learning_rate": 4.921052631578948e-06,
      "loss": 0.2793,
      "step": 1925
    },
    {
      "epoch": 2.7357954545454546,
      "grad_norm": 0.21479233728138492,
      "learning_rate": 4.894736842105263e-06,
      "loss": 0.2835,
      "step": 1926
    },
    {
      "epoch": 2.737215909090909,
      "grad_norm": 0.20586744132448184,
      "learning_rate": 4.868421052631579e-06,
      "loss": 0.2837,
      "step": 1927
    },
    {
      "epoch": 2.7386363636363638,
      "grad_norm": 0.19076719331557926,
      "learning_rate": 4.842105263157895e-06,
      "loss": 0.275,
      "step": 1928
    },
    {
      "epoch": 2.7400568181818183,
      "grad_norm": 0.20353900517403026,
      "learning_rate": 4.815789473684211e-06,
      "loss": 0.2867,
      "step": 1929
    },
    {
      "epoch": 2.7414772727272725,
      "grad_norm": 0.20541826387015413,
      "learning_rate": 4.789473684210526e-06,
      "loss": 0.2639,
      "step": 1930
    },
    {
      "epoch": 2.7428977272727275,
      "grad_norm": 0.21281732189020275,
      "learning_rate": 4.763157894736842e-06,
      "loss": 0.2903,
      "step": 1931
    },
    {
      "epoch": 2.7443181818181817,
      "grad_norm": 0.2304138007096311,
      "learning_rate": 4.736842105263159e-06,
      "loss": 0.2861,
      "step": 1932
    },
    {
      "epoch": 2.7457386363636362,
      "grad_norm": 0.20305237256527317,
      "learning_rate": 4.7105263157894736e-06,
      "loss": 0.2942,
      "step": 1933
    },
    {
      "epoch": 2.747159090909091,
      "grad_norm": 0.20965683834973114,
      "learning_rate": 4.684210526315789e-06,
      "loss": 0.2941,
      "step": 1934
    },
    {
      "epoch": 2.7485795454545454,
      "grad_norm": 0.20846629835478905,
      "learning_rate": 4.657894736842105e-06,
      "loss": 0.2823,
      "step": 1935
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.21788987994451658,
      "learning_rate": 4.631578947368422e-06,
      "loss": 0.2888,
      "step": 1936
    },
    {
      "epoch": 2.7514204545454546,
      "grad_norm": 0.20179499717989083,
      "learning_rate": 4.605263157894737e-06,
      "loss": 0.3,
      "step": 1937
    },
    {
      "epoch": 2.752840909090909,
      "grad_norm": 0.2054050728601467,
      "learning_rate": 4.578947368421053e-06,
      "loss": 0.287,
      "step": 1938
    },
    {
      "epoch": 2.7542613636363638,
      "grad_norm": 0.22367787289532434,
      "learning_rate": 4.5526315789473685e-06,
      "loss": 0.302,
      "step": 1939
    },
    {
      "epoch": 2.7556818181818183,
      "grad_norm": 0.20067525357157467,
      "learning_rate": 4.526315789473685e-06,
      "loss": 0.2778,
      "step": 1940
    },
    {
      "epoch": 2.7571022727272725,
      "grad_norm": 0.20594322074770757,
      "learning_rate": 4.5e-06,
      "loss": 0.2844,
      "step": 1941
    },
    {
      "epoch": 2.7585227272727275,
      "grad_norm": 0.2068316538096572,
      "learning_rate": 4.473684210526316e-06,
      "loss": 0.2866,
      "step": 1942
    },
    {
      "epoch": 2.7599431818181817,
      "grad_norm": 0.21512598699086483,
      "learning_rate": 4.447368421052632e-06,
      "loss": 0.2862,
      "step": 1943
    },
    {
      "epoch": 2.7613636363636362,
      "grad_norm": 0.20885914474721298,
      "learning_rate": 4.4210526315789476e-06,
      "loss": 0.2619,
      "step": 1944
    },
    {
      "epoch": 2.762784090909091,
      "grad_norm": 0.19808330123026852,
      "learning_rate": 4.394736842105263e-06,
      "loss": 0.2732,
      "step": 1945
    },
    {
      "epoch": 2.7642045454545454,
      "grad_norm": 0.2083215537935415,
      "learning_rate": 4.368421052631579e-06,
      "loss": 0.2792,
      "step": 1946
    },
    {
      "epoch": 2.765625,
      "grad_norm": 0.19665119514949395,
      "learning_rate": 4.342105263157895e-06,
      "loss": 0.2721,
      "step": 1947
    },
    {
      "epoch": 2.7670454545454546,
      "grad_norm": 0.1919478229289917,
      "learning_rate": 4.315789473684211e-06,
      "loss": 0.2808,
      "step": 1948
    },
    {
      "epoch": 2.768465909090909,
      "grad_norm": 0.21793768415854792,
      "learning_rate": 4.289473684210527e-06,
      "loss": 0.2934,
      "step": 1949
    },
    {
      "epoch": 2.7698863636363638,
      "grad_norm": 0.19332661909766305,
      "learning_rate": 4.2631578947368425e-06,
      "loss": 0.2703,
      "step": 1950
    },
    {
      "epoch": 2.7713068181818183,
      "grad_norm": 0.21738993615710614,
      "learning_rate": 4.2368421052631575e-06,
      "loss": 0.279,
      "step": 1951
    },
    {
      "epoch": 2.7727272727272725,
      "grad_norm": 0.21898954864688688,
      "learning_rate": 4.210526315789474e-06,
      "loss": 0.2824,
      "step": 1952
    },
    {
      "epoch": 2.7741477272727275,
      "grad_norm": 0.19621221096463126,
      "learning_rate": 4.18421052631579e-06,
      "loss": 0.2832,
      "step": 1953
    },
    {
      "epoch": 2.7755681818181817,
      "grad_norm": 0.20748098635576076,
      "learning_rate": 4.157894736842106e-06,
      "loss": 0.2875,
      "step": 1954
    },
    {
      "epoch": 2.7769886363636362,
      "grad_norm": 0.20062506925523238,
      "learning_rate": 4.131578947368421e-06,
      "loss": 0.2876,
      "step": 1955
    },
    {
      "epoch": 2.778409090909091,
      "grad_norm": 0.20892021791285817,
      "learning_rate": 4.105263157894737e-06,
      "loss": 0.282,
      "step": 1956
    },
    {
      "epoch": 2.7798295454545454,
      "grad_norm": 0.2610103389347364,
      "learning_rate": 4.078947368421053e-06,
      "loss": 0.3008,
      "step": 1957
    },
    {
      "epoch": 2.78125,
      "grad_norm": 0.20074301489532062,
      "learning_rate": 4.052631578947368e-06,
      "loss": 0.2878,
      "step": 1958
    },
    {
      "epoch": 2.7826704545454546,
      "grad_norm": 0.20246012015423864,
      "learning_rate": 4.026315789473684e-06,
      "loss": 0.2846,
      "step": 1959
    },
    {
      "epoch": 2.784090909090909,
      "grad_norm": 0.21537937121146622,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.2838,
      "step": 1960
    },
    {
      "epoch": 2.7855113636363638,
      "grad_norm": 0.21258194780641027,
      "learning_rate": 3.9736842105263165e-06,
      "loss": 0.2735,
      "step": 1961
    },
    {
      "epoch": 2.7869318181818183,
      "grad_norm": 0.19053337038976553,
      "learning_rate": 3.9473684210526315e-06,
      "loss": 0.2781,
      "step": 1962
    },
    {
      "epoch": 2.7883522727272725,
      "grad_norm": 0.20686418190764394,
      "learning_rate": 3.921052631578947e-06,
      "loss": 0.2857,
      "step": 1963
    },
    {
      "epoch": 2.7897727272727275,
      "grad_norm": 0.21308611490673418,
      "learning_rate": 3.894736842105264e-06,
      "loss": 0.2941,
      "step": 1964
    },
    {
      "epoch": 2.7911931818181817,
      "grad_norm": 0.2007436115624069,
      "learning_rate": 3.868421052631579e-06,
      "loss": 0.2729,
      "step": 1965
    },
    {
      "epoch": 2.7926136363636362,
      "grad_norm": 0.20565599713538615,
      "learning_rate": 3.842105263157895e-06,
      "loss": 0.2821,
      "step": 1966
    },
    {
      "epoch": 2.794034090909091,
      "grad_norm": 0.20376091645281544,
      "learning_rate": 3.8157894736842105e-06,
      "loss": 0.2665,
      "step": 1967
    },
    {
      "epoch": 2.7954545454545454,
      "grad_norm": 0.2004821488573183,
      "learning_rate": 3.789473684210527e-06,
      "loss": 0.2805,
      "step": 1968
    },
    {
      "epoch": 2.796875,
      "grad_norm": 0.20183995982304137,
      "learning_rate": 3.763157894736842e-06,
      "loss": 0.2825,
      "step": 1969
    },
    {
      "epoch": 2.7982954545454546,
      "grad_norm": 0.21048014988231792,
      "learning_rate": 3.736842105263158e-06,
      "loss": 0.2926,
      "step": 1970
    },
    {
      "epoch": 2.799715909090909,
      "grad_norm": 0.20633837350853834,
      "learning_rate": 3.710526315789474e-06,
      "loss": 0.2628,
      "step": 1971
    },
    {
      "epoch": 2.8011363636363638,
      "grad_norm": 0.1901861087260353,
      "learning_rate": 3.6842105263157892e-06,
      "loss": 0.2768,
      "step": 1972
    },
    {
      "epoch": 2.8025568181818183,
      "grad_norm": 0.2220971014526733,
      "learning_rate": 3.6578947368421055e-06,
      "loss": 0.2846,
      "step": 1973
    },
    {
      "epoch": 2.8039772727272725,
      "grad_norm": 0.19441909907660848,
      "learning_rate": 3.6315789473684213e-06,
      "loss": 0.2812,
      "step": 1974
    },
    {
      "epoch": 2.8053977272727275,
      "grad_norm": 0.20859690799183736,
      "learning_rate": 3.605263157894737e-06,
      "loss": 0.2907,
      "step": 1975
    },
    {
      "epoch": 2.8068181818181817,
      "grad_norm": 0.22014830619859965,
      "learning_rate": 3.5789473684210525e-06,
      "loss": 0.2876,
      "step": 1976
    },
    {
      "epoch": 2.8082386363636362,
      "grad_norm": 0.23637450234018614,
      "learning_rate": 3.5526315789473687e-06,
      "loss": 0.2807,
      "step": 1977
    },
    {
      "epoch": 2.809659090909091,
      "grad_norm": 0.20602110952768035,
      "learning_rate": 3.5263157894736846e-06,
      "loss": 0.2849,
      "step": 1978
    },
    {
      "epoch": 2.8110795454545454,
      "grad_norm": 0.2245785333901269,
      "learning_rate": 3.5000000000000004e-06,
      "loss": 0.3032,
      "step": 1979
    },
    {
      "epoch": 2.8125,
      "grad_norm": 0.20309493042670132,
      "learning_rate": 3.4736842105263158e-06,
      "loss": 0.2888,
      "step": 1980
    },
    {
      "epoch": 2.8139204545454546,
      "grad_norm": 0.20250075154317773,
      "learning_rate": 3.4473684210526316e-06,
      "loss": 0.2753,
      "step": 1981
    },
    {
      "epoch": 2.815340909090909,
      "grad_norm": 0.2045473423347507,
      "learning_rate": 3.421052631578948e-06,
      "loss": 0.293,
      "step": 1982
    },
    {
      "epoch": 2.8167613636363638,
      "grad_norm": 0.2184529803407239,
      "learning_rate": 3.3947368421052632e-06,
      "loss": 0.2848,
      "step": 1983
    },
    {
      "epoch": 2.8181818181818183,
      "grad_norm": 0.19676155870828071,
      "learning_rate": 3.368421052631579e-06,
      "loss": 0.2714,
      "step": 1984
    },
    {
      "epoch": 2.8196022727272725,
      "grad_norm": 0.20949729808657366,
      "learning_rate": 3.342105263157895e-06,
      "loss": 0.2925,
      "step": 1985
    },
    {
      "epoch": 2.8210227272727275,
      "grad_norm": 0.21453380758244733,
      "learning_rate": 3.315789473684211e-06,
      "loss": 0.276,
      "step": 1986
    },
    {
      "epoch": 2.8224431818181817,
      "grad_norm": 0.1968672332232558,
      "learning_rate": 3.2894736842105265e-06,
      "loss": 0.2936,
      "step": 1987
    },
    {
      "epoch": 2.8238636363636362,
      "grad_norm": 0.21266860235868437,
      "learning_rate": 3.2631578947368423e-06,
      "loss": 0.2894,
      "step": 1988
    },
    {
      "epoch": 2.825284090909091,
      "grad_norm": 0.20585893696994773,
      "learning_rate": 3.236842105263158e-06,
      "loss": 0.2735,
      "step": 1989
    },
    {
      "epoch": 2.8267045454545454,
      "grad_norm": 0.21638445726659453,
      "learning_rate": 3.2105263157894735e-06,
      "loss": 0.2905,
      "step": 1990
    },
    {
      "epoch": 2.828125,
      "grad_norm": 0.19010842542201806,
      "learning_rate": 3.1842105263157893e-06,
      "loss": 0.2702,
      "step": 1991
    },
    {
      "epoch": 2.8295454545454546,
      "grad_norm": 0.22280584840256798,
      "learning_rate": 3.1578947368421056e-06,
      "loss": 0.287,
      "step": 1992
    },
    {
      "epoch": 2.830965909090909,
      "grad_norm": 0.20157905852127164,
      "learning_rate": 3.1315789473684214e-06,
      "loss": 0.2808,
      "step": 1993
    },
    {
      "epoch": 2.8323863636363638,
      "grad_norm": 0.19404634073491525,
      "learning_rate": 3.1052631578947372e-06,
      "loss": 0.2685,
      "step": 1994
    },
    {
      "epoch": 2.8338068181818183,
      "grad_norm": 0.19991663115839176,
      "learning_rate": 3.0789473684210526e-06,
      "loss": 0.2812,
      "step": 1995
    },
    {
      "epoch": 2.8352272727272725,
      "grad_norm": 0.19722560976198839,
      "learning_rate": 3.0526315789473684e-06,
      "loss": 0.2813,
      "step": 1996
    },
    {
      "epoch": 2.8366477272727275,
      "grad_norm": 0.19938890186096103,
      "learning_rate": 3.0263157894736843e-06,
      "loss": 0.2828,
      "step": 1997
    },
    {
      "epoch": 2.8380681818181817,
      "grad_norm": 0.19290625125124006,
      "learning_rate": 3e-06,
      "loss": 0.2822,
      "step": 1998
    },
    {
      "epoch": 2.8394886363636362,
      "grad_norm": 0.20365439840896538,
      "learning_rate": 2.973684210526316e-06,
      "loss": 0.2814,
      "step": 1999
    },
    {
      "epoch": 2.840909090909091,
      "grad_norm": 0.20376219117796,
      "learning_rate": 2.9473684210526317e-06,
      "loss": 0.2673,
      "step": 2000
    },
    {
      "epoch": 2.8423295454545454,
      "grad_norm": 0.21413260634797335,
      "learning_rate": 2.9210526315789475e-06,
      "loss": 0.2987,
      "step": 2001
    },
    {
      "epoch": 2.84375,
      "grad_norm": 0.21001146955194128,
      "learning_rate": 2.8947368421052634e-06,
      "loss": 0.2832,
      "step": 2002
    },
    {
      "epoch": 2.8451704545454546,
      "grad_norm": 0.2143530184549882,
      "learning_rate": 2.8684210526315787e-06,
      "loss": 0.2859,
      "step": 2003
    },
    {
      "epoch": 2.846590909090909,
      "grad_norm": 0.2188736086939502,
      "learning_rate": 2.842105263157895e-06,
      "loss": 0.2914,
      "step": 2004
    },
    {
      "epoch": 2.8480113636363638,
      "grad_norm": 0.20366609582229284,
      "learning_rate": 2.8157894736842104e-06,
      "loss": 0.2902,
      "step": 2005
    },
    {
      "epoch": 2.8494318181818183,
      "grad_norm": 0.19567923780033136,
      "learning_rate": 2.7894736842105266e-06,
      "loss": 0.2846,
      "step": 2006
    },
    {
      "epoch": 2.8508522727272725,
      "grad_norm": 0.19879884713150214,
      "learning_rate": 2.763157894736842e-06,
      "loss": 0.2875,
      "step": 2007
    },
    {
      "epoch": 2.8522727272727275,
      "grad_norm": 0.1975376217214479,
      "learning_rate": 2.7368421052631583e-06,
      "loss": 0.273,
      "step": 2008
    },
    {
      "epoch": 2.8536931818181817,
      "grad_norm": 0.1984231723177305,
      "learning_rate": 2.7105263157894737e-06,
      "loss": 0.2716,
      "step": 2009
    },
    {
      "epoch": 2.8551136363636362,
      "grad_norm": 0.19283098340535837,
      "learning_rate": 2.68421052631579e-06,
      "loss": 0.2811,
      "step": 2010
    },
    {
      "epoch": 2.856534090909091,
      "grad_norm": 0.19088201447900757,
      "learning_rate": 2.6578947368421053e-06,
      "loss": 0.2768,
      "step": 2011
    },
    {
      "epoch": 2.8579545454545454,
      "grad_norm": 0.20105589259522966,
      "learning_rate": 2.631578947368421e-06,
      "loss": 0.2875,
      "step": 2012
    },
    {
      "epoch": 2.859375,
      "grad_norm": 0.1970411413327219,
      "learning_rate": 2.605263157894737e-06,
      "loss": 0.2757,
      "step": 2013
    },
    {
      "epoch": 2.8607954545454546,
      "grad_norm": 0.20811771483203506,
      "learning_rate": 2.5789473684210527e-06,
      "loss": 0.2965,
      "step": 2014
    },
    {
      "epoch": 2.862215909090909,
      "grad_norm": 0.1914324844116993,
      "learning_rate": 2.5526315789473686e-06,
      "loss": 0.2915,
      "step": 2015
    },
    {
      "epoch": 2.8636363636363638,
      "grad_norm": 0.19596753488107757,
      "learning_rate": 2.5263157894736844e-06,
      "loss": 0.2889,
      "step": 2016
    },
    {
      "epoch": 2.8650568181818183,
      "grad_norm": 0.2255712802087179,
      "learning_rate": 2.5e-06,
      "loss": 0.2809,
      "step": 2017
    },
    {
      "epoch": 2.8664772727272725,
      "grad_norm": 0.18268994398336938,
      "learning_rate": 2.473684210526316e-06,
      "loss": 0.2746,
      "step": 2018
    },
    {
      "epoch": 2.8678977272727275,
      "grad_norm": 0.19234535127684443,
      "learning_rate": 2.4473684210526314e-06,
      "loss": 0.2806,
      "step": 2019
    },
    {
      "epoch": 2.8693181818181817,
      "grad_norm": 0.18887339511078519,
      "learning_rate": 2.4210526315789477e-06,
      "loss": 0.282,
      "step": 2020
    },
    {
      "epoch": 2.8707386363636362,
      "grad_norm": 0.19311644125176683,
      "learning_rate": 2.394736842105263e-06,
      "loss": 0.2575,
      "step": 2021
    },
    {
      "epoch": 2.872159090909091,
      "grad_norm": 0.21063410198823587,
      "learning_rate": 2.3684210526315793e-06,
      "loss": 0.2991,
      "step": 2022
    },
    {
      "epoch": 2.8735795454545454,
      "grad_norm": 0.2001669634643022,
      "learning_rate": 2.3421052631578947e-06,
      "loss": 0.2946,
      "step": 2023
    },
    {
      "epoch": 2.875,
      "grad_norm": 0.20484602557495404,
      "learning_rate": 2.315789473684211e-06,
      "loss": 0.2795,
      "step": 2024
    },
    {
      "epoch": 2.8764204545454546,
      "grad_norm": 0.21303454020601148,
      "learning_rate": 2.2894736842105263e-06,
      "loss": 0.3045,
      "step": 2025
    },
    {
      "epoch": 2.877840909090909,
      "grad_norm": 0.1947790494123628,
      "learning_rate": 2.2631578947368426e-06,
      "loss": 0.289,
      "step": 2026
    },
    {
      "epoch": 2.8792613636363638,
      "grad_norm": 0.1881496344735473,
      "learning_rate": 2.236842105263158e-06,
      "loss": 0.2883,
      "step": 2027
    },
    {
      "epoch": 2.8806818181818183,
      "grad_norm": 0.2113558113324988,
      "learning_rate": 2.2105263157894738e-06,
      "loss": 0.3029,
      "step": 2028
    },
    {
      "epoch": 2.8821022727272725,
      "grad_norm": 0.2000661024732429,
      "learning_rate": 2.1842105263157896e-06,
      "loss": 0.2819,
      "step": 2029
    },
    {
      "epoch": 2.8835227272727275,
      "grad_norm": 0.2071094357374062,
      "learning_rate": 2.1578947368421054e-06,
      "loss": 0.2859,
      "step": 2030
    },
    {
      "epoch": 2.8849431818181817,
      "grad_norm": 0.18780975225677882,
      "learning_rate": 2.1315789473684212e-06,
      "loss": 0.2812,
      "step": 2031
    },
    {
      "epoch": 2.8863636363636362,
      "grad_norm": 0.1975525802507232,
      "learning_rate": 2.105263157894737e-06,
      "loss": 0.2856,
      "step": 2032
    },
    {
      "epoch": 2.887784090909091,
      "grad_norm": 0.19736782560917102,
      "learning_rate": 2.078947368421053e-06,
      "loss": 0.2864,
      "step": 2033
    },
    {
      "epoch": 2.8892045454545454,
      "grad_norm": 0.19105300844482653,
      "learning_rate": 2.0526315789473687e-06,
      "loss": 0.2891,
      "step": 2034
    },
    {
      "epoch": 2.890625,
      "grad_norm": 0.2050108685111397,
      "learning_rate": 2.026315789473684e-06,
      "loss": 0.2913,
      "step": 2035
    },
    {
      "epoch": 2.8920454545454546,
      "grad_norm": 0.20956926195188966,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.2756,
      "step": 2036
    },
    {
      "epoch": 2.893465909090909,
      "grad_norm": 0.2000274541082262,
      "learning_rate": 1.9736842105263157e-06,
      "loss": 0.2941,
      "step": 2037
    },
    {
      "epoch": 2.8948863636363638,
      "grad_norm": 0.187375911973632,
      "learning_rate": 1.947368421052632e-06,
      "loss": 0.2828,
      "step": 2038
    },
    {
      "epoch": 2.8963068181818183,
      "grad_norm": 0.1914874618414179,
      "learning_rate": 1.9210526315789474e-06,
      "loss": 0.2755,
      "step": 2039
    },
    {
      "epoch": 2.8977272727272725,
      "grad_norm": 0.203592862908335,
      "learning_rate": 1.8947368421052634e-06,
      "loss": 0.2949,
      "step": 2040
    },
    {
      "epoch": 2.8991477272727275,
      "grad_norm": 0.20433175271579576,
      "learning_rate": 1.868421052631579e-06,
      "loss": 0.2858,
      "step": 2041
    },
    {
      "epoch": 2.9005681818181817,
      "grad_norm": 0.20163099846205548,
      "learning_rate": 1.8421052631578946e-06,
      "loss": 0.2893,
      "step": 2042
    },
    {
      "epoch": 2.9019886363636362,
      "grad_norm": 0.19247483859414624,
      "learning_rate": 1.8157894736842106e-06,
      "loss": 0.2897,
      "step": 2043
    },
    {
      "epoch": 2.903409090909091,
      "grad_norm": 0.18808182783223734,
      "learning_rate": 1.7894736842105262e-06,
      "loss": 0.2747,
      "step": 2044
    },
    {
      "epoch": 2.9048295454545454,
      "grad_norm": 0.19495756320795976,
      "learning_rate": 1.7631578947368423e-06,
      "loss": 0.2921,
      "step": 2045
    },
    {
      "epoch": 2.90625,
      "grad_norm": 0.18369377957942246,
      "learning_rate": 1.7368421052631579e-06,
      "loss": 0.2765,
      "step": 2046
    },
    {
      "epoch": 2.9076704545454546,
      "grad_norm": 0.1961532283966688,
      "learning_rate": 1.710526315789474e-06,
      "loss": 0.2876,
      "step": 2047
    },
    {
      "epoch": 2.909090909090909,
      "grad_norm": 0.22350119076316433,
      "learning_rate": 1.6842105263157895e-06,
      "loss": 0.2755,
      "step": 2048
    },
    {
      "epoch": 2.9105113636363638,
      "grad_norm": 0.18811828437526543,
      "learning_rate": 1.6578947368421056e-06,
      "loss": 0.2762,
      "step": 2049
    },
    {
      "epoch": 2.9119318181818183,
      "grad_norm": 0.19789661645295165,
      "learning_rate": 1.6315789473684212e-06,
      "loss": 0.2814,
      "step": 2050
    },
    {
      "epoch": 2.9133522727272725,
      "grad_norm": 0.2095640003211725,
      "learning_rate": 1.6052631578947368e-06,
      "loss": 0.2848,
      "step": 2051
    },
    {
      "epoch": 2.9147727272727275,
      "grad_norm": 0.18985298360197234,
      "learning_rate": 1.5789473684210528e-06,
      "loss": 0.2803,
      "step": 2052
    },
    {
      "epoch": 2.9161931818181817,
      "grad_norm": 0.1882727159885737,
      "learning_rate": 1.5526315789473686e-06,
      "loss": 0.2762,
      "step": 2053
    },
    {
      "epoch": 2.9176136363636362,
      "grad_norm": 0.232186825408911,
      "learning_rate": 1.5263157894736842e-06,
      "loss": 0.2803,
      "step": 2054
    },
    {
      "epoch": 2.919034090909091,
      "grad_norm": 0.19451564807357025,
      "learning_rate": 1.5e-06,
      "loss": 0.2852,
      "step": 2055
    },
    {
      "epoch": 2.9204545454545454,
      "grad_norm": 0.19491283823539432,
      "learning_rate": 1.4736842105263159e-06,
      "loss": 0.2873,
      "step": 2056
    },
    {
      "epoch": 2.921875,
      "grad_norm": 0.19121281421006264,
      "learning_rate": 1.4473684210526317e-06,
      "loss": 0.2863,
      "step": 2057
    },
    {
      "epoch": 2.9232954545454546,
      "grad_norm": 0.1999151019149978,
      "learning_rate": 1.4210526315789475e-06,
      "loss": 0.2822,
      "step": 2058
    },
    {
      "epoch": 2.924715909090909,
      "grad_norm": 0.1992298250671759,
      "learning_rate": 1.3947368421052633e-06,
      "loss": 0.2764,
      "step": 2059
    },
    {
      "epoch": 2.9261363636363638,
      "grad_norm": 0.19278939598482114,
      "learning_rate": 1.3684210526315791e-06,
      "loss": 0.2848,
      "step": 2060
    },
    {
      "epoch": 2.9275568181818183,
      "grad_norm": 0.19680469105279094,
      "learning_rate": 1.342105263157895e-06,
      "loss": 0.2937,
      "step": 2061
    },
    {
      "epoch": 2.9289772727272725,
      "grad_norm": 0.20797347573712824,
      "learning_rate": 1.3157894736842106e-06,
      "loss": 0.299,
      "step": 2062
    },
    {
      "epoch": 2.9303977272727275,
      "grad_norm": 0.194246976615525,
      "learning_rate": 1.2894736842105264e-06,
      "loss": 0.2736,
      "step": 2063
    },
    {
      "epoch": 2.9318181818181817,
      "grad_norm": 0.2023729617580181,
      "learning_rate": 1.2631578947368422e-06,
      "loss": 0.287,
      "step": 2064
    },
    {
      "epoch": 2.9332386363636362,
      "grad_norm": 0.185729543831043,
      "learning_rate": 1.236842105263158e-06,
      "loss": 0.2821,
      "step": 2065
    },
    {
      "epoch": 2.934659090909091,
      "grad_norm": 0.19553671162826516,
      "learning_rate": 1.2105263157894738e-06,
      "loss": 0.28,
      "step": 2066
    },
    {
      "epoch": 2.9360795454545454,
      "grad_norm": 0.19878750719658708,
      "learning_rate": 1.1842105263157896e-06,
      "loss": 0.2861,
      "step": 2067
    },
    {
      "epoch": 2.9375,
      "grad_norm": 0.18672676007331623,
      "learning_rate": 1.1578947368421055e-06,
      "loss": 0.2907,
      "step": 2068
    },
    {
      "epoch": 2.9389204545454546,
      "grad_norm": 0.20040847990606275,
      "learning_rate": 1.1315789473684213e-06,
      "loss": 0.2895,
      "step": 2069
    },
    {
      "epoch": 2.940340909090909,
      "grad_norm": 0.19906118356822988,
      "learning_rate": 1.1052631578947369e-06,
      "loss": 0.3006,
      "step": 2070
    },
    {
      "epoch": 2.9417613636363638,
      "grad_norm": 0.18762849759297928,
      "learning_rate": 1.0789473684210527e-06,
      "loss": 0.2834,
      "step": 2071
    },
    {
      "epoch": 2.9431818181818183,
      "grad_norm": 0.18401755548741236,
      "learning_rate": 1.0526315789473685e-06,
      "loss": 0.2733,
      "step": 2072
    },
    {
      "epoch": 2.9446022727272725,
      "grad_norm": 0.19673662807122536,
      "learning_rate": 1.0263157894736843e-06,
      "loss": 0.2939,
      "step": 2073
    },
    {
      "epoch": 2.9460227272727275,
      "grad_norm": 0.1996559688873919,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.2872,
      "step": 2074
    },
    {
      "epoch": 2.9474431818181817,
      "grad_norm": 0.19339809201794914,
      "learning_rate": 9.73684210526316e-07,
      "loss": 0.2781,
      "step": 2075
    },
    {
      "epoch": 2.9488636363636362,
      "grad_norm": 0.1916205614419561,
      "learning_rate": 9.473684210526317e-07,
      "loss": 0.2935,
      "step": 2076
    },
    {
      "epoch": 2.950284090909091,
      "grad_norm": 0.19186322907136866,
      "learning_rate": 9.210526315789473e-07,
      "loss": 0.2924,
      "step": 2077
    },
    {
      "epoch": 2.9517045454545454,
      "grad_norm": 0.1970059642263826,
      "learning_rate": 8.947368421052631e-07,
      "loss": 0.2786,
      "step": 2078
    },
    {
      "epoch": 2.953125,
      "grad_norm": 0.20092664616405115,
      "learning_rate": 8.684210526315789e-07,
      "loss": 0.2867,
      "step": 2079
    },
    {
      "epoch": 2.9545454545454546,
      "grad_norm": 0.18763021770238275,
      "learning_rate": 8.421052631578948e-07,
      "loss": 0.2858,
      "step": 2080
    },
    {
      "epoch": 2.955965909090909,
      "grad_norm": 0.18526415545453082,
      "learning_rate": 8.157894736842106e-07,
      "loss": 0.2873,
      "step": 2081
    },
    {
      "epoch": 2.9573863636363638,
      "grad_norm": 0.19872727378907906,
      "learning_rate": 7.894736842105264e-07,
      "loss": 0.2974,
      "step": 2082
    },
    {
      "epoch": 2.9588068181818183,
      "grad_norm": 0.20447015927371373,
      "learning_rate": 7.631578947368421e-07,
      "loss": 0.2919,
      "step": 2083
    },
    {
      "epoch": 2.9602272727272725,
      "grad_norm": 0.19410410066955255,
      "learning_rate": 7.368421052631579e-07,
      "loss": 0.2857,
      "step": 2084
    },
    {
      "epoch": 2.9616477272727275,
      "grad_norm": 0.18299881906080168,
      "learning_rate": 7.105263157894737e-07,
      "loss": 0.2771,
      "step": 2085
    },
    {
      "epoch": 2.9630681818181817,
      "grad_norm": 0.18978268711113566,
      "learning_rate": 6.842105263157896e-07,
      "loss": 0.2836,
      "step": 2086
    },
    {
      "epoch": 2.9644886363636362,
      "grad_norm": 0.19098261881408551,
      "learning_rate": 6.578947368421053e-07,
      "loss": 0.275,
      "step": 2087
    },
    {
      "epoch": 2.965909090909091,
      "grad_norm": 0.18270139296441668,
      "learning_rate": 6.315789473684211e-07,
      "loss": 0.2794,
      "step": 2088
    },
    {
      "epoch": 2.9673295454545454,
      "grad_norm": 0.18578835537302887,
      "learning_rate": 6.052631578947369e-07,
      "loss": 0.2834,
      "step": 2089
    },
    {
      "epoch": 2.96875,
      "grad_norm": 0.19491061861208503,
      "learning_rate": 5.789473684210527e-07,
      "loss": 0.2886,
      "step": 2090
    },
    {
      "epoch": 2.9701704545454546,
      "grad_norm": 0.18831817514270036,
      "learning_rate": 5.526315789473684e-07,
      "loss": 0.2765,
      "step": 2091
    },
    {
      "epoch": 2.971590909090909,
      "grad_norm": 0.20683709515479873,
      "learning_rate": 5.263157894736843e-07,
      "loss": 0.2826,
      "step": 2092
    },
    {
      "epoch": 2.9730113636363638,
      "grad_norm": 0.20210877509098826,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.291,
      "step": 2093
    },
    {
      "epoch": 2.9744318181818183,
      "grad_norm": 0.1925765157493675,
      "learning_rate": 4.7368421052631585e-07,
      "loss": 0.2883,
      "step": 2094
    },
    {
      "epoch": 2.9758522727272725,
      "grad_norm": 0.18335082571509506,
      "learning_rate": 4.4736842105263156e-07,
      "loss": 0.2678,
      "step": 2095
    },
    {
      "epoch": 2.9772727272727275,
      "grad_norm": 0.1954818718722415,
      "learning_rate": 4.210526315789474e-07,
      "loss": 0.2748,
      "step": 2096
    },
    {
      "epoch": 2.9786931818181817,
      "grad_norm": 0.20904335565640544,
      "learning_rate": 3.947368421052632e-07,
      "loss": 0.2771,
      "step": 2097
    },
    {
      "epoch": 2.9801136363636362,
      "grad_norm": 0.1909605640138579,
      "learning_rate": 3.6842105263157896e-07,
      "loss": 0.2831,
      "step": 2098
    },
    {
      "epoch": 2.981534090909091,
      "grad_norm": 0.4094959713752424,
      "learning_rate": 3.421052631578948e-07,
      "loss": 0.2858,
      "step": 2099
    },
    {
      "epoch": 2.9829545454545454,
      "grad_norm": 0.19780613116057918,
      "learning_rate": 3.1578947368421055e-07,
      "loss": 0.2805,
      "step": 2100
    },
    {
      "epoch": 2.984375,
      "grad_norm": 0.1979485917345104,
      "learning_rate": 2.8947368421052637e-07,
      "loss": 0.3051,
      "step": 2101
    },
    {
      "epoch": 2.9857954545454546,
      "grad_norm": 0.1974265392405128,
      "learning_rate": 2.6315789473684213e-07,
      "loss": 0.2904,
      "step": 2102
    },
    {
      "epoch": 2.987215909090909,
      "grad_norm": 0.19687702926366793,
      "learning_rate": 2.3684210526315792e-07,
      "loss": 0.2847,
      "step": 2103
    },
    {
      "epoch": 2.9886363636363638,
      "grad_norm": 0.18457677995670124,
      "learning_rate": 2.105263157894737e-07,
      "loss": 0.2639,
      "step": 2104
    },
    {
      "epoch": 2.9900568181818183,
      "grad_norm": 0.18581907355108282,
      "learning_rate": 1.8421052631578948e-07,
      "loss": 0.2845,
      "step": 2105
    },
    {
      "epoch": 2.9914772727272725,
      "grad_norm": 0.20939886277038056,
      "learning_rate": 1.5789473684210527e-07,
      "loss": 0.287,
      "step": 2106
    },
    {
      "epoch": 2.9928977272727275,
      "grad_norm": 0.18292060179770495,
      "learning_rate": 1.3157894736842107e-07,
      "loss": 0.2695,
      "step": 2107
    },
    {
      "epoch": 2.9943181818181817,
      "grad_norm": 0.19426554688334796,
      "learning_rate": 1.0526315789473685e-07,
      "loss": 0.2816,
      "step": 2108
    },
    {
      "epoch": 2.9957386363636362,
      "grad_norm": 0.1909226201117013,
      "learning_rate": 7.894736842105264e-08,
      "loss": 0.2886,
      "step": 2109
    },
    {
      "epoch": 2.997159090909091,
      "grad_norm": 0.1867495400167528,
      "learning_rate": 5.263157894736842e-08,
      "loss": 0.2903,
      "step": 2110
    },
    {
      "epoch": 2.9985795454545454,
      "grad_norm": 0.19384411950991948,
      "learning_rate": 2.631578947368421e-08,
      "loss": 0.2907,
      "step": 2111
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.1958539394313963,
      "learning_rate": 0.0,
      "loss": 0.2491,
      "step": 2112
    },
    {
      "epoch": 3.0,
      "step": 2112,
      "total_flos": 1.8047985149710172e+18,
      "train_loss": 0.4428125687761966,
      "train_runtime": 122652.2385,
      "train_samples_per_second": 0.275,
      "train_steps_per_second": 0.017
    }
  ],
  "logging_steps": 1,
  "max_steps": 2112,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "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": 1.8047985149710172e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}