{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5745257452574526,
  "eval_steps": 500,
  "global_step": 1484,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00038714672861014324,
      "grad_norm": 0.048257023096084595,
      "learning_rate": 2e-05,
      "loss": 0.0394,
      "step": 1
    },
    {
      "epoch": 0.0007742934572202865,
      "grad_norm": 0.048200540244579315,
      "learning_rate": 4e-05,
      "loss": 0.048,
      "step": 2
    },
    {
      "epoch": 0.0011614401858304297,
      "grad_norm": 0.10289892554283142,
      "learning_rate": 6e-05,
      "loss": 0.0864,
      "step": 3
    },
    {
      "epoch": 0.001548586914440573,
      "grad_norm": 0.10379760712385178,
      "learning_rate": 8e-05,
      "loss": 0.0887,
      "step": 4
    },
    {
      "epoch": 0.0019357336430507162,
      "grad_norm": 0.14089909195899963,
      "learning_rate": 0.0001,
      "loss": 0.1072,
      "step": 5
    },
    {
      "epoch": 0.0023228803716608595,
      "grad_norm": 0.16513778269290924,
      "learning_rate": 9.999988720152121e-05,
      "loss": 0.1346,
      "step": 6
    },
    {
      "epoch": 0.0027100271002710027,
      "grad_norm": 0.17860881984233856,
      "learning_rate": 9.999954880659377e-05,
      "loss": 0.1306,
      "step": 7
    },
    {
      "epoch": 0.003097173828881146,
      "grad_norm": 0.2414817363023758,
      "learning_rate": 9.999898481674448e-05,
      "loss": 0.1593,
      "step": 8
    },
    {
      "epoch": 0.003484320557491289,
      "grad_norm": 0.2995632588863373,
      "learning_rate": 9.999819523451806e-05,
      "loss": 0.1684,
      "step": 9
    },
    {
      "epoch": 0.0038714672861014324,
      "grad_norm": 0.2831806242465973,
      "learning_rate": 9.999718006347704e-05,
      "loss": 0.1538,
      "step": 10
    },
    {
      "epoch": 0.004258614014711576,
      "grad_norm": 0.3746040463447571,
      "learning_rate": 9.999593930820181e-05,
      "loss": 0.1573,
      "step": 11
    },
    {
      "epoch": 0.004645760743321719,
      "grad_norm": 0.5143625736236572,
      "learning_rate": 9.999447297429058e-05,
      "loss": 0.1596,
      "step": 12
    },
    {
      "epoch": 0.005032907471931862,
      "grad_norm": 0.5252905488014221,
      "learning_rate": 9.999278106835936e-05,
      "loss": 0.1503,
      "step": 13
    },
    {
      "epoch": 0.005420054200542005,
      "grad_norm": 0.43592771887779236,
      "learning_rate": 9.999086359804195e-05,
      "loss": 0.1396,
      "step": 14
    },
    {
      "epoch": 0.005807200929152149,
      "grad_norm": 0.7996945381164551,
      "learning_rate": 9.998872057198982e-05,
      "loss": 0.1226,
      "step": 15
    },
    {
      "epoch": 0.006194347657762292,
      "grad_norm": 0.8167794942855835,
      "learning_rate": 9.998635199987219e-05,
      "loss": 0.1368,
      "step": 16
    },
    {
      "epoch": 0.006581494386372435,
      "grad_norm": 0.8105260729789734,
      "learning_rate": 9.998375789237593e-05,
      "loss": 0.1199,
      "step": 17
    },
    {
      "epoch": 0.006968641114982578,
      "grad_norm": 0.7661561369895935,
      "learning_rate": 9.998093826120548e-05,
      "loss": 0.1011,
      "step": 18
    },
    {
      "epoch": 0.007355787843592722,
      "grad_norm": 0.6599623560905457,
      "learning_rate": 9.997789311908284e-05,
      "loss": 0.0966,
      "step": 19
    },
    {
      "epoch": 0.007742934572202865,
      "grad_norm": 0.8486294150352478,
      "learning_rate": 9.997462247974752e-05,
      "loss": 0.0979,
      "step": 20
    },
    {
      "epoch": 0.008130081300813009,
      "grad_norm": 0.6950157284736633,
      "learning_rate": 9.997112635795643e-05,
      "loss": 0.0894,
      "step": 21
    },
    {
      "epoch": 0.008517228029423151,
      "grad_norm": 0.4815932512283325,
      "learning_rate": 9.996740476948385e-05,
      "loss": 0.0649,
      "step": 22
    },
    {
      "epoch": 0.008904374758033295,
      "grad_norm": 0.41461682319641113,
      "learning_rate": 9.996345773112139e-05,
      "loss": 0.0704,
      "step": 23
    },
    {
      "epoch": 0.009291521486643438,
      "grad_norm": 0.42949068546295166,
      "learning_rate": 9.995928526067784e-05,
      "loss": 0.0715,
      "step": 24
    },
    {
      "epoch": 0.009678668215253582,
      "grad_norm": 0.3920012414455414,
      "learning_rate": 9.99548873769791e-05,
      "loss": 0.048,
      "step": 25
    },
    {
      "epoch": 0.010065814943863724,
      "grad_norm": 0.261152058839798,
      "learning_rate": 9.995026409986821e-05,
      "loss": 0.0343,
      "step": 26
    },
    {
      "epoch": 0.010452961672473868,
      "grad_norm": 0.3788264989852905,
      "learning_rate": 9.994541545020509e-05,
      "loss": 0.0403,
      "step": 27
    },
    {
      "epoch": 0.01084010840108401,
      "grad_norm": 0.8691251873970032,
      "learning_rate": 9.994034144986653e-05,
      "loss": 0.063,
      "step": 28
    },
    {
      "epoch": 0.011227255129694155,
      "grad_norm": 0.5656071305274963,
      "learning_rate": 9.993504212174613e-05,
      "loss": 0.0415,
      "step": 29
    },
    {
      "epoch": 0.011614401858304297,
      "grad_norm": 0.42919209599494934,
      "learning_rate": 9.992951748975412e-05,
      "loss": 0.0229,
      "step": 30
    },
    {
      "epoch": 0.012001548586914441,
      "grad_norm": 0.46598193049430847,
      "learning_rate": 9.992376757881735e-05,
      "loss": 0.0208,
      "step": 31
    },
    {
      "epoch": 0.012388695315524584,
      "grad_norm": 0.5426745414733887,
      "learning_rate": 9.991779241487899e-05,
      "loss": 0.0188,
      "step": 32
    },
    {
      "epoch": 0.012775842044134728,
      "grad_norm": 0.8520743250846863,
      "learning_rate": 9.991159202489871e-05,
      "loss": 0.0278,
      "step": 33
    },
    {
      "epoch": 0.01316298877274487,
      "grad_norm": 0.6999525427818298,
      "learning_rate": 9.990516643685222e-05,
      "loss": 0.0145,
      "step": 34
    },
    {
      "epoch": 0.013550135501355014,
      "grad_norm": 1.101209282875061,
      "learning_rate": 9.989851567973139e-05,
      "loss": 0.0096,
      "step": 35
    },
    {
      "epoch": 0.013937282229965157,
      "grad_norm": 1.42173433303833,
      "learning_rate": 9.989163978354408e-05,
      "loss": 0.01,
      "step": 36
    },
    {
      "epoch": 0.014324428958575301,
      "grad_norm": 0.1809816062450409,
      "learning_rate": 9.988453877931386e-05,
      "loss": 0.0026,
      "step": 37
    },
    {
      "epoch": 0.014711575687185443,
      "grad_norm": 0.5789023637771606,
      "learning_rate": 9.987721269908006e-05,
      "loss": 0.0264,
      "step": 38
    },
    {
      "epoch": 0.015098722415795587,
      "grad_norm": 1.4514374732971191,
      "learning_rate": 9.98696615758975e-05,
      "loss": 0.0197,
      "step": 39
    },
    {
      "epoch": 0.01548586914440573,
      "grad_norm": 1.6745890378952026,
      "learning_rate": 9.986188544383639e-05,
      "loss": 0.0398,
      "step": 40
    },
    {
      "epoch": 0.015873015873015872,
      "grad_norm": 0.31723207235336304,
      "learning_rate": 9.985388433798215e-05,
      "loss": 0.0207,
      "step": 41
    },
    {
      "epoch": 0.016260162601626018,
      "grad_norm": 0.8627777695655823,
      "learning_rate": 9.984565829443531e-05,
      "loss": 0.0182,
      "step": 42
    },
    {
      "epoch": 0.01664730933023616,
      "grad_norm": 0.023634975776076317,
      "learning_rate": 9.983720735031125e-05,
      "loss": 0.0004,
      "step": 43
    },
    {
      "epoch": 0.017034456058846303,
      "grad_norm": 0.45953628420829773,
      "learning_rate": 9.982853154374014e-05,
      "loss": 0.0111,
      "step": 44
    },
    {
      "epoch": 0.017421602787456445,
      "grad_norm": 0.2831597328186035,
      "learning_rate": 9.98196309138667e-05,
      "loss": 0.0109,
      "step": 45
    },
    {
      "epoch": 0.01780874951606659,
      "grad_norm": 0.7697696089744568,
      "learning_rate": 9.981050550084998e-05,
      "loss": 0.0427,
      "step": 46
    },
    {
      "epoch": 0.018195896244676733,
      "grad_norm": 0.5054185390472412,
      "learning_rate": 9.980115534586335e-05,
      "loss": 0.0093,
      "step": 47
    },
    {
      "epoch": 0.018583042973286876,
      "grad_norm": 0.16788819432258606,
      "learning_rate": 9.979158049109407e-05,
      "loss": 0.0024,
      "step": 48
    },
    {
      "epoch": 0.018970189701897018,
      "grad_norm": 2.8662874698638916,
      "learning_rate": 9.978178097974337e-05,
      "loss": 0.0919,
      "step": 49
    },
    {
      "epoch": 0.019357336430507164,
      "grad_norm": 2.2201218605041504,
      "learning_rate": 9.977175685602602e-05,
      "loss": 0.2433,
      "step": 50
    },
    {
      "epoch": 0.019744483159117306,
      "grad_norm": 0.08052156120538712,
      "learning_rate": 9.976150816517024e-05,
      "loss": 0.0113,
      "step": 51
    },
    {
      "epoch": 0.02013162988772745,
      "grad_norm": 0.10142353922128677,
      "learning_rate": 9.975103495341753e-05,
      "loss": 0.0095,
      "step": 52
    },
    {
      "epoch": 0.02051877661633759,
      "grad_norm": 0.14654894173145294,
      "learning_rate": 9.974033726802237e-05,
      "loss": 0.0141,
      "step": 53
    },
    {
      "epoch": 0.020905923344947737,
      "grad_norm": 0.19414274394512177,
      "learning_rate": 9.972941515725206e-05,
      "loss": 0.0129,
      "step": 54
    },
    {
      "epoch": 0.02129307007355788,
      "grad_norm": 0.24451220035552979,
      "learning_rate": 9.971826867038651e-05,
      "loss": 0.0141,
      "step": 55
    },
    {
      "epoch": 0.02168021680216802,
      "grad_norm": 0.1221214234828949,
      "learning_rate": 9.970689785771798e-05,
      "loss": 0.0025,
      "step": 56
    },
    {
      "epoch": 0.022067363530778164,
      "grad_norm": 0.4560745656490326,
      "learning_rate": 9.96953027705509e-05,
      "loss": 0.0127,
      "step": 57
    },
    {
      "epoch": 0.02245451025938831,
      "grad_norm": 0.2530369758605957,
      "learning_rate": 9.968348346120159e-05,
      "loss": 0.0148,
      "step": 58
    },
    {
      "epoch": 0.022841656987998452,
      "grad_norm": 0.7455848455429077,
      "learning_rate": 9.967143998299803e-05,
      "loss": 0.0069,
      "step": 59
    },
    {
      "epoch": 0.023228803716608595,
      "grad_norm": 0.10115598142147064,
      "learning_rate": 9.965917239027972e-05,
      "loss": 0.0094,
      "step": 60
    },
    {
      "epoch": 0.023615950445218737,
      "grad_norm": 0.15474660694599152,
      "learning_rate": 9.964668073839724e-05,
      "loss": 0.0079,
      "step": 61
    },
    {
      "epoch": 0.024003097173828883,
      "grad_norm": 0.020637422800064087,
      "learning_rate": 9.963396508371218e-05,
      "loss": 0.0004,
      "step": 62
    },
    {
      "epoch": 0.024390243902439025,
      "grad_norm": 0.14796803891658783,
      "learning_rate": 9.96210254835968e-05,
      "loss": 0.001,
      "step": 63
    },
    {
      "epoch": 0.024777390631049168,
      "grad_norm": 0.28956761956214905,
      "learning_rate": 9.960786199643377e-05,
      "loss": 0.0013,
      "step": 64
    },
    {
      "epoch": 0.02516453735965931,
      "grad_norm": 0.020133867859840393,
      "learning_rate": 9.959447468161598e-05,
      "loss": 0.0002,
      "step": 65
    },
    {
      "epoch": 0.025551684088269456,
      "grad_norm": 0.43387070298194885,
      "learning_rate": 9.958086359954616e-05,
      "loss": 0.0038,
      "step": 66
    },
    {
      "epoch": 0.025938830816879598,
      "grad_norm": 0.12942586839199066,
      "learning_rate": 9.956702881163669e-05,
      "loss": 0.0018,
      "step": 67
    },
    {
      "epoch": 0.02632597754548974,
      "grad_norm": 0.24969282746315002,
      "learning_rate": 9.955297038030927e-05,
      "loss": 0.0067,
      "step": 68
    },
    {
      "epoch": 0.026713124274099883,
      "grad_norm": 0.06376656889915466,
      "learning_rate": 9.953868836899472e-05,
      "loss": 0.0008,
      "step": 69
    },
    {
      "epoch": 0.02710027100271003,
      "grad_norm": 0.03575746715068817,
      "learning_rate": 9.952418284213257e-05,
      "loss": 0.0009,
      "step": 70
    },
    {
      "epoch": 0.02748741773132017,
      "grad_norm": 0.19091981649398804,
      "learning_rate": 9.95094538651709e-05,
      "loss": 0.0046,
      "step": 71
    },
    {
      "epoch": 0.027874564459930314,
      "grad_norm": 0.03124951757490635,
      "learning_rate": 9.949450150456596e-05,
      "loss": 0.0005,
      "step": 72
    },
    {
      "epoch": 0.028261711188540456,
      "grad_norm": 0.10794883221387863,
      "learning_rate": 9.947932582778188e-05,
      "loss": 0.0009,
      "step": 73
    },
    {
      "epoch": 0.028648857917150602,
      "grad_norm": 0.054995935410261154,
      "learning_rate": 9.946392690329039e-05,
      "loss": 0.0006,
      "step": 74
    },
    {
      "epoch": 0.029036004645760744,
      "grad_norm": 0.12434786558151245,
      "learning_rate": 9.944830480057049e-05,
      "loss": 0.0015,
      "step": 75
    },
    {
      "epoch": 0.029423151374370887,
      "grad_norm": 0.36718836426734924,
      "learning_rate": 9.943245959010819e-05,
      "loss": 0.0123,
      "step": 76
    },
    {
      "epoch": 0.02981029810298103,
      "grad_norm": 0.002867339877411723,
      "learning_rate": 9.941639134339608e-05,
      "loss": 0.0001,
      "step": 77
    },
    {
      "epoch": 0.030197444831591175,
      "grad_norm": 0.02242063172161579,
      "learning_rate": 9.940010013293313e-05,
      "loss": 0.0003,
      "step": 78
    },
    {
      "epoch": 0.030584591560201317,
      "grad_norm": 0.6994889378547668,
      "learning_rate": 9.938358603222429e-05,
      "loss": 0.005,
      "step": 79
    },
    {
      "epoch": 0.03097173828881146,
      "grad_norm": 0.06029118597507477,
      "learning_rate": 9.936684911578018e-05,
      "loss": 0.0009,
      "step": 80
    },
    {
      "epoch": 0.0313588850174216,
      "grad_norm": 0.43654757738113403,
      "learning_rate": 9.934988945911674e-05,
      "loss": 0.0041,
      "step": 81
    },
    {
      "epoch": 0.031746031746031744,
      "grad_norm": 1.1972216367721558,
      "learning_rate": 9.93327071387549e-05,
      "loss": 0.0098,
      "step": 82
    },
    {
      "epoch": 0.03213317847464189,
      "grad_norm": 0.008753047324717045,
      "learning_rate": 9.931530223222027e-05,
      "loss": 0.0001,
      "step": 83
    },
    {
      "epoch": 0.032520325203252036,
      "grad_norm": 0.019141845405101776,
      "learning_rate": 9.929767481804271e-05,
      "loss": 0.0003,
      "step": 84
    },
    {
      "epoch": 0.03290747193186218,
      "grad_norm": 0.13906456530094147,
      "learning_rate": 9.927982497575605e-05,
      "loss": 0.0017,
      "step": 85
    },
    {
      "epoch": 0.03329461866047232,
      "grad_norm": 0.018300075083971024,
      "learning_rate": 9.926175278589769e-05,
      "loss": 0.0004,
      "step": 86
    },
    {
      "epoch": 0.03368176538908246,
      "grad_norm": 0.4637622535228729,
      "learning_rate": 9.924345833000824e-05,
      "loss": 0.0056,
      "step": 87
    },
    {
      "epoch": 0.034068912117692605,
      "grad_norm": 0.00441858172416687,
      "learning_rate": 9.922494169063119e-05,
      "loss": 0.0001,
      "step": 88
    },
    {
      "epoch": 0.03445605884630275,
      "grad_norm": 0.0403720885515213,
      "learning_rate": 9.920620295131247e-05,
      "loss": 0.0007,
      "step": 89
    },
    {
      "epoch": 0.03484320557491289,
      "grad_norm": 0.17813202738761902,
      "learning_rate": 9.918724219660013e-05,
      "loss": 0.0015,
      "step": 90
    },
    {
      "epoch": 0.03523035230352303,
      "grad_norm": 0.058298081159591675,
      "learning_rate": 9.916805951204397e-05,
      "loss": 0.0004,
      "step": 91
    },
    {
      "epoch": 0.03561749903213318,
      "grad_norm": 0.29420581459999084,
      "learning_rate": 9.91486549841951e-05,
      "loss": 0.0176,
      "step": 92
    },
    {
      "epoch": 0.036004645760743324,
      "grad_norm": 0.005454974714666605,
      "learning_rate": 9.912902870060552e-05,
      "loss": 0.0001,
      "step": 93
    },
    {
      "epoch": 0.03639179248935347,
      "grad_norm": 0.01605985127389431,
      "learning_rate": 9.910918074982787e-05,
      "loss": 0.0002,
      "step": 94
    },
    {
      "epoch": 0.03677893921796361,
      "grad_norm": 0.08237680792808533,
      "learning_rate": 9.908911122141486e-05,
      "loss": 0.0011,
      "step": 95
    },
    {
      "epoch": 0.03716608594657375,
      "grad_norm": 0.6426149606704712,
      "learning_rate": 9.906882020591903e-05,
      "loss": 0.0214,
      "step": 96
    },
    {
      "epoch": 0.037553232675183894,
      "grad_norm": 0.21502873301506042,
      "learning_rate": 9.904830779489216e-05,
      "loss": 0.0019,
      "step": 97
    },
    {
      "epoch": 0.037940379403794036,
      "grad_norm": 0.7887445092201233,
      "learning_rate": 9.902757408088503e-05,
      "loss": 0.0369,
      "step": 98
    },
    {
      "epoch": 0.03832752613240418,
      "grad_norm": 0.013632730580866337,
      "learning_rate": 9.900661915744689e-05,
      "loss": 0.0001,
      "step": 99
    },
    {
      "epoch": 0.03871467286101433,
      "grad_norm": 5.337656021118164,
      "learning_rate": 9.898544311912508e-05,
      "loss": 0.163,
      "step": 100
    },
    {
      "epoch": 0.03910181958962447,
      "grad_norm": 0.07259741425514221,
      "learning_rate": 9.896404606146456e-05,
      "loss": 0.0103,
      "step": 101
    },
    {
      "epoch": 0.03948896631823461,
      "grad_norm": 0.1117100939154625,
      "learning_rate": 9.894242808100762e-05,
      "loss": 0.0058,
      "step": 102
    },
    {
      "epoch": 0.039876113046844755,
      "grad_norm": 0.15869347751140594,
      "learning_rate": 9.892058927529322e-05,
      "loss": 0.009,
      "step": 103
    },
    {
      "epoch": 0.0402632597754549,
      "grad_norm": 0.15626105666160583,
      "learning_rate": 9.889852974285673e-05,
      "loss": 0.0049,
      "step": 104
    },
    {
      "epoch": 0.04065040650406504,
      "grad_norm": 0.20995132625102997,
      "learning_rate": 9.887624958322946e-05,
      "loss": 0.0105,
      "step": 105
    },
    {
      "epoch": 0.04103755323267518,
      "grad_norm": 0.1864119917154312,
      "learning_rate": 9.885374889693809e-05,
      "loss": 0.0128,
      "step": 106
    },
    {
      "epoch": 0.041424699961285325,
      "grad_norm": 0.16312390565872192,
      "learning_rate": 9.883102778550434e-05,
      "loss": 0.0106,
      "step": 107
    },
    {
      "epoch": 0.041811846689895474,
      "grad_norm": 0.07751648128032684,
      "learning_rate": 9.880808635144452e-05,
      "loss": 0.0017,
      "step": 108
    },
    {
      "epoch": 0.042198993418505616,
      "grad_norm": 0.014390280470252037,
      "learning_rate": 9.878492469826896e-05,
      "loss": 0.0003,
      "step": 109
    },
    {
      "epoch": 0.04258614014711576,
      "grad_norm": 0.1289769858121872,
      "learning_rate": 9.876154293048163e-05,
      "loss": 0.0026,
      "step": 110
    },
    {
      "epoch": 0.0429732868757259,
      "grad_norm": 0.25492918491363525,
      "learning_rate": 9.873794115357966e-05,
      "loss": 0.0093,
      "step": 111
    },
    {
      "epoch": 0.04336043360433604,
      "grad_norm": 0.05008332058787346,
      "learning_rate": 9.87141194740528e-05,
      "loss": 0.0011,
      "step": 112
    },
    {
      "epoch": 0.043747580332946186,
      "grad_norm": 0.24651627242565155,
      "learning_rate": 9.869007799938305e-05,
      "loss": 0.0034,
      "step": 113
    },
    {
      "epoch": 0.04413472706155633,
      "grad_norm": 0.004503064788877964,
      "learning_rate": 9.86658168380441e-05,
      "loss": 0.0001,
      "step": 114
    },
    {
      "epoch": 0.04452187379016647,
      "grad_norm": 0.02013341151177883,
      "learning_rate": 9.864133609950078e-05,
      "loss": 0.0003,
      "step": 115
    },
    {
      "epoch": 0.04490902051877662,
      "grad_norm": 0.4879964590072632,
      "learning_rate": 9.861663589420871e-05,
      "loss": 0.019,
      "step": 116
    },
    {
      "epoch": 0.04529616724738676,
      "grad_norm": 0.4585297405719757,
      "learning_rate": 9.859171633361372e-05,
      "loss": 0.022,
      "step": 117
    },
    {
      "epoch": 0.045683313975996905,
      "grad_norm": 0.45999088883399963,
      "learning_rate": 9.856657753015135e-05,
      "loss": 0.0196,
      "step": 118
    },
    {
      "epoch": 0.04607046070460705,
      "grad_norm": 0.15138907730579376,
      "learning_rate": 9.854121959724636e-05,
      "loss": 0.0004,
      "step": 119
    },
    {
      "epoch": 0.04645760743321719,
      "grad_norm": 0.05731305480003357,
      "learning_rate": 9.851564264931219e-05,
      "loss": 0.0004,
      "step": 120
    },
    {
      "epoch": 0.04684475416182733,
      "grad_norm": 0.6728512644767761,
      "learning_rate": 9.848984680175049e-05,
      "loss": 0.0094,
      "step": 121
    },
    {
      "epoch": 0.047231900890437474,
      "grad_norm": 0.009215263649821281,
      "learning_rate": 9.846383217095053e-05,
      "loss": 0.0002,
      "step": 122
    },
    {
      "epoch": 0.047619047619047616,
      "grad_norm": 0.006808656267821789,
      "learning_rate": 9.843759887428873e-05,
      "loss": 0.0001,
      "step": 123
    },
    {
      "epoch": 0.048006194347657766,
      "grad_norm": 0.005876703653484583,
      "learning_rate": 9.841114703012817e-05,
      "loss": 0.0001,
      "step": 124
    },
    {
      "epoch": 0.04839334107626791,
      "grad_norm": 0.019576961174607277,
      "learning_rate": 9.838447675781794e-05,
      "loss": 0.0003,
      "step": 125
    },
    {
      "epoch": 0.04878048780487805,
      "grad_norm": 0.008768095634877682,
      "learning_rate": 9.835758817769269e-05,
      "loss": 0.0002,
      "step": 126
    },
    {
      "epoch": 0.04916763453348819,
      "grad_norm": 0.023376451805233955,
      "learning_rate": 9.833048141107203e-05,
      "loss": 0.0004,
      "step": 127
    },
    {
      "epoch": 0.049554781262098335,
      "grad_norm": 0.065276600420475,
      "learning_rate": 9.830315658026011e-05,
      "loss": 0.0008,
      "step": 128
    },
    {
      "epoch": 0.04994192799070848,
      "grad_norm": 0.240321084856987,
      "learning_rate": 9.827561380854482e-05,
      "loss": 0.0014,
      "step": 129
    },
    {
      "epoch": 0.05032907471931862,
      "grad_norm": 0.01600070670247078,
      "learning_rate": 9.824785322019753e-05,
      "loss": 0.0002,
      "step": 130
    },
    {
      "epoch": 0.05071622144792876,
      "grad_norm": 0.009562276303768158,
      "learning_rate": 9.821987494047229e-05,
      "loss": 0.0003,
      "step": 131
    },
    {
      "epoch": 0.05110336817653891,
      "grad_norm": 0.009312965907156467,
      "learning_rate": 9.819167909560543e-05,
      "loss": 0.0002,
      "step": 132
    },
    {
      "epoch": 0.051490514905149054,
      "grad_norm": 0.016949571669101715,
      "learning_rate": 9.816326581281485e-05,
      "loss": 0.0003,
      "step": 133
    },
    {
      "epoch": 0.051877661633759196,
      "grad_norm": 0.011933255940675735,
      "learning_rate": 9.813463522029957e-05,
      "loss": 0.0003,
      "step": 134
    },
    {
      "epoch": 0.05226480836236934,
      "grad_norm": 0.03951014205813408,
      "learning_rate": 9.810578744723911e-05,
      "loss": 0.0007,
      "step": 135
    },
    {
      "epoch": 0.05265195509097948,
      "grad_norm": 0.008960571140050888,
      "learning_rate": 9.807672262379282e-05,
      "loss": 0.0001,
      "step": 136
    },
    {
      "epoch": 0.053039101819589624,
      "grad_norm": 0.1299860030412674,
      "learning_rate": 9.804744088109943e-05,
      "loss": 0.0038,
      "step": 137
    },
    {
      "epoch": 0.053426248548199766,
      "grad_norm": 0.036717738956213,
      "learning_rate": 9.801794235127636e-05,
      "loss": 0.0004,
      "step": 138
    },
    {
      "epoch": 0.05381339527680991,
      "grad_norm": 0.008997570723295212,
      "learning_rate": 9.798822716741925e-05,
      "loss": 0.0002,
      "step": 139
    },
    {
      "epoch": 0.05420054200542006,
      "grad_norm": 0.03509435057640076,
      "learning_rate": 9.795829546360114e-05,
      "loss": 0.0005,
      "step": 140
    },
    {
      "epoch": 0.0545876887340302,
      "grad_norm": 0.034250613301992416,
      "learning_rate": 9.792814737487207e-05,
      "loss": 0.0006,
      "step": 141
    },
    {
      "epoch": 0.05497483546264034,
      "grad_norm": 0.0243786983191967,
      "learning_rate": 9.789778303725838e-05,
      "loss": 0.0005,
      "step": 142
    },
    {
      "epoch": 0.055361982191250485,
      "grad_norm": 0.0226341113448143,
      "learning_rate": 9.786720258776213e-05,
      "loss": 0.0004,
      "step": 143
    },
    {
      "epoch": 0.05574912891986063,
      "grad_norm": 0.23691046237945557,
      "learning_rate": 9.783640616436044e-05,
      "loss": 0.0013,
      "step": 144
    },
    {
      "epoch": 0.05613627564847077,
      "grad_norm": 0.010031294077634811,
      "learning_rate": 9.780539390600489e-05,
      "loss": 0.0002,
      "step": 145
    },
    {
      "epoch": 0.05652342237708091,
      "grad_norm": 0.07630310952663422,
      "learning_rate": 9.777416595262091e-05,
      "loss": 0.0013,
      "step": 146
    },
    {
      "epoch": 0.056910569105691054,
      "grad_norm": 0.25514844059944153,
      "learning_rate": 9.774272244510712e-05,
      "loss": 0.0045,
      "step": 147
    },
    {
      "epoch": 0.057297715834301204,
      "grad_norm": 0.03640531376004219,
      "learning_rate": 9.771106352533472e-05,
      "loss": 0.0007,
      "step": 148
    },
    {
      "epoch": 0.057684862562911346,
      "grad_norm": 0.07202103734016418,
      "learning_rate": 9.767918933614682e-05,
      "loss": 0.0016,
      "step": 149
    },
    {
      "epoch": 0.05807200929152149,
      "grad_norm": 0.12926505506038666,
      "learning_rate": 9.764710002135784e-05,
      "loss": 0.0023,
      "step": 150
    },
    {
      "epoch": 0.05845915602013163,
      "grad_norm": 0.05395910516381264,
      "learning_rate": 9.761479572575279e-05,
      "loss": 0.0121,
      "step": 151
    },
    {
      "epoch": 0.05884630274874177,
      "grad_norm": 0.06644894927740097,
      "learning_rate": 9.75822765950867e-05,
      "loss": 0.0046,
      "step": 152
    },
    {
      "epoch": 0.059233449477351915,
      "grad_norm": 0.02791532874107361,
      "learning_rate": 9.754954277608391e-05,
      "loss": 0.0009,
      "step": 153
    },
    {
      "epoch": 0.05962059620596206,
      "grad_norm": 0.03686601668596268,
      "learning_rate": 9.751659441643742e-05,
      "loss": 0.0016,
      "step": 154
    },
    {
      "epoch": 0.0600077429345722,
      "grad_norm": 0.04866962507367134,
      "learning_rate": 9.748343166480823e-05,
      "loss": 0.0016,
      "step": 155
    },
    {
      "epoch": 0.06039488966318235,
      "grad_norm": 0.06715308874845505,
      "learning_rate": 9.745005467082464e-05,
      "loss": 0.0011,
      "step": 156
    },
    {
      "epoch": 0.06078203639179249,
      "grad_norm": 0.10314395278692245,
      "learning_rate": 9.741646358508164e-05,
      "loss": 0.0038,
      "step": 157
    },
    {
      "epoch": 0.061169183120402634,
      "grad_norm": 0.02877042070031166,
      "learning_rate": 9.738265855914013e-05,
      "loss": 0.0005,
      "step": 158
    },
    {
      "epoch": 0.06155632984901278,
      "grad_norm": 0.07983982563018799,
      "learning_rate": 9.734863974552635e-05,
      "loss": 0.0019,
      "step": 159
    },
    {
      "epoch": 0.06194347657762292,
      "grad_norm": 0.006283288821578026,
      "learning_rate": 9.731440729773114e-05,
      "loss": 0.0001,
      "step": 160
    },
    {
      "epoch": 0.06233062330623306,
      "grad_norm": 0.004482879303395748,
      "learning_rate": 9.727996137020918e-05,
      "loss": 0.0001,
      "step": 161
    },
    {
      "epoch": 0.0627177700348432,
      "grad_norm": 0.0071263727732002735,
      "learning_rate": 9.72453021183784e-05,
      "loss": 0.0001,
      "step": 162
    },
    {
      "epoch": 0.06310491676345335,
      "grad_norm": 0.0014257959555834532,
      "learning_rate": 9.721042969861928e-05,
      "loss": 0.0,
      "step": 163
    },
    {
      "epoch": 0.06349206349206349,
      "grad_norm": 0.06279278546571732,
      "learning_rate": 9.717534426827404e-05,
      "loss": 0.0013,
      "step": 164
    },
    {
      "epoch": 0.06387921022067364,
      "grad_norm": 0.10048985481262207,
      "learning_rate": 9.714004598564598e-05,
      "loss": 0.0015,
      "step": 165
    },
    {
      "epoch": 0.06426635694928377,
      "grad_norm": 0.010703545063734055,
      "learning_rate": 9.710453500999882e-05,
      "loss": 0.0002,
      "step": 166
    },
    {
      "epoch": 0.06465350367789392,
      "grad_norm": 0.0008502176497131586,
      "learning_rate": 9.706881150155591e-05,
      "loss": 0.0,
      "step": 167
    },
    {
      "epoch": 0.06504065040650407,
      "grad_norm": 0.035037811845541,
      "learning_rate": 9.703287562149959e-05,
      "loss": 0.0003,
      "step": 168
    },
    {
      "epoch": 0.06542779713511421,
      "grad_norm": 0.0808122381567955,
      "learning_rate": 9.69967275319703e-05,
      "loss": 0.0007,
      "step": 169
    },
    {
      "epoch": 0.06581494386372436,
      "grad_norm": 0.0047109900042414665,
      "learning_rate": 9.696036739606606e-05,
      "loss": 0.0001,
      "step": 170
    },
    {
      "epoch": 0.06620209059233449,
      "grad_norm": 0.30203601717948914,
      "learning_rate": 9.692379537784157e-05,
      "loss": 0.0048,
      "step": 171
    },
    {
      "epoch": 0.06658923732094464,
      "grad_norm": 0.0027131803799420595,
      "learning_rate": 9.688701164230758e-05,
      "loss": 0.0001,
      "step": 172
    },
    {
      "epoch": 0.06697638404955478,
      "grad_norm": 0.002335549332201481,
      "learning_rate": 9.685001635543005e-05,
      "loss": 0.0001,
      "step": 173
    },
    {
      "epoch": 0.06736353077816493,
      "grad_norm": 0.0023098927922546864,
      "learning_rate": 9.681280968412947e-05,
      "loss": 0.0001,
      "step": 174
    },
    {
      "epoch": 0.06775067750677506,
      "grad_norm": 0.001849909545853734,
      "learning_rate": 9.677539179628005e-05,
      "loss": 0.0,
      "step": 175
    },
    {
      "epoch": 0.06813782423538521,
      "grad_norm": 0.32040995359420776,
      "learning_rate": 9.673776286070906e-05,
      "loss": 0.0028,
      "step": 176
    },
    {
      "epoch": 0.06852497096399536,
      "grad_norm": 0.0077479127794504166,
      "learning_rate": 9.669992304719595e-05,
      "loss": 0.0001,
      "step": 177
    },
    {
      "epoch": 0.0689121176926055,
      "grad_norm": 0.020282207056879997,
      "learning_rate": 9.666187252647165e-05,
      "loss": 0.0002,
      "step": 178
    },
    {
      "epoch": 0.06929926442121565,
      "grad_norm": 0.00533844530582428,
      "learning_rate": 9.662361147021779e-05,
      "loss": 0.0001,
      "step": 179
    },
    {
      "epoch": 0.06968641114982578,
      "grad_norm": 0.0043952250853180885,
      "learning_rate": 9.658514005106596e-05,
      "loss": 0.0001,
      "step": 180
    },
    {
      "epoch": 0.07007355787843593,
      "grad_norm": 0.09598931670188904,
      "learning_rate": 9.654645844259683e-05,
      "loss": 0.0023,
      "step": 181
    },
    {
      "epoch": 0.07046070460704607,
      "grad_norm": 0.006287871859967709,
      "learning_rate": 9.650756681933947e-05,
      "loss": 0.0001,
      "step": 182
    },
    {
      "epoch": 0.07084785133565621,
      "grad_norm": 0.03524954989552498,
      "learning_rate": 9.646846535677055e-05,
      "loss": 0.0002,
      "step": 183
    },
    {
      "epoch": 0.07123499806426636,
      "grad_norm": 0.005163317546248436,
      "learning_rate": 9.642915423131342e-05,
      "loss": 0.0001,
      "step": 184
    },
    {
      "epoch": 0.0716221447928765,
      "grad_norm": 0.03801601007580757,
      "learning_rate": 9.638963362033756e-05,
      "loss": 0.0009,
      "step": 185
    },
    {
      "epoch": 0.07200929152148665,
      "grad_norm": 0.009458637796342373,
      "learning_rate": 9.634990370215752e-05,
      "loss": 0.0001,
      "step": 186
    },
    {
      "epoch": 0.07239643825009678,
      "grad_norm": 0.07404721528291702,
      "learning_rate": 9.630996465603228e-05,
      "loss": 0.0011,
      "step": 187
    },
    {
      "epoch": 0.07278358497870693,
      "grad_norm": 0.1175776869058609,
      "learning_rate": 9.626981666216439e-05,
      "loss": 0.001,
      "step": 188
    },
    {
      "epoch": 0.07317073170731707,
      "grad_norm": 0.07584076374769211,
      "learning_rate": 9.622945990169916e-05,
      "loss": 0.0007,
      "step": 189
    },
    {
      "epoch": 0.07355787843592722,
      "grad_norm": 0.01918666809797287,
      "learning_rate": 9.618889455672384e-05,
      "loss": 0.0001,
      "step": 190
    },
    {
      "epoch": 0.07394502516453735,
      "grad_norm": 1.1152503490447998,
      "learning_rate": 9.614812081026678e-05,
      "loss": 0.0111,
      "step": 191
    },
    {
      "epoch": 0.0743321718931475,
      "grad_norm": 0.0046646371483802795,
      "learning_rate": 9.610713884629666e-05,
      "loss": 0.0,
      "step": 192
    },
    {
      "epoch": 0.07471931862175765,
      "grad_norm": 0.0027072590310126543,
      "learning_rate": 9.60659488497216e-05,
      "loss": 0.0001,
      "step": 193
    },
    {
      "epoch": 0.07510646535036779,
      "grad_norm": 0.4910131096839905,
      "learning_rate": 9.602455100638836e-05,
      "loss": 0.0041,
      "step": 194
    },
    {
      "epoch": 0.07549361207897794,
      "grad_norm": 0.0024883951991796494,
      "learning_rate": 9.598294550308149e-05,
      "loss": 0.0,
      "step": 195
    },
    {
      "epoch": 0.07588075880758807,
      "grad_norm": 0.012495968490839005,
      "learning_rate": 9.594113252752248e-05,
      "loss": 0.0001,
      "step": 196
    },
    {
      "epoch": 0.07626790553619822,
      "grad_norm": 0.012010253965854645,
      "learning_rate": 9.589911226836896e-05,
      "loss": 0.0001,
      "step": 197
    },
    {
      "epoch": 0.07665505226480836,
      "grad_norm": 0.019101502373814583,
      "learning_rate": 9.585688491521374e-05,
      "loss": 0.0002,
      "step": 198
    },
    {
      "epoch": 0.0770421989934185,
      "grad_norm": 0.018270643427968025,
      "learning_rate": 9.58144506585841e-05,
      "loss": 0.0004,
      "step": 199
    },
    {
      "epoch": 0.07742934572202866,
      "grad_norm": 1.0924277305603027,
      "learning_rate": 9.577180968994082e-05,
      "loss": 0.0119,
      "step": 200
    },
    {
      "epoch": 0.07781649245063879,
      "grad_norm": 0.05194557085633278,
      "learning_rate": 9.572896220167735e-05,
      "loss": 0.0083,
      "step": 201
    },
    {
      "epoch": 0.07820363917924894,
      "grad_norm": 0.0748182013630867,
      "learning_rate": 9.568590838711895e-05,
      "loss": 0.0026,
      "step": 202
    },
    {
      "epoch": 0.07859078590785908,
      "grad_norm": 0.04829065874218941,
      "learning_rate": 9.564264844052182e-05,
      "loss": 0.0009,
      "step": 203
    },
    {
      "epoch": 0.07897793263646923,
      "grad_norm": 0.07577355951070786,
      "learning_rate": 9.559918255707219e-05,
      "loss": 0.0031,
      "step": 204
    },
    {
      "epoch": 0.07936507936507936,
      "grad_norm": 0.09440341591835022,
      "learning_rate": 9.555551093288548e-05,
      "loss": 0.0033,
      "step": 205
    },
    {
      "epoch": 0.07975222609368951,
      "grad_norm": 0.028996804729104042,
      "learning_rate": 9.551163376500543e-05,
      "loss": 0.0002,
      "step": 206
    },
    {
      "epoch": 0.08013937282229965,
      "grad_norm": 0.12111848592758179,
      "learning_rate": 9.546755125140313e-05,
      "loss": 0.0018,
      "step": 207
    },
    {
      "epoch": 0.0805265195509098,
      "grad_norm": 0.01013265922665596,
      "learning_rate": 9.542326359097619e-05,
      "loss": 0.0001,
      "step": 208
    },
    {
      "epoch": 0.08091366627951994,
      "grad_norm": 0.04623102769255638,
      "learning_rate": 9.537877098354786e-05,
      "loss": 0.0007,
      "step": 209
    },
    {
      "epoch": 0.08130081300813008,
      "grad_norm": 0.0021863309666514397,
      "learning_rate": 9.533407362986606e-05,
      "loss": 0.0,
      "step": 210
    },
    {
      "epoch": 0.08168795973674023,
      "grad_norm": 0.0072447373531758785,
      "learning_rate": 9.528917173160255e-05,
      "loss": 0.0001,
      "step": 211
    },
    {
      "epoch": 0.08207510646535036,
      "grad_norm": 0.011722790077328682,
      "learning_rate": 9.524406549135194e-05,
      "loss": 0.0002,
      "step": 212
    },
    {
      "epoch": 0.08246225319396051,
      "grad_norm": 0.08522871136665344,
      "learning_rate": 9.519875511263086e-05,
      "loss": 0.0006,
      "step": 213
    },
    {
      "epoch": 0.08284939992257065,
      "grad_norm": 0.03702247142791748,
      "learning_rate": 9.515324079987697e-05,
      "loss": 0.0007,
      "step": 214
    },
    {
      "epoch": 0.0832365466511808,
      "grad_norm": 0.004982550162822008,
      "learning_rate": 9.51075227584481e-05,
      "loss": 0.0001,
      "step": 215
    },
    {
      "epoch": 0.08362369337979095,
      "grad_norm": 0.0032351913396269083,
      "learning_rate": 9.506160119462124e-05,
      "loss": 0.0,
      "step": 216
    },
    {
      "epoch": 0.08401084010840108,
      "grad_norm": 0.008003685623407364,
      "learning_rate": 9.501547631559172e-05,
      "loss": 0.0001,
      "step": 217
    },
    {
      "epoch": 0.08439798683701123,
      "grad_norm": 0.21871095895767212,
      "learning_rate": 9.496914832947215e-05,
      "loss": 0.0045,
      "step": 218
    },
    {
      "epoch": 0.08478513356562137,
      "grad_norm": 0.05447891354560852,
      "learning_rate": 9.492261744529163e-05,
      "loss": 0.0008,
      "step": 219
    },
    {
      "epoch": 0.08517228029423152,
      "grad_norm": 0.019385039806365967,
      "learning_rate": 9.487588387299464e-05,
      "loss": 0.0003,
      "step": 220
    },
    {
      "epoch": 0.08555942702284165,
      "grad_norm": 0.009612143971025944,
      "learning_rate": 9.482894782344025e-05,
      "loss": 0.0002,
      "step": 221
    },
    {
      "epoch": 0.0859465737514518,
      "grad_norm": 0.1183067038655281,
      "learning_rate": 9.478180950840103e-05,
      "loss": 0.0007,
      "step": 222
    },
    {
      "epoch": 0.08633372048006194,
      "grad_norm": 0.2591310143470764,
      "learning_rate": 9.47344691405622e-05,
      "loss": 0.0008,
      "step": 223
    },
    {
      "epoch": 0.08672086720867209,
      "grad_norm": 0.0012734796619042754,
      "learning_rate": 9.468692693352063e-05,
      "loss": 0.0,
      "step": 224
    },
    {
      "epoch": 0.08710801393728224,
      "grad_norm": 1.0743658542633057,
      "learning_rate": 9.463918310178384e-05,
      "loss": 0.0083,
      "step": 225
    },
    {
      "epoch": 0.08749516066589237,
      "grad_norm": 0.0012743185507133603,
      "learning_rate": 9.459123786076912e-05,
      "loss": 0.0,
      "step": 226
    },
    {
      "epoch": 0.08788230739450252,
      "grad_norm": 0.007091084029525518,
      "learning_rate": 9.454309142680247e-05,
      "loss": 0.0,
      "step": 227
    },
    {
      "epoch": 0.08826945412311266,
      "grad_norm": 0.16258011758327484,
      "learning_rate": 9.449474401711766e-05,
      "loss": 0.0007,
      "step": 228
    },
    {
      "epoch": 0.0886566008517228,
      "grad_norm": 0.02502388134598732,
      "learning_rate": 9.444619584985526e-05,
      "loss": 0.0002,
      "step": 229
    },
    {
      "epoch": 0.08904374758033294,
      "grad_norm": 0.020117763429880142,
      "learning_rate": 9.439744714406167e-05,
      "loss": 0.0001,
      "step": 230
    },
    {
      "epoch": 0.08943089430894309,
      "grad_norm": 0.025272773578763008,
      "learning_rate": 9.434849811968807e-05,
      "loss": 0.0001,
      "step": 231
    },
    {
      "epoch": 0.08981804103755324,
      "grad_norm": 0.0023280701134353876,
      "learning_rate": 9.429934899758948e-05,
      "loss": 0.0,
      "step": 232
    },
    {
      "epoch": 0.09020518776616337,
      "grad_norm": 0.04557230323553085,
      "learning_rate": 9.424999999952375e-05,
      "loss": 0.0003,
      "step": 233
    },
    {
      "epoch": 0.09059233449477352,
      "grad_norm": 0.3395751714706421,
      "learning_rate": 9.420045134815056e-05,
      "loss": 0.0009,
      "step": 234
    },
    {
      "epoch": 0.09097948122338366,
      "grad_norm": 0.3556441068649292,
      "learning_rate": 9.41507032670304e-05,
      "loss": 0.0203,
      "step": 235
    },
    {
      "epoch": 0.09136662795199381,
      "grad_norm": 0.021049408242106438,
      "learning_rate": 9.410075598062358e-05,
      "loss": 0.0001,
      "step": 236
    },
    {
      "epoch": 0.09175377468060394,
      "grad_norm": 0.0035629859194159508,
      "learning_rate": 9.405060971428923e-05,
      "loss": 0.0,
      "step": 237
    },
    {
      "epoch": 0.0921409214092141,
      "grad_norm": 1.088625431060791,
      "learning_rate": 9.400026469428424e-05,
      "loss": 0.003,
      "step": 238
    },
    {
      "epoch": 0.09252806813782423,
      "grad_norm": 0.34952932596206665,
      "learning_rate": 9.39497211477623e-05,
      "loss": 0.0006,
      "step": 239
    },
    {
      "epoch": 0.09291521486643438,
      "grad_norm": 0.01403273455798626,
      "learning_rate": 9.38989793027728e-05,
      "loss": 0.0002,
      "step": 240
    },
    {
      "epoch": 0.09330236159504453,
      "grad_norm": 0.0016716537065804005,
      "learning_rate": 9.384803938825988e-05,
      "loss": 0.0,
      "step": 241
    },
    {
      "epoch": 0.09368950832365466,
      "grad_norm": 0.006439753342419863,
      "learning_rate": 9.379690163406128e-05,
      "loss": 0.0,
      "step": 242
    },
    {
      "epoch": 0.09407665505226481,
      "grad_norm": 0.0021656707394868135,
      "learning_rate": 9.374556627090749e-05,
      "loss": 0.0,
      "step": 243
    },
    {
      "epoch": 0.09446380178087495,
      "grad_norm": 0.0023138655815273523,
      "learning_rate": 9.369403353042052e-05,
      "loss": 0.0,
      "step": 244
    },
    {
      "epoch": 0.0948509485094851,
      "grad_norm": 0.002230309648439288,
      "learning_rate": 9.364230364511296e-05,
      "loss": 0.0001,
      "step": 245
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 0.014607599936425686,
      "learning_rate": 9.35903768483869e-05,
      "loss": 0.0001,
      "step": 246
    },
    {
      "epoch": 0.09562524196670538,
      "grad_norm": 0.005560693796724081,
      "learning_rate": 9.353825337453291e-05,
      "loss": 0.0001,
      "step": 247
    },
    {
      "epoch": 0.09601238869531553,
      "grad_norm": 0.013218767009675503,
      "learning_rate": 9.348593345872891e-05,
      "loss": 0.0002,
      "step": 248
    },
    {
      "epoch": 0.09639953542392567,
      "grad_norm": 0.05328545346856117,
      "learning_rate": 9.343341733703919e-05,
      "loss": 0.0004,
      "step": 249
    },
    {
      "epoch": 0.09678668215253582,
      "grad_norm": 0.0031676662620157003,
      "learning_rate": 9.338070524641329e-05,
      "loss": 0.0001,
      "step": 250
    },
    {
      "epoch": 0.09717382888114595,
      "grad_norm": 0.050222184509038925,
      "learning_rate": 9.332779742468496e-05,
      "loss": 0.0157,
      "step": 251
    },
    {
      "epoch": 0.0975609756097561,
      "grad_norm": 0.028248561546206474,
      "learning_rate": 9.327469411057106e-05,
      "loss": 0.0007,
      "step": 252
    },
    {
      "epoch": 0.09794812233836624,
      "grad_norm": 0.03661990538239479,
      "learning_rate": 9.322139554367052e-05,
      "loss": 0.0005,
      "step": 253
    },
    {
      "epoch": 0.09833526906697639,
      "grad_norm": 0.04206279292702675,
      "learning_rate": 9.316790196446324e-05,
      "loss": 0.0013,
      "step": 254
    },
    {
      "epoch": 0.09872241579558652,
      "grad_norm": 0.11051138490438461,
      "learning_rate": 9.311421361430898e-05,
      "loss": 0.0047,
      "step": 255
    },
    {
      "epoch": 0.09910956252419667,
      "grad_norm": 0.08012690395116806,
      "learning_rate": 9.306033073544631e-05,
      "loss": 0.0006,
      "step": 256
    },
    {
      "epoch": 0.09949670925280682,
      "grad_norm": 0.05237819254398346,
      "learning_rate": 9.300625357099151e-05,
      "loss": 0.0008,
      "step": 257
    },
    {
      "epoch": 0.09988385598141696,
      "grad_norm": 0.014332195743918419,
      "learning_rate": 9.295198236493745e-05,
      "loss": 0.0002,
      "step": 258
    },
    {
      "epoch": 0.1002710027100271,
      "grad_norm": 0.05660046637058258,
      "learning_rate": 9.289751736215251e-05,
      "loss": 0.0004,
      "step": 259
    },
    {
      "epoch": 0.10065814943863724,
      "grad_norm": 0.09471816569566727,
      "learning_rate": 9.284285880837946e-05,
      "loss": 0.0009,
      "step": 260
    },
    {
      "epoch": 0.10104529616724739,
      "grad_norm": 0.0016173557378351688,
      "learning_rate": 9.27880069502344e-05,
      "loss": 0.0,
      "step": 261
    },
    {
      "epoch": 0.10143244289585752,
      "grad_norm": 0.1207282617688179,
      "learning_rate": 9.273296203520553e-05,
      "loss": 0.0009,
      "step": 262
    },
    {
      "epoch": 0.10181958962446767,
      "grad_norm": 0.0016727615147829056,
      "learning_rate": 9.267772431165218e-05,
      "loss": 0.0,
      "step": 263
    },
    {
      "epoch": 0.10220673635307782,
      "grad_norm": 0.010696031153202057,
      "learning_rate": 9.262229402880362e-05,
      "loss": 0.0001,
      "step": 264
    },
    {
      "epoch": 0.10259388308168796,
      "grad_norm": 0.003082268638536334,
      "learning_rate": 9.256667143675789e-05,
      "loss": 0.0001,
      "step": 265
    },
    {
      "epoch": 0.10298102981029811,
      "grad_norm": 0.019705908372998238,
      "learning_rate": 9.251085678648072e-05,
      "loss": 0.0001,
      "step": 266
    },
    {
      "epoch": 0.10336817653890824,
      "grad_norm": 0.0005683451890945435,
      "learning_rate": 9.245485032980445e-05,
      "loss": 0.0,
      "step": 267
    },
    {
      "epoch": 0.10375532326751839,
      "grad_norm": 0.0019504778319969773,
      "learning_rate": 9.239865231942678e-05,
      "loss": 0.0,
      "step": 268
    },
    {
      "epoch": 0.10414246999612853,
      "grad_norm": 0.002394254319369793,
      "learning_rate": 9.234226300890974e-05,
      "loss": 0.0,
      "step": 269
    },
    {
      "epoch": 0.10452961672473868,
      "grad_norm": 0.00082309142453596,
      "learning_rate": 9.228568265267845e-05,
      "loss": 0.0,
      "step": 270
    },
    {
      "epoch": 0.10491676345334881,
      "grad_norm": 0.002265011891722679,
      "learning_rate": 9.222891150602005e-05,
      "loss": 0.0001,
      "step": 271
    },
    {
      "epoch": 0.10530391018195896,
      "grad_norm": 0.0030987069476395845,
      "learning_rate": 9.217194982508247e-05,
      "loss": 0.0,
      "step": 272
    },
    {
      "epoch": 0.10569105691056911,
      "grad_norm": 0.001322443364188075,
      "learning_rate": 9.211479786687338e-05,
      "loss": 0.0,
      "step": 273
    },
    {
      "epoch": 0.10607820363917925,
      "grad_norm": 0.010089612565934658,
      "learning_rate": 9.205745588925891e-05,
      "loss": 0.0001,
      "step": 274
    },
    {
      "epoch": 0.1064653503677894,
      "grad_norm": 0.3834461569786072,
      "learning_rate": 9.199992415096261e-05,
      "loss": 0.0114,
      "step": 275
    },
    {
      "epoch": 0.10685249709639953,
      "grad_norm": 0.0012483131140470505,
      "learning_rate": 9.194220291156413e-05,
      "loss": 0.0,
      "step": 276
    },
    {
      "epoch": 0.10723964382500968,
      "grad_norm": 0.0014100170228630304,
      "learning_rate": 9.188429243149824e-05,
      "loss": 0.0,
      "step": 277
    },
    {
      "epoch": 0.10762679055361982,
      "grad_norm": 0.0015941811725497246,
      "learning_rate": 9.182619297205348e-05,
      "loss": 0.0,
      "step": 278
    },
    {
      "epoch": 0.10801393728222997,
      "grad_norm": 0.0005449766758829355,
      "learning_rate": 9.176790479537108e-05,
      "loss": 0.0,
      "step": 279
    },
    {
      "epoch": 0.10840108401084012,
      "grad_norm": 0.1997094601392746,
      "learning_rate": 9.170942816444375e-05,
      "loss": 0.0027,
      "step": 280
    },
    {
      "epoch": 0.10878823073945025,
      "grad_norm": 0.001990030286833644,
      "learning_rate": 9.165076334311446e-05,
      "loss": 0.0,
      "step": 281
    },
    {
      "epoch": 0.1091753774680604,
      "grad_norm": 0.009841017425060272,
      "learning_rate": 9.159191059607537e-05,
      "loss": 0.0001,
      "step": 282
    },
    {
      "epoch": 0.10956252419667054,
      "grad_norm": 0.0011333709117025137,
      "learning_rate": 9.153287018886644e-05,
      "loss": 0.0,
      "step": 283
    },
    {
      "epoch": 0.10994967092528068,
      "grad_norm": 0.0004827795783057809,
      "learning_rate": 9.147364238787443e-05,
      "loss": 0.0,
      "step": 284
    },
    {
      "epoch": 0.11033681765389082,
      "grad_norm": 0.0046328590251505375,
      "learning_rate": 9.141422746033156e-05,
      "loss": 0.0,
      "step": 285
    },
    {
      "epoch": 0.11072396438250097,
      "grad_norm": 0.00042733096051961184,
      "learning_rate": 9.135462567431438e-05,
      "loss": 0.0,
      "step": 286
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 0.023281529545783997,
      "learning_rate": 9.12948372987425e-05,
      "loss": 0.0001,
      "step": 287
    },
    {
      "epoch": 0.11149825783972125,
      "grad_norm": 0.0010597595246508718,
      "learning_rate": 9.123486260337744e-05,
      "loss": 0.0,
      "step": 288
    },
    {
      "epoch": 0.1118854045683314,
      "grad_norm": 0.006935764104127884,
      "learning_rate": 9.117470185882139e-05,
      "loss": 0.0001,
      "step": 289
    },
    {
      "epoch": 0.11227255129694154,
      "grad_norm": 0.011282574385404587,
      "learning_rate": 9.111435533651596e-05,
      "loss": 0.0001,
      "step": 290
    },
    {
      "epoch": 0.11265969802555169,
      "grad_norm": 0.006138740107417107,
      "learning_rate": 9.105382330874099e-05,
      "loss": 0.0001,
      "step": 291
    },
    {
      "epoch": 0.11304684475416182,
      "grad_norm": 0.0003823502629529685,
      "learning_rate": 9.099310604861329e-05,
      "loss": 0.0,
      "step": 292
    },
    {
      "epoch": 0.11343399148277197,
      "grad_norm": 0.009164854884147644,
      "learning_rate": 9.093220383008545e-05,
      "loss": 0.0001,
      "step": 293
    },
    {
      "epoch": 0.11382113821138211,
      "grad_norm": 0.005529541522264481,
      "learning_rate": 9.087111692794459e-05,
      "loss": 0.0001,
      "step": 294
    },
    {
      "epoch": 0.11420828493999226,
      "grad_norm": 0.04618334770202637,
      "learning_rate": 9.08098456178111e-05,
      "loss": 0.0009,
      "step": 295
    },
    {
      "epoch": 0.11459543166860241,
      "grad_norm": 0.005164325702935457,
      "learning_rate": 9.074839017613736e-05,
      "loss": 0.0001,
      "step": 296
    },
    {
      "epoch": 0.11498257839721254,
      "grad_norm": 0.005010255612432957,
      "learning_rate": 9.068675088020661e-05,
      "loss": 0.0001,
      "step": 297
    },
    {
      "epoch": 0.11536972512582269,
      "grad_norm": 0.0010323580354452133,
      "learning_rate": 9.062492800813162e-05,
      "loss": 0.0,
      "step": 298
    },
    {
      "epoch": 0.11575687185443283,
      "grad_norm": 0.0012561334297060966,
      "learning_rate": 9.056292183885343e-05,
      "loss": 0.0,
      "step": 299
    },
    {
      "epoch": 0.11614401858304298,
      "grad_norm": 0.02239901013672352,
      "learning_rate": 9.050073265214005e-05,
      "loss": 0.0003,
      "step": 300
    },
    {
      "epoch": 0.11653116531165311,
      "grad_norm": 0.03204142302274704,
      "learning_rate": 9.043836072858535e-05,
      "loss": 0.0118,
      "step": 301
    },
    {
      "epoch": 0.11691831204026326,
      "grad_norm": 0.01608111336827278,
      "learning_rate": 9.037580634960764e-05,
      "loss": 0.0005,
      "step": 302
    },
    {
      "epoch": 0.1173054587688734,
      "grad_norm": 0.006757182534784079,
      "learning_rate": 9.031306979744847e-05,
      "loss": 0.0001,
      "step": 303
    },
    {
      "epoch": 0.11769260549748355,
      "grad_norm": 0.12025570124387741,
      "learning_rate": 9.025015135517137e-05,
      "loss": 0.0013,
      "step": 304
    },
    {
      "epoch": 0.1180797522260937,
      "grad_norm": 0.03977954015135765,
      "learning_rate": 9.01870513066605e-05,
      "loss": 0.0009,
      "step": 305
    },
    {
      "epoch": 0.11846689895470383,
      "grad_norm": 0.0013618416851386428,
      "learning_rate": 9.012376993661944e-05,
      "loss": 0.0,
      "step": 306
    },
    {
      "epoch": 0.11885404568331398,
      "grad_norm": 0.018894262611865997,
      "learning_rate": 9.006030753056989e-05,
      "loss": 0.0006,
      "step": 307
    },
    {
      "epoch": 0.11924119241192412,
      "grad_norm": 0.12205176800489426,
      "learning_rate": 8.999666437485035e-05,
      "loss": 0.0008,
      "step": 308
    },
    {
      "epoch": 0.11962833914053426,
      "grad_norm": 0.0020962394773960114,
      "learning_rate": 8.993284075661487e-05,
      "loss": 0.0001,
      "step": 309
    },
    {
      "epoch": 0.1200154858691444,
      "grad_norm": 0.0007874541915953159,
      "learning_rate": 8.986883696383175e-05,
      "loss": 0.0,
      "step": 310
    },
    {
      "epoch": 0.12040263259775455,
      "grad_norm": 0.015855029225349426,
      "learning_rate": 8.980465328528219e-05,
      "loss": 0.0002,
      "step": 311
    },
    {
      "epoch": 0.1207897793263647,
      "grad_norm": 0.004300241824239492,
      "learning_rate": 8.974029001055906e-05,
      "loss": 0.0001,
      "step": 312
    },
    {
      "epoch": 0.12117692605497483,
      "grad_norm": 0.06699482351541519,
      "learning_rate": 8.967574743006551e-05,
      "loss": 0.0011,
      "step": 313
    },
    {
      "epoch": 0.12156407278358498,
      "grad_norm": 0.0414130873978138,
      "learning_rate": 8.961102583501377e-05,
      "loss": 0.0009,
      "step": 314
    },
    {
      "epoch": 0.12195121951219512,
      "grad_norm": 0.0015767719596624374,
      "learning_rate": 8.95461255174237e-05,
      "loss": 0.0,
      "step": 315
    },
    {
      "epoch": 0.12233836624080527,
      "grad_norm": 0.006629679352045059,
      "learning_rate": 8.948104677012161e-05,
      "loss": 0.0001,
      "step": 316
    },
    {
      "epoch": 0.1227255129694154,
      "grad_norm": 0.03726351261138916,
      "learning_rate": 8.941578988673887e-05,
      "loss": 0.0002,
      "step": 317
    },
    {
      "epoch": 0.12311265969802555,
      "grad_norm": 0.01188661903142929,
      "learning_rate": 8.935035516171051e-05,
      "loss": 0.0001,
      "step": 318
    },
    {
      "epoch": 0.12349980642663569,
      "grad_norm": 0.018168246373534203,
      "learning_rate": 8.928474289027407e-05,
      "loss": 0.0001,
      "step": 319
    },
    {
      "epoch": 0.12388695315524584,
      "grad_norm": 0.0020913630723953247,
      "learning_rate": 8.921895336846813e-05,
      "loss": 0.0,
      "step": 320
    },
    {
      "epoch": 0.12427409988385599,
      "grad_norm": 0.008128736168146133,
      "learning_rate": 8.915298689313097e-05,
      "loss": 0.0001,
      "step": 321
    },
    {
      "epoch": 0.12466124661246612,
      "grad_norm": 0.02947946824133396,
      "learning_rate": 8.908684376189935e-05,
      "loss": 0.0004,
      "step": 322
    },
    {
      "epoch": 0.12504839334107626,
      "grad_norm": 0.0005556530086323619,
      "learning_rate": 8.902052427320704e-05,
      "loss": 0.0,
      "step": 323
    },
    {
      "epoch": 0.1254355400696864,
      "grad_norm": 0.00042274571023881435,
      "learning_rate": 8.895402872628352e-05,
      "loss": 0.0,
      "step": 324
    },
    {
      "epoch": 0.12582268679829656,
      "grad_norm": 0.15616343915462494,
      "learning_rate": 8.888735742115268e-05,
      "loss": 0.0031,
      "step": 325
    },
    {
      "epoch": 0.1262098335269067,
      "grad_norm": 0.0006327281007543206,
      "learning_rate": 8.88205106586314e-05,
      "loss": 0.0,
      "step": 326
    },
    {
      "epoch": 0.12659698025551683,
      "grad_norm": 0.0013813385739922523,
      "learning_rate": 8.875348874032817e-05,
      "loss": 0.0,
      "step": 327
    },
    {
      "epoch": 0.12698412698412698,
      "grad_norm": 0.0075355893932282925,
      "learning_rate": 8.868629196864182e-05,
      "loss": 0.0,
      "step": 328
    },
    {
      "epoch": 0.12737127371273713,
      "grad_norm": 0.0004084401880390942,
      "learning_rate": 8.861892064676009e-05,
      "loss": 0.0,
      "step": 329
    },
    {
      "epoch": 0.12775842044134728,
      "grad_norm": 0.0006156496237963438,
      "learning_rate": 8.855137507865832e-05,
      "loss": 0.0,
      "step": 330
    },
    {
      "epoch": 0.12814556716995743,
      "grad_norm": 0.0027660855557769537,
      "learning_rate": 8.848365556909796e-05,
      "loss": 0.0,
      "step": 331
    },
    {
      "epoch": 0.12853271389856755,
      "grad_norm": 0.04597468674182892,
      "learning_rate": 8.841576242362533e-05,
      "loss": 0.0002,
      "step": 332
    },
    {
      "epoch": 0.1289198606271777,
      "grad_norm": 0.006770620588213205,
      "learning_rate": 8.834769594857019e-05,
      "loss": 0.0001,
      "step": 333
    },
    {
      "epoch": 0.12930700735578785,
      "grad_norm": 0.0005002229590900242,
      "learning_rate": 8.827945645104429e-05,
      "loss": 0.0,
      "step": 334
    },
    {
      "epoch": 0.129694154084398,
      "grad_norm": 0.0015748875448480248,
      "learning_rate": 8.821104423894016e-05,
      "loss": 0.0,
      "step": 335
    },
    {
      "epoch": 0.13008130081300814,
      "grad_norm": 0.020414279773831367,
      "learning_rate": 8.814245962092946e-05,
      "loss": 0.0003,
      "step": 336
    },
    {
      "epoch": 0.13046844754161827,
      "grad_norm": 0.024940207600593567,
      "learning_rate": 8.807370290646186e-05,
      "loss": 0.0001,
      "step": 337
    },
    {
      "epoch": 0.13085559427022841,
      "grad_norm": 0.0011114513035863638,
      "learning_rate": 8.800477440576347e-05,
      "loss": 0.0,
      "step": 338
    },
    {
      "epoch": 0.13124274099883856,
      "grad_norm": 0.006404312793165445,
      "learning_rate": 8.793567442983548e-05,
      "loss": 0.0001,
      "step": 339
    },
    {
      "epoch": 0.1316298877274487,
      "grad_norm": 0.0012291868915781379,
      "learning_rate": 8.786640329045278e-05,
      "loss": 0.0,
      "step": 340
    },
    {
      "epoch": 0.13201703445605883,
      "grad_norm": 0.027518250048160553,
      "learning_rate": 8.779696130016253e-05,
      "loss": 0.0002,
      "step": 341
    },
    {
      "epoch": 0.13240418118466898,
      "grad_norm": 0.000955001509282738,
      "learning_rate": 8.772734877228276e-05,
      "loss": 0.0,
      "step": 342
    },
    {
      "epoch": 0.13279132791327913,
      "grad_norm": 0.0012375194346532226,
      "learning_rate": 8.765756602090101e-05,
      "loss": 0.0,
      "step": 343
    },
    {
      "epoch": 0.13317847464188928,
      "grad_norm": 0.0009470246150158346,
      "learning_rate": 8.758761336087274e-05,
      "loss": 0.0,
      "step": 344
    },
    {
      "epoch": 0.13356562137049943,
      "grad_norm": 0.01968814805150032,
      "learning_rate": 8.751749110782012e-05,
      "loss": 0.0002,
      "step": 345
    },
    {
      "epoch": 0.13395276809910955,
      "grad_norm": 0.5433384776115417,
      "learning_rate": 8.74471995781305e-05,
      "loss": 0.0105,
      "step": 346
    },
    {
      "epoch": 0.1343399148277197,
      "grad_norm": 0.0011947732418775558,
      "learning_rate": 8.737673908895498e-05,
      "loss": 0.0,
      "step": 347
    },
    {
      "epoch": 0.13472706155632985,
      "grad_norm": 0.005630856845527887,
      "learning_rate": 8.7306109958207e-05,
      "loss": 0.0001,
      "step": 348
    },
    {
      "epoch": 0.13511420828494,
      "grad_norm": 0.00635663652792573,
      "learning_rate": 8.72353125045609e-05,
      "loss": 0.0001,
      "step": 349
    },
    {
      "epoch": 0.13550135501355012,
      "grad_norm": 0.8739410638809204,
      "learning_rate": 8.716434704745046e-05,
      "loss": 0.0276,
      "step": 350
    },
    {
      "epoch": 0.13588850174216027,
      "grad_norm": 0.05545734986662865,
      "learning_rate": 8.709321390706756e-05,
      "loss": 0.0078,
      "step": 351
    },
    {
      "epoch": 0.13627564847077042,
      "grad_norm": 0.027743853628635406,
      "learning_rate": 8.702191340436054e-05,
      "loss": 0.0007,
      "step": 352
    },
    {
      "epoch": 0.13666279519938057,
      "grad_norm": 0.10148515552282333,
      "learning_rate": 8.695044586103296e-05,
      "loss": 0.0048,
      "step": 353
    },
    {
      "epoch": 0.13704994192799072,
      "grad_norm": 0.031402137130498886,
      "learning_rate": 8.687881159954202e-05,
      "loss": 0.0008,
      "step": 354
    },
    {
      "epoch": 0.13743708865660084,
      "grad_norm": 0.055899184197187424,
      "learning_rate": 8.680701094309716e-05,
      "loss": 0.0008,
      "step": 355
    },
    {
      "epoch": 0.137824235385211,
      "grad_norm": 0.024489685893058777,
      "learning_rate": 8.673504421565856e-05,
      "loss": 0.0004,
      "step": 356
    },
    {
      "epoch": 0.13821138211382114,
      "grad_norm": 0.013303971849381924,
      "learning_rate": 8.666291174193573e-05,
      "loss": 0.0001,
      "step": 357
    },
    {
      "epoch": 0.1385985288424313,
      "grad_norm": 0.009390785358846188,
      "learning_rate": 8.659061384738598e-05,
      "loss": 0.0001,
      "step": 358
    },
    {
      "epoch": 0.1389856755710414,
      "grad_norm": 0.04934685677289963,
      "learning_rate": 8.651815085821303e-05,
      "loss": 0.0003,
      "step": 359
    },
    {
      "epoch": 0.13937282229965156,
      "grad_norm": 0.001061944873072207,
      "learning_rate": 8.644552310136546e-05,
      "loss": 0.0,
      "step": 360
    },
    {
      "epoch": 0.1397599690282617,
      "grad_norm": 0.005462823435664177,
      "learning_rate": 8.637273090453532e-05,
      "loss": 0.0001,
      "step": 361
    },
    {
      "epoch": 0.14014711575687186,
      "grad_norm": 0.0016542257508262992,
      "learning_rate": 8.629977459615655e-05,
      "loss": 0.0,
      "step": 362
    },
    {
      "epoch": 0.140534262485482,
      "grad_norm": 0.0004302252782508731,
      "learning_rate": 8.622665450540357e-05,
      "loss": 0.0,
      "step": 363
    },
    {
      "epoch": 0.14092140921409213,
      "grad_norm": 0.022549882531166077,
      "learning_rate": 8.615337096218979e-05,
      "loss": 0.0004,
      "step": 364
    },
    {
      "epoch": 0.14130855594270228,
      "grad_norm": 0.024535585194826126,
      "learning_rate": 8.60799242971661e-05,
      "loss": 0.0001,
      "step": 365
    },
    {
      "epoch": 0.14169570267131243,
      "grad_norm": 0.0006051576347090304,
      "learning_rate": 8.600631484171938e-05,
      "loss": 0.0,
      "step": 366
    },
    {
      "epoch": 0.14208284939992258,
      "grad_norm": 0.0036598185542970896,
      "learning_rate": 8.593254292797099e-05,
      "loss": 0.0,
      "step": 367
    },
    {
      "epoch": 0.14246999612853273,
      "grad_norm": 0.11391391605138779,
      "learning_rate": 8.585860888877537e-05,
      "loss": 0.0015,
      "step": 368
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.026237154379487038,
      "learning_rate": 8.578451305771836e-05,
      "loss": 0.0003,
      "step": 369
    },
    {
      "epoch": 0.143244289585753,
      "grad_norm": 0.005311033688485622,
      "learning_rate": 8.571025576911587e-05,
      "loss": 0.0001,
      "step": 370
    },
    {
      "epoch": 0.14363143631436315,
      "grad_norm": 0.0019114494789391756,
      "learning_rate": 8.563583735801224e-05,
      "loss": 0.0,
      "step": 371
    },
    {
      "epoch": 0.1440185830429733,
      "grad_norm": 0.0007387467776425183,
      "learning_rate": 8.556125816017882e-05,
      "loss": 0.0,
      "step": 372
    },
    {
      "epoch": 0.14440572977158342,
      "grad_norm": 0.17123480141162872,
      "learning_rate": 8.548651851211245e-05,
      "loss": 0.0016,
      "step": 373
    },
    {
      "epoch": 0.14479287650019357,
      "grad_norm": 0.0009499336592853069,
      "learning_rate": 8.541161875103381e-05,
      "loss": 0.0,
      "step": 374
    },
    {
      "epoch": 0.14518002322880372,
      "grad_norm": 0.001388967502862215,
      "learning_rate": 8.533655921488612e-05,
      "loss": 0.0,
      "step": 375
    },
    {
      "epoch": 0.14556716995741387,
      "grad_norm": 0.0017165900208055973,
      "learning_rate": 8.526134024233341e-05,
      "loss": 0.0,
      "step": 376
    },
    {
      "epoch": 0.14595431668602402,
      "grad_norm": 0.001004920806735754,
      "learning_rate": 8.518596217275912e-05,
      "loss": 0.0,
      "step": 377
    },
    {
      "epoch": 0.14634146341463414,
      "grad_norm": 0.0011910515604540706,
      "learning_rate": 8.511042534626449e-05,
      "loss": 0.0,
      "step": 378
    },
    {
      "epoch": 0.1467286101432443,
      "grad_norm": 0.020283464342355728,
      "learning_rate": 8.503473010366713e-05,
      "loss": 0.0001,
      "step": 379
    },
    {
      "epoch": 0.14711575687185444,
      "grad_norm": 0.0019051478011533618,
      "learning_rate": 8.495887678649933e-05,
      "loss": 0.0,
      "step": 380
    },
    {
      "epoch": 0.14750290360046459,
      "grad_norm": 0.0011362056247889996,
      "learning_rate": 8.488286573700665e-05,
      "loss": 0.0,
      "step": 381
    },
    {
      "epoch": 0.1478900503290747,
      "grad_norm": 0.02182379551231861,
      "learning_rate": 8.480669729814635e-05,
      "loss": 0.0002,
      "step": 382
    },
    {
      "epoch": 0.14827719705768486,
      "grad_norm": 0.006234586238861084,
      "learning_rate": 8.473037181358574e-05,
      "loss": 0.0001,
      "step": 383
    },
    {
      "epoch": 0.148664343786295,
      "grad_norm": 0.0006982333725318313,
      "learning_rate": 8.465388962770077e-05,
      "loss": 0.0,
      "step": 384
    },
    {
      "epoch": 0.14905149051490515,
      "grad_norm": 0.0026596917305141687,
      "learning_rate": 8.457725108557446e-05,
      "loss": 0.0,
      "step": 385
    },
    {
      "epoch": 0.1494386372435153,
      "grad_norm": 0.0008504387806169689,
      "learning_rate": 8.450045653299521e-05,
      "loss": 0.0,
      "step": 386
    },
    {
      "epoch": 0.14982578397212543,
      "grad_norm": 0.00914840493351221,
      "learning_rate": 8.442350631645536e-05,
      "loss": 0.0001,
      "step": 387
    },
    {
      "epoch": 0.15021293070073558,
      "grad_norm": 0.001677824417129159,
      "learning_rate": 8.434640078314962e-05,
      "loss": 0.0,
      "step": 388
    },
    {
      "epoch": 0.15060007742934572,
      "grad_norm": 0.002033672295510769,
      "learning_rate": 8.426914028097349e-05,
      "loss": 0.0,
      "step": 389
    },
    {
      "epoch": 0.15098722415795587,
      "grad_norm": 1.3247064352035522,
      "learning_rate": 8.419172515852159e-05,
      "loss": 0.0154,
      "step": 390
    },
    {
      "epoch": 0.151374370886566,
      "grad_norm": 0.0004959017969667912,
      "learning_rate": 8.41141557650863e-05,
      "loss": 0.0,
      "step": 391
    },
    {
      "epoch": 0.15176151761517614,
      "grad_norm": 0.0022033858112990856,
      "learning_rate": 8.403643245065598e-05,
      "loss": 0.0,
      "step": 392
    },
    {
      "epoch": 0.1521486643437863,
      "grad_norm": 0.00043081503827124834,
      "learning_rate": 8.39585555659135e-05,
      "loss": 0.0,
      "step": 393
    },
    {
      "epoch": 0.15253581107239644,
      "grad_norm": 0.0013331023510545492,
      "learning_rate": 8.388052546223461e-05,
      "loss": 0.0,
      "step": 394
    },
    {
      "epoch": 0.1529229578010066,
      "grad_norm": 0.0025198757648468018,
      "learning_rate": 8.380234249168641e-05,
      "loss": 0.0,
      "step": 395
    },
    {
      "epoch": 0.15331010452961671,
      "grad_norm": 0.008715974166989326,
      "learning_rate": 8.37240070070257e-05,
      "loss": 0.0001,
      "step": 396
    },
    {
      "epoch": 0.15369725125822686,
      "grad_norm": 0.0013087478000670671,
      "learning_rate": 8.364551936169742e-05,
      "loss": 0.0,
      "step": 397
    },
    {
      "epoch": 0.154084397986837,
      "grad_norm": 0.00940459594130516,
      "learning_rate": 8.356687990983306e-05,
      "loss": 0.0001,
      "step": 398
    },
    {
      "epoch": 0.15447154471544716,
      "grad_norm": 0.007965102791786194,
      "learning_rate": 8.348808900624901e-05,
      "loss": 0.0001,
      "step": 399
    },
    {
      "epoch": 0.1548586914440573,
      "grad_norm": 0.31034138798713684,
      "learning_rate": 8.340914700644506e-05,
      "loss": 0.0072,
      "step": 400
    },
    {
      "epoch": 0.15524583817266743,
      "grad_norm": 0.0318954698741436,
      "learning_rate": 8.333005426660272e-05,
      "loss": 0.0048,
      "step": 401
    },
    {
      "epoch": 0.15563298490127758,
      "grad_norm": 0.018714942038059235,
      "learning_rate": 8.32508111435836e-05,
      "loss": 0.0006,
      "step": 402
    },
    {
      "epoch": 0.15602013162988773,
      "grad_norm": 0.0029496820643544197,
      "learning_rate": 8.317141799492782e-05,
      "loss": 0.0001,
      "step": 403
    },
    {
      "epoch": 0.15640727835849788,
      "grad_norm": 0.01938321255147457,
      "learning_rate": 8.309187517885249e-05,
      "loss": 0.0003,
      "step": 404
    },
    {
      "epoch": 0.156794425087108,
      "grad_norm": 0.05222015455365181,
      "learning_rate": 8.301218305424994e-05,
      "loss": 0.0009,
      "step": 405
    },
    {
      "epoch": 0.15718157181571815,
      "grad_norm": 0.0011752565624192357,
      "learning_rate": 8.293234198068619e-05,
      "loss": 0.0,
      "step": 406
    },
    {
      "epoch": 0.1575687185443283,
      "grad_norm": 0.0020070699974894524,
      "learning_rate": 8.285235231839928e-05,
      "loss": 0.0001,
      "step": 407
    },
    {
      "epoch": 0.15795586527293845,
      "grad_norm": 0.04195886477828026,
      "learning_rate": 8.27722144282977e-05,
      "loss": 0.0037,
      "step": 408
    },
    {
      "epoch": 0.1583430120015486,
      "grad_norm": 0.0009243409149348736,
      "learning_rate": 8.269192867195878e-05,
      "loss": 0.0,
      "step": 409
    },
    {
      "epoch": 0.15873015873015872,
      "grad_norm": 0.0021821213886141777,
      "learning_rate": 8.261149541162691e-05,
      "loss": 0.0,
      "step": 410
    },
    {
      "epoch": 0.15911730545876887,
      "grad_norm": 0.0726943165063858,
      "learning_rate": 8.25309150102121e-05,
      "loss": 0.01,
      "step": 411
    },
    {
      "epoch": 0.15950445218737902,
      "grad_norm": 0.03284567967057228,
      "learning_rate": 8.245018783128823e-05,
      "loss": 0.0005,
      "step": 412
    },
    {
      "epoch": 0.15989159891598917,
      "grad_norm": 0.0005924602737650275,
      "learning_rate": 8.236931423909138e-05,
      "loss": 0.0,
      "step": 413
    },
    {
      "epoch": 0.1602787456445993,
      "grad_norm": 0.00531816016882658,
      "learning_rate": 8.228829459851832e-05,
      "loss": 0.0001,
      "step": 414
    },
    {
      "epoch": 0.16066589237320944,
      "grad_norm": 0.0028056297451257706,
      "learning_rate": 8.22071292751247e-05,
      "loss": 0.0,
      "step": 415
    },
    {
      "epoch": 0.1610530391018196,
      "grad_norm": 0.02348213456571102,
      "learning_rate": 8.212581863512353e-05,
      "loss": 0.0003,
      "step": 416
    },
    {
      "epoch": 0.16144018583042974,
      "grad_norm": 0.03243598714470863,
      "learning_rate": 8.204436304538349e-05,
      "loss": 0.0003,
      "step": 417
    },
    {
      "epoch": 0.1618273325590399,
      "grad_norm": 0.016708970069885254,
      "learning_rate": 8.196276287342723e-05,
      "loss": 0.0001,
      "step": 418
    },
    {
      "epoch": 0.16221447928765,
      "grad_norm": 0.010106835514307022,
      "learning_rate": 8.188101848742975e-05,
      "loss": 0.0001,
      "step": 419
    },
    {
      "epoch": 0.16260162601626016,
      "grad_norm": 0.007768136449158192,
      "learning_rate": 8.179913025621676e-05,
      "loss": 0.0,
      "step": 420
    },
    {
      "epoch": 0.1629887727448703,
      "grad_norm": 0.006284730043262243,
      "learning_rate": 8.171709854926298e-05,
      "loss": 0.0001,
      "step": 421
    },
    {
      "epoch": 0.16337591947348046,
      "grad_norm": 0.0018183101201429963,
      "learning_rate": 8.163492373669048e-05,
      "loss": 0.0,
      "step": 422
    },
    {
      "epoch": 0.16376306620209058,
      "grad_norm": 0.002550381701439619,
      "learning_rate": 8.155260618926699e-05,
      "loss": 0.0,
      "step": 423
    },
    {
      "epoch": 0.16415021293070073,
      "grad_norm": 0.0016546697588637471,
      "learning_rate": 8.14701462784043e-05,
      "loss": 0.0,
      "step": 424
    },
    {
      "epoch": 0.16453735965931088,
      "grad_norm": 0.002344394801184535,
      "learning_rate": 8.138754437615651e-05,
      "loss": 0.0,
      "step": 425
    },
    {
      "epoch": 0.16492450638792103,
      "grad_norm": 0.003373808925971389,
      "learning_rate": 8.130480085521837e-05,
      "loss": 0.0001,
      "step": 426
    },
    {
      "epoch": 0.16531165311653118,
      "grad_norm": 0.0002938521502073854,
      "learning_rate": 8.12219160889236e-05,
      "loss": 0.0,
      "step": 427
    },
    {
      "epoch": 0.1656987998451413,
      "grad_norm": 0.008733021095395088,
      "learning_rate": 8.113889045124324e-05,
      "loss": 0.0001,
      "step": 428
    },
    {
      "epoch": 0.16608594657375145,
      "grad_norm": 0.0003386593016330153,
      "learning_rate": 8.105572431678392e-05,
      "loss": 0.0,
      "step": 429
    },
    {
      "epoch": 0.1664730933023616,
      "grad_norm": 0.0021686386317014694,
      "learning_rate": 8.097241806078615e-05,
      "loss": 0.0,
      "step": 430
    },
    {
      "epoch": 0.16686024003097175,
      "grad_norm": 0.001093542668968439,
      "learning_rate": 8.088897205912272e-05,
      "loss": 0.0,
      "step": 431
    },
    {
      "epoch": 0.1672473867595819,
      "grad_norm": 0.0008129264460876584,
      "learning_rate": 8.080538668829688e-05,
      "loss": 0.0,
      "step": 432
    },
    {
      "epoch": 0.16763453348819202,
      "grad_norm": 0.000851363642141223,
      "learning_rate": 8.072166232544078e-05,
      "loss": 0.0,
      "step": 433
    },
    {
      "epoch": 0.16802168021680217,
      "grad_norm": 0.0010424451902508736,
      "learning_rate": 8.063779934831361e-05,
      "loss": 0.0,
      "step": 434
    },
    {
      "epoch": 0.16840882694541232,
      "grad_norm": 0.001159070641733706,
      "learning_rate": 8.055379813530002e-05,
      "loss": 0.0,
      "step": 435
    },
    {
      "epoch": 0.16879597367402246,
      "grad_norm": 0.001520829158835113,
      "learning_rate": 8.046965906540841e-05,
      "loss": 0.0001,
      "step": 436
    },
    {
      "epoch": 0.1691831204026326,
      "grad_norm": 0.03214537724852562,
      "learning_rate": 8.038538251826913e-05,
      "loss": 0.0002,
      "step": 437
    },
    {
      "epoch": 0.16957026713124274,
      "grad_norm": 0.000738432165235281,
      "learning_rate": 8.030096887413279e-05,
      "loss": 0.0,
      "step": 438
    },
    {
      "epoch": 0.16995741385985288,
      "grad_norm": 0.0010283209849148989,
      "learning_rate": 8.021641851386867e-05,
      "loss": 0.0,
      "step": 439
    },
    {
      "epoch": 0.17034456058846303,
      "grad_norm": 0.00038005216629244387,
      "learning_rate": 8.013173181896283e-05,
      "loss": 0.0,
      "step": 440
    },
    {
      "epoch": 0.17073170731707318,
      "grad_norm": 0.00039252720307558775,
      "learning_rate": 8.004690917151647e-05,
      "loss": 0.0,
      "step": 441
    },
    {
      "epoch": 0.1711188540456833,
      "grad_norm": 0.0010534483008086681,
      "learning_rate": 7.996195095424425e-05,
      "loss": 0.0,
      "step": 442
    },
    {
      "epoch": 0.17150600077429345,
      "grad_norm": 0.0006456160335801542,
      "learning_rate": 7.987685755047242e-05,
      "loss": 0.0,
      "step": 443
    },
    {
      "epoch": 0.1718931475029036,
      "grad_norm": 0.015872713178396225,
      "learning_rate": 7.97916293441373e-05,
      "loss": 0.0002,
      "step": 444
    },
    {
      "epoch": 0.17228029423151375,
      "grad_norm": 0.0021046926267445087,
      "learning_rate": 7.970626671978336e-05,
      "loss": 0.0,
      "step": 445
    },
    {
      "epoch": 0.17266744096012387,
      "grad_norm": 0.0005791055737063289,
      "learning_rate": 7.962077006256154e-05,
      "loss": 0.0,
      "step": 446
    },
    {
      "epoch": 0.17305458768873402,
      "grad_norm": 0.008445385843515396,
      "learning_rate": 7.953513975822755e-05,
      "loss": 0.0001,
      "step": 447
    },
    {
      "epoch": 0.17344173441734417,
      "grad_norm": 0.018906451761722565,
      "learning_rate": 7.944937619314016e-05,
      "loss": 0.0001,
      "step": 448
    },
    {
      "epoch": 0.17382888114595432,
      "grad_norm": 0.005337848793715239,
      "learning_rate": 7.93634797542593e-05,
      "loss": 0.0001,
      "step": 449
    },
    {
      "epoch": 0.17421602787456447,
      "grad_norm": 1.4769160747528076,
      "learning_rate": 7.927745082914453e-05,
      "loss": 0.0365,
      "step": 450
    },
    {
      "epoch": 0.1746031746031746,
      "grad_norm": 0.07269283384084702,
      "learning_rate": 7.919128980595309e-05,
      "loss": 0.0047,
      "step": 451
    },
    {
      "epoch": 0.17499032133178474,
      "grad_norm": 0.03315652534365654,
      "learning_rate": 7.910499707343828e-05,
      "loss": 0.001,
      "step": 452
    },
    {
      "epoch": 0.1753774680603949,
      "grad_norm": 0.013138361275196075,
      "learning_rate": 7.901857302094766e-05,
      "loss": 0.0002,
      "step": 453
    },
    {
      "epoch": 0.17576461478900504,
      "grad_norm": 0.6827967166900635,
      "learning_rate": 7.89320180384213e-05,
      "loss": 0.0006,
      "step": 454
    },
    {
      "epoch": 0.17615176151761516,
      "grad_norm": 0.023769304156303406,
      "learning_rate": 7.884533251638999e-05,
      "loss": 0.0003,
      "step": 455
    },
    {
      "epoch": 0.1765389082462253,
      "grad_norm": 0.06922416388988495,
      "learning_rate": 7.875851684597358e-05,
      "loss": 0.0029,
      "step": 456
    },
    {
      "epoch": 0.17692605497483546,
      "grad_norm": 0.03714381158351898,
      "learning_rate": 7.867157141887906e-05,
      "loss": 0.0003,
      "step": 457
    },
    {
      "epoch": 0.1773132017034456,
      "grad_norm": 0.029482917860150337,
      "learning_rate": 7.85844966273989e-05,
      "loss": 0.0004,
      "step": 458
    },
    {
      "epoch": 0.17770034843205576,
      "grad_norm": 0.0012693512253463268,
      "learning_rate": 7.849729286440928e-05,
      "loss": 0.0,
      "step": 459
    },
    {
      "epoch": 0.17808749516066588,
      "grad_norm": 0.0013141317758709192,
      "learning_rate": 7.840996052336827e-05,
      "loss": 0.0001,
      "step": 460
    },
    {
      "epoch": 0.17847464188927603,
      "grad_norm": 0.002732696942985058,
      "learning_rate": 7.832249999831407e-05,
      "loss": 0.0,
      "step": 461
    },
    {
      "epoch": 0.17886178861788618,
      "grad_norm": 0.005494223441928625,
      "learning_rate": 7.823491168386326e-05,
      "loss": 0.0001,
      "step": 462
    },
    {
      "epoch": 0.17924893534649633,
      "grad_norm": 0.0013144632102921605,
      "learning_rate": 7.814719597520894e-05,
      "loss": 0.0,
      "step": 463
    },
    {
      "epoch": 0.17963608207510648,
      "grad_norm": 0.0018678493797779083,
      "learning_rate": 7.805935326811912e-05,
      "loss": 0.0,
      "step": 464
    },
    {
      "epoch": 0.1800232288037166,
      "grad_norm": 0.004181758500635624,
      "learning_rate": 7.79713839589347e-05,
      "loss": 0.0001,
      "step": 465
    },
    {
      "epoch": 0.18041037553232675,
      "grad_norm": 0.0013430188409984112,
      "learning_rate": 7.788328844456789e-05,
      "loss": 0.0,
      "step": 466
    },
    {
      "epoch": 0.1807975222609369,
      "grad_norm": 0.00033934798557311296,
      "learning_rate": 7.779506712250024e-05,
      "loss": 0.0,
      "step": 467
    },
    {
      "epoch": 0.18118466898954705,
      "grad_norm": 0.0014731610426679254,
      "learning_rate": 7.770672039078102e-05,
      "loss": 0.0,
      "step": 468
    },
    {
      "epoch": 0.18157181571815717,
      "grad_norm": 0.012643665075302124,
      "learning_rate": 7.76182486480253e-05,
      "loss": 0.0002,
      "step": 469
    },
    {
      "epoch": 0.18195896244676732,
      "grad_norm": 0.19011537730693817,
      "learning_rate": 7.75296522934122e-05,
      "loss": 0.0057,
      "step": 470
    },
    {
      "epoch": 0.18234610917537747,
      "grad_norm": 0.002415846334770322,
      "learning_rate": 7.744093172668307e-05,
      "loss": 0.0,
      "step": 471
    },
    {
      "epoch": 0.18273325590398762,
      "grad_norm": 0.0007457759347744286,
      "learning_rate": 7.735208734813975e-05,
      "loss": 0.0,
      "step": 472
    },
    {
      "epoch": 0.18312040263259777,
      "grad_norm": 0.0009572989656589925,
      "learning_rate": 7.726311955864262e-05,
      "loss": 0.0,
      "step": 473
    },
    {
      "epoch": 0.1835075493612079,
      "grad_norm": 0.0011215334525331855,
      "learning_rate": 7.717402875960896e-05,
      "loss": 0.0,
      "step": 474
    },
    {
      "epoch": 0.18389469608981804,
      "grad_norm": 0.0025596783962100744,
      "learning_rate": 7.708481535301102e-05,
      "loss": 0.0,
      "step": 475
    },
    {
      "epoch": 0.1842818428184282,
      "grad_norm": 0.007137306500226259,
      "learning_rate": 7.699547974137426e-05,
      "loss": 0.0001,
      "step": 476
    },
    {
      "epoch": 0.18466898954703834,
      "grad_norm": 0.0028858983423560858,
      "learning_rate": 7.690602232777555e-05,
      "loss": 0.0,
      "step": 477
    },
    {
      "epoch": 0.18505613627564846,
      "grad_norm": 0.0008354281890206039,
      "learning_rate": 7.681644351584129e-05,
      "loss": 0.0,
      "step": 478
    },
    {
      "epoch": 0.1854432830042586,
      "grad_norm": 0.005381579976528883,
      "learning_rate": 7.672674370974559e-05,
      "loss": 0.0001,
      "step": 479
    },
    {
      "epoch": 0.18583042973286876,
      "grad_norm": 0.001957010943442583,
      "learning_rate": 7.663692331420857e-05,
      "loss": 0.0,
      "step": 480
    },
    {
      "epoch": 0.1862175764614789,
      "grad_norm": 0.001346597564406693,
      "learning_rate": 7.654698273449435e-05,
      "loss": 0.0,
      "step": 481
    },
    {
      "epoch": 0.18660472319008906,
      "grad_norm": 0.0013993862085044384,
      "learning_rate": 7.645692237640938e-05,
      "loss": 0.0,
      "step": 482
    },
    {
      "epoch": 0.18699186991869918,
      "grad_norm": 0.0008532855426892638,
      "learning_rate": 7.636674264630049e-05,
      "loss": 0.0,
      "step": 483
    },
    {
      "epoch": 0.18737901664730933,
      "grad_norm": 0.0010875778971239924,
      "learning_rate": 7.627644395105315e-05,
      "loss": 0.0,
      "step": 484
    },
    {
      "epoch": 0.18776616337591948,
      "grad_norm": 0.0006594783044420183,
      "learning_rate": 7.618602669808959e-05,
      "loss": 0.0,
      "step": 485
    },
    {
      "epoch": 0.18815331010452963,
      "grad_norm": 0.000321607367368415,
      "learning_rate": 7.609549129536693e-05,
      "loss": 0.0,
      "step": 486
    },
    {
      "epoch": 0.18854045683313975,
      "grad_norm": 0.0007030332926660776,
      "learning_rate": 7.60048381513754e-05,
      "loss": 0.0,
      "step": 487
    },
    {
      "epoch": 0.1889276035617499,
      "grad_norm": 0.006779205519706011,
      "learning_rate": 7.59140676751365e-05,
      "loss": 0.0001,
      "step": 488
    },
    {
      "epoch": 0.18931475029036005,
      "grad_norm": 0.0013242230052128434,
      "learning_rate": 7.582318027620105e-05,
      "loss": 0.0,
      "step": 489
    },
    {
      "epoch": 0.1897018970189702,
      "grad_norm": 0.018140189349651337,
      "learning_rate": 7.573217636464751e-05,
      "loss": 0.0003,
      "step": 490
    },
    {
      "epoch": 0.19008904374758034,
      "grad_norm": 0.0014813413145020604,
      "learning_rate": 7.564105635107996e-05,
      "loss": 0.0,
      "step": 491
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 0.0025619007647037506,
      "learning_rate": 7.554982064662637e-05,
      "loss": 0.0001,
      "step": 492
    },
    {
      "epoch": 0.19086333720480061,
      "grad_norm": 0.0014394291210919619,
      "learning_rate": 7.545846966293668e-05,
      "loss": 0.0,
      "step": 493
    },
    {
      "epoch": 0.19125048393341076,
      "grad_norm": 0.0014465994900092483,
      "learning_rate": 7.536700381218098e-05,
      "loss": 0.0,
      "step": 494
    },
    {
      "epoch": 0.1916376306620209,
      "grad_norm": 0.0030286244582384825,
      "learning_rate": 7.527542350704759e-05,
      "loss": 0.0001,
      "step": 495
    },
    {
      "epoch": 0.19202477739063106,
      "grad_norm": 0.09401986002922058,
      "learning_rate": 7.518372916074132e-05,
      "loss": 0.0004,
      "step": 496
    },
    {
      "epoch": 0.19241192411924118,
      "grad_norm": 0.017964961007237434,
      "learning_rate": 7.509192118698146e-05,
      "loss": 0.0002,
      "step": 497
    },
    {
      "epoch": 0.19279907084785133,
      "grad_norm": 0.010520697571337223,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.0001,
      "step": 498
    },
    {
      "epoch": 0.19318621757646148,
      "grad_norm": 2.2694268226623535,
      "learning_rate": 7.490796601453976e-05,
      "loss": 0.1216,
      "step": 499
    },
    {
      "epoch": 0.19357336430507163,
      "grad_norm": 0.40960317850112915,
      "learning_rate": 7.481581964585245e-05,
      "loss": 0.0043,
      "step": 500
    },
    {
      "epoch": 0.19396051103368175,
      "grad_norm": 0.0367998331785202,
      "learning_rate": 7.472356130969689e-05,
      "loss": 0.005,
      "step": 501
    },
    {
      "epoch": 0.1943476577622919,
      "grad_norm": 0.0033928700722754,
      "learning_rate": 7.46311914223371e-05,
      "loss": 0.0001,
      "step": 502
    },
    {
      "epoch": 0.19473480449090205,
      "grad_norm": 0.009577157907187939,
      "learning_rate": 7.453871040054037e-05,
      "loss": 0.0002,
      "step": 503
    },
    {
      "epoch": 0.1951219512195122,
      "grad_norm": 0.0033961099106818438,
      "learning_rate": 7.444611866157544e-05,
      "loss": 0.0001,
      "step": 504
    },
    {
      "epoch": 0.19550909794812235,
      "grad_norm": 0.015139970928430557,
      "learning_rate": 7.435341662321062e-05,
      "loss": 0.0004,
      "step": 505
    },
    {
      "epoch": 0.19589624467673247,
      "grad_norm": 0.001446568756364286,
      "learning_rate": 7.426060470371185e-05,
      "loss": 0.0001,
      "step": 506
    },
    {
      "epoch": 0.19628339140534262,
      "grad_norm": 0.03114064782857895,
      "learning_rate": 7.416768332184088e-05,
      "loss": 0.0009,
      "step": 507
    },
    {
      "epoch": 0.19667053813395277,
      "grad_norm": 0.012780173681676388,
      "learning_rate": 7.407465289685333e-05,
      "loss": 0.0003,
      "step": 508
    },
    {
      "epoch": 0.19705768486256292,
      "grad_norm": 0.07623110711574554,
      "learning_rate": 7.39815138484968e-05,
      "loss": 0.0003,
      "step": 509
    },
    {
      "epoch": 0.19744483159117304,
      "grad_norm": 0.03304853290319443,
      "learning_rate": 7.388826659700902e-05,
      "loss": 0.0002,
      "step": 510
    },
    {
      "epoch": 0.1978319783197832,
      "grad_norm": 0.002419049385935068,
      "learning_rate": 7.379491156311589e-05,
      "loss": 0.0001,
      "step": 511
    },
    {
      "epoch": 0.19821912504839334,
      "grad_norm": 0.008822989650070667,
      "learning_rate": 7.370144916802969e-05,
      "loss": 0.0002,
      "step": 512
    },
    {
      "epoch": 0.1986062717770035,
      "grad_norm": 0.0024873509537428617,
      "learning_rate": 7.360787983344704e-05,
      "loss": 0.0001,
      "step": 513
    },
    {
      "epoch": 0.19899341850561364,
      "grad_norm": 0.2240598499774933,
      "learning_rate": 7.351420398154705e-05,
      "loss": 0.0047,
      "step": 514
    },
    {
      "epoch": 0.19938056523422376,
      "grad_norm": 0.03293812274932861,
      "learning_rate": 7.342042203498951e-05,
      "loss": 0.0004,
      "step": 515
    },
    {
      "epoch": 0.1997677119628339,
      "grad_norm": 0.0035280841402709484,
      "learning_rate": 7.332653441691285e-05,
      "loss": 0.0001,
      "step": 516
    },
    {
      "epoch": 0.20015485869144406,
      "grad_norm": 0.33550676703453064,
      "learning_rate": 7.323254155093225e-05,
      "loss": 0.0161,
      "step": 517
    },
    {
      "epoch": 0.2005420054200542,
      "grad_norm": 0.009800751693546772,
      "learning_rate": 7.313844386113784e-05,
      "loss": 0.0001,
      "step": 518
    },
    {
      "epoch": 0.20092915214866433,
      "grad_norm": 0.13843496143817902,
      "learning_rate": 7.304424177209268e-05,
      "loss": 0.0008,
      "step": 519
    },
    {
      "epoch": 0.20131629887727448,
      "grad_norm": 0.07233725488185883,
      "learning_rate": 7.294993570883082e-05,
      "loss": 0.0004,
      "step": 520
    },
    {
      "epoch": 0.20170344560588463,
      "grad_norm": 0.005212654825299978,
      "learning_rate": 7.285552609685551e-05,
      "loss": 0.0001,
      "step": 521
    },
    {
      "epoch": 0.20209059233449478,
      "grad_norm": 0.002458937931805849,
      "learning_rate": 7.276101336213719e-05,
      "loss": 0.0001,
      "step": 522
    },
    {
      "epoch": 0.20247773906310493,
      "grad_norm": 0.011983088217675686,
      "learning_rate": 7.266639793111151e-05,
      "loss": 0.0003,
      "step": 523
    },
    {
      "epoch": 0.20286488579171505,
      "grad_norm": 0.003091826569288969,
      "learning_rate": 7.257168023067759e-05,
      "loss": 0.0001,
      "step": 524
    },
    {
      "epoch": 0.2032520325203252,
      "grad_norm": 0.06563227623701096,
      "learning_rate": 7.247686068819593e-05,
      "loss": 0.0004,
      "step": 525
    },
    {
      "epoch": 0.20363917924893535,
      "grad_norm": 0.002093779155984521,
      "learning_rate": 7.238193973148651e-05,
      "loss": 0.0001,
      "step": 526
    },
    {
      "epoch": 0.2040263259775455,
      "grad_norm": 0.048385411500930786,
      "learning_rate": 7.228691778882693e-05,
      "loss": 0.0008,
      "step": 527
    },
    {
      "epoch": 0.20441347270615565,
      "grad_norm": 0.00616808095946908,
      "learning_rate": 7.219179528895038e-05,
      "loss": 0.0001,
      "step": 528
    },
    {
      "epoch": 0.20480061943476577,
      "grad_norm": 0.003878154791891575,
      "learning_rate": 7.209657266104385e-05,
      "loss": 0.0001,
      "step": 529
    },
    {
      "epoch": 0.20518776616337592,
      "grad_norm": 0.008397274650633335,
      "learning_rate": 7.200125033474598e-05,
      "loss": 0.0002,
      "step": 530
    },
    {
      "epoch": 0.20557491289198607,
      "grad_norm": 0.08736693114042282,
      "learning_rate": 7.190582874014535e-05,
      "loss": 0.001,
      "step": 531
    },
    {
      "epoch": 0.20596205962059622,
      "grad_norm": 0.004129898734390736,
      "learning_rate": 7.181030830777837e-05,
      "loss": 0.0001,
      "step": 532
    },
    {
      "epoch": 0.20634920634920634,
      "grad_norm": 0.011208189651370049,
      "learning_rate": 7.171468946862744e-05,
      "loss": 0.0003,
      "step": 533
    },
    {
      "epoch": 0.2067363530778165,
      "grad_norm": 0.008657621219754219,
      "learning_rate": 7.161897265411891e-05,
      "loss": 0.0003,
      "step": 534
    },
    {
      "epoch": 0.20712349980642664,
      "grad_norm": 0.002974047791212797,
      "learning_rate": 7.152315829612125e-05,
      "loss": 0.0001,
      "step": 535
    },
    {
      "epoch": 0.20751064653503679,
      "grad_norm": 0.007309361360967159,
      "learning_rate": 7.142724682694299e-05,
      "loss": 0.0002,
      "step": 536
    },
    {
      "epoch": 0.20789779326364694,
      "grad_norm": 0.25275537371635437,
      "learning_rate": 7.133123867933087e-05,
      "loss": 0.0023,
      "step": 537
    },
    {
      "epoch": 0.20828493999225706,
      "grad_norm": 0.010795282199978828,
      "learning_rate": 7.12351342864678e-05,
      "loss": 0.0001,
      "step": 538
    },
    {
      "epoch": 0.2086720867208672,
      "grad_norm": 0.0217526163905859,
      "learning_rate": 7.113893408197092e-05,
      "loss": 0.0002,
      "step": 539
    },
    {
      "epoch": 0.20905923344947736,
      "grad_norm": 0.013393939472734928,
      "learning_rate": 7.104263849988976e-05,
      "loss": 0.0002,
      "step": 540
    },
    {
      "epoch": 0.2094463801780875,
      "grad_norm": 0.01084110513329506,
      "learning_rate": 7.094624797470407e-05,
      "loss": 0.0002,
      "step": 541
    },
    {
      "epoch": 0.20983352690669763,
      "grad_norm": 0.003147432114928961,
      "learning_rate": 7.084976294132208e-05,
      "loss": 0.0001,
      "step": 542
    },
    {
      "epoch": 0.21022067363530778,
      "grad_norm": 0.0023937560617923737,
      "learning_rate": 7.075318383507836e-05,
      "loss": 0.0001,
      "step": 543
    },
    {
      "epoch": 0.21060782036391792,
      "grad_norm": 0.0013290519127622247,
      "learning_rate": 7.065651109173197e-05,
      "loss": 0.0,
      "step": 544
    },
    {
      "epoch": 0.21099496709252807,
      "grad_norm": 0.16888383030891418,
      "learning_rate": 7.055974514746446e-05,
      "loss": 0.0018,
      "step": 545
    },
    {
      "epoch": 0.21138211382113822,
      "grad_norm": 0.014458446763455868,
      "learning_rate": 7.046288643887784e-05,
      "loss": 0.0001,
      "step": 546
    },
    {
      "epoch": 0.21176926054974834,
      "grad_norm": 0.002114094328135252,
      "learning_rate": 7.036593540299274e-05,
      "loss": 0.0,
      "step": 547
    },
    {
      "epoch": 0.2121564072783585,
      "grad_norm": 0.0411420576274395,
      "learning_rate": 7.026889247724635e-05,
      "loss": 0.0005,
      "step": 548
    },
    {
      "epoch": 0.21254355400696864,
      "grad_norm": 0.0021586837247014046,
      "learning_rate": 7.017175809949044e-05,
      "loss": 0.0001,
      "step": 549
    },
    {
      "epoch": 0.2129307007355788,
      "grad_norm": 0.08301404863595963,
      "learning_rate": 7.007453270798937e-05,
      "loss": 0.0017,
      "step": 550
    },
    {
      "epoch": 0.21331784746418891,
      "grad_norm": 0.0510292649269104,
      "learning_rate": 6.997721674141823e-05,
      "loss": 0.01,
      "step": 551
    },
    {
      "epoch": 0.21370499419279906,
      "grad_norm": 0.010133273899555206,
      "learning_rate": 6.987981063886074e-05,
      "loss": 0.0006,
      "step": 552
    },
    {
      "epoch": 0.2140921409214092,
      "grad_norm": 0.001888690865598619,
      "learning_rate": 6.978231483980728e-05,
      "loss": 0.0001,
      "step": 553
    },
    {
      "epoch": 0.21447928765001936,
      "grad_norm": 0.017150606960058212,
      "learning_rate": 6.968472978415301e-05,
      "loss": 0.0002,
      "step": 554
    },
    {
      "epoch": 0.2148664343786295,
      "grad_norm": 0.0014273010892793536,
      "learning_rate": 6.95870559121957e-05,
      "loss": 0.0,
      "step": 555
    },
    {
      "epoch": 0.21525358110723963,
      "grad_norm": 0.009298505261540413,
      "learning_rate": 6.948929366463396e-05,
      "loss": 0.0003,
      "step": 556
    },
    {
      "epoch": 0.21564072783584978,
      "grad_norm": 0.01824180595576763,
      "learning_rate": 6.939144348256511e-05,
      "loss": 0.0002,
      "step": 557
    },
    {
      "epoch": 0.21602787456445993,
      "grad_norm": 0.008332878351211548,
      "learning_rate": 6.92935058074832e-05,
      "loss": 0.0001,
      "step": 558
    },
    {
      "epoch": 0.21641502129307008,
      "grad_norm": 0.0005384304095059633,
      "learning_rate": 6.919548108127705e-05,
      "loss": 0.0,
      "step": 559
    },
    {
      "epoch": 0.21680216802168023,
      "grad_norm": 0.9911210536956787,
      "learning_rate": 6.909736974622827e-05,
      "loss": 0.1797,
      "step": 560
    },
    {
      "epoch": 0.21718931475029035,
      "grad_norm": 0.002593154553323984,
      "learning_rate": 6.899917224500925e-05,
      "loss": 0.0,
      "step": 561
    },
    {
      "epoch": 0.2175764614789005,
      "grad_norm": 0.025043247267603874,
      "learning_rate": 6.890088902068111e-05,
      "loss": 0.0001,
      "step": 562
    },
    {
      "epoch": 0.21796360820751065,
      "grad_norm": 0.0018312917090952396,
      "learning_rate": 6.88025205166918e-05,
      "loss": 0.0,
      "step": 563
    },
    {
      "epoch": 0.2183507549361208,
      "grad_norm": 0.0029724184423685074,
      "learning_rate": 6.870406717687402e-05,
      "loss": 0.0,
      "step": 564
    },
    {
      "epoch": 0.21873790166473092,
      "grad_norm": 0.0035826992243528366,
      "learning_rate": 6.860552944544325e-05,
      "loss": 0.0001,
      "step": 565
    },
    {
      "epoch": 0.21912504839334107,
      "grad_norm": 0.103156678378582,
      "learning_rate": 6.850690776699573e-05,
      "loss": 0.0007,
      "step": 566
    },
    {
      "epoch": 0.21951219512195122,
      "grad_norm": 0.003371761180460453,
      "learning_rate": 6.840820258650646e-05,
      "loss": 0.0001,
      "step": 567
    },
    {
      "epoch": 0.21989934185056137,
      "grad_norm": 0.0007010716944932938,
      "learning_rate": 6.830941434932726e-05,
      "loss": 0.0,
      "step": 568
    },
    {
      "epoch": 0.22028648857917152,
      "grad_norm": 0.061280257999897,
      "learning_rate": 6.821054350118458e-05,
      "loss": 0.0003,
      "step": 569
    },
    {
      "epoch": 0.22067363530778164,
      "grad_norm": 0.000763554242439568,
      "learning_rate": 6.811159048817772e-05,
      "loss": 0.0,
      "step": 570
    },
    {
      "epoch": 0.2210607820363918,
      "grad_norm": 0.028872434049844742,
      "learning_rate": 6.801255575677663e-05,
      "loss": 0.0002,
      "step": 571
    },
    {
      "epoch": 0.22144792876500194,
      "grad_norm": 0.005142708774656057,
      "learning_rate": 6.791343975382e-05,
      "loss": 0.0,
      "step": 572
    },
    {
      "epoch": 0.2218350754936121,
      "grad_norm": 0.0014610120560973883,
      "learning_rate": 6.78142429265132e-05,
      "loss": 0.0,
      "step": 573
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 0.0011419376824051142,
      "learning_rate": 6.771496572242627e-05,
      "loss": 0.0,
      "step": 574
    },
    {
      "epoch": 0.22260936895083236,
      "grad_norm": 0.005002702586352825,
      "learning_rate": 6.761560858949193e-05,
      "loss": 0.0001,
      "step": 575
    },
    {
      "epoch": 0.2229965156794425,
      "grad_norm": 0.001541085192002356,
      "learning_rate": 6.75161719760035e-05,
      "loss": 0.0,
      "step": 576
    },
    {
      "epoch": 0.22338366240805266,
      "grad_norm": 0.0024027032777667046,
      "learning_rate": 6.741665633061297e-05,
      "loss": 0.0,
      "step": 577
    },
    {
      "epoch": 0.2237708091366628,
      "grad_norm": 0.014390048570930958,
      "learning_rate": 6.731706210232883e-05,
      "loss": 0.0,
      "step": 578
    },
    {
      "epoch": 0.22415795586527293,
      "grad_norm": 0.005124339833855629,
      "learning_rate": 6.72173897405142e-05,
      "loss": 0.0001,
      "step": 579
    },
    {
      "epoch": 0.22454510259388308,
      "grad_norm": 0.005504544824361801,
      "learning_rate": 6.711763969488471e-05,
      "loss": 0.0,
      "step": 580
    },
    {
      "epoch": 0.22493224932249323,
      "grad_norm": 0.0024512598756700754,
      "learning_rate": 6.701781241550649e-05,
      "loss": 0.0,
      "step": 581
    },
    {
      "epoch": 0.22531939605110338,
      "grad_norm": 0.0021401650737971067,
      "learning_rate": 6.691790835279417e-05,
      "loss": 0.0,
      "step": 582
    },
    {
      "epoch": 0.2257065427797135,
      "grad_norm": 0.0038466479163616896,
      "learning_rate": 6.681792795750875e-05,
      "loss": 0.0,
      "step": 583
    },
    {
      "epoch": 0.22609368950832365,
      "grad_norm": 0.006864097900688648,
      "learning_rate": 6.671787168075577e-05,
      "loss": 0.0001,
      "step": 584
    },
    {
      "epoch": 0.2264808362369338,
      "grad_norm": 0.00382824894040823,
      "learning_rate": 6.661773997398298e-05,
      "loss": 0.0001,
      "step": 585
    },
    {
      "epoch": 0.22686798296554395,
      "grad_norm": 0.0022931089624762535,
      "learning_rate": 6.651753328897862e-05,
      "loss": 0.0,
      "step": 586
    },
    {
      "epoch": 0.2272551296941541,
      "grad_norm": 0.01140713132917881,
      "learning_rate": 6.64172520778691e-05,
      "loss": 0.0001,
      "step": 587
    },
    {
      "epoch": 0.22764227642276422,
      "grad_norm": 0.0010691037168726325,
      "learning_rate": 6.631689679311716e-05,
      "loss": 0.0,
      "step": 588
    },
    {
      "epoch": 0.22802942315137437,
      "grad_norm": 0.009856486693024635,
      "learning_rate": 6.621646788751978e-05,
      "loss": 0.0001,
      "step": 589
    },
    {
      "epoch": 0.22841656987998452,
      "grad_norm": 0.0005757238250225782,
      "learning_rate": 6.6115965814206e-05,
      "loss": 0.0,
      "step": 590
    },
    {
      "epoch": 0.22880371660859466,
      "grad_norm": 0.032714247703552246,
      "learning_rate": 6.60153910266351e-05,
      "loss": 0.0001,
      "step": 591
    },
    {
      "epoch": 0.22919086333720481,
      "grad_norm": 0.0019227054435759783,
      "learning_rate": 6.591474397859438e-05,
      "loss": 0.0,
      "step": 592
    },
    {
      "epoch": 0.22957801006581494,
      "grad_norm": 0.020541738718748093,
      "learning_rate": 6.581402512419724e-05,
      "loss": 0.0002,
      "step": 593
    },
    {
      "epoch": 0.22996515679442509,
      "grad_norm": 0.000846848648507148,
      "learning_rate": 6.571323491788098e-05,
      "loss": 0.0,
      "step": 594
    },
    {
      "epoch": 0.23035230352303523,
      "grad_norm": 0.006502535659819841,
      "learning_rate": 6.561237381440491e-05,
      "loss": 0.0001,
      "step": 595
    },
    {
      "epoch": 0.23073945025164538,
      "grad_norm": 0.002776126377284527,
      "learning_rate": 6.551144226884816e-05,
      "loss": 0.0,
      "step": 596
    },
    {
      "epoch": 0.2311265969802555,
      "grad_norm": 0.008968628011643887,
      "learning_rate": 6.541044073660773e-05,
      "loss": 0.0001,
      "step": 597
    },
    {
      "epoch": 0.23151374370886565,
      "grad_norm": 0.0025781530421227217,
      "learning_rate": 6.530936967339641e-05,
      "loss": 0.0,
      "step": 598
    },
    {
      "epoch": 0.2319008904374758,
      "grad_norm": 0.012835104949772358,
      "learning_rate": 6.520822953524064e-05,
      "loss": 0.0001,
      "step": 599
    },
    {
      "epoch": 0.23228803716608595,
      "grad_norm": 0.0038900829385966063,
      "learning_rate": 6.510702077847863e-05,
      "loss": 0.0001,
      "step": 600
    },
    {
      "epoch": 0.2326751838946961,
      "grad_norm": 0.02674148790538311,
      "learning_rate": 6.500574385975811e-05,
      "loss": 0.0034,
      "step": 601
    },
    {
      "epoch": 0.23306233062330622,
      "grad_norm": 0.013194573111832142,
      "learning_rate": 6.490439923603436e-05,
      "loss": 0.0003,
      "step": 602
    },
    {
      "epoch": 0.23344947735191637,
      "grad_norm": 0.0062402160838246346,
      "learning_rate": 6.480298736456814e-05,
      "loss": 0.0001,
      "step": 603
    },
    {
      "epoch": 0.23383662408052652,
      "grad_norm": 0.009061329998075962,
      "learning_rate": 6.470150870292369e-05,
      "loss": 0.0001,
      "step": 604
    },
    {
      "epoch": 0.23422377080913667,
      "grad_norm": 0.00190588622353971,
      "learning_rate": 6.459996370896653e-05,
      "loss": 0.0,
      "step": 605
    },
    {
      "epoch": 0.2346109175377468,
      "grad_norm": 0.003754197619855404,
      "learning_rate": 6.449835284086147e-05,
      "loss": 0.0001,
      "step": 606
    },
    {
      "epoch": 0.23499806426635694,
      "grad_norm": 0.005926064681261778,
      "learning_rate": 6.439667655707062e-05,
      "loss": 0.0001,
      "step": 607
    },
    {
      "epoch": 0.2353852109949671,
      "grad_norm": 0.004422381520271301,
      "learning_rate": 6.429493531635115e-05,
      "loss": 0.0001,
      "step": 608
    },
    {
      "epoch": 0.23577235772357724,
      "grad_norm": 0.01841561682522297,
      "learning_rate": 6.419312957775334e-05,
      "loss": 0.0003,
      "step": 609
    },
    {
      "epoch": 0.2361595044521874,
      "grad_norm": 0.0049548279494047165,
      "learning_rate": 6.409125980061853e-05,
      "loss": 0.0001,
      "step": 610
    },
    {
      "epoch": 0.2365466511807975,
      "grad_norm": 0.002091424772515893,
      "learning_rate": 6.39893264445769e-05,
      "loss": 0.0,
      "step": 611
    },
    {
      "epoch": 0.23693379790940766,
      "grad_norm": 0.0005297800526022911,
      "learning_rate": 6.388732996954559e-05,
      "loss": 0.0,
      "step": 612
    },
    {
      "epoch": 0.2373209446380178,
      "grad_norm": 0.0013506063260138035,
      "learning_rate": 6.378527083572646e-05,
      "loss": 0.0,
      "step": 613
    },
    {
      "epoch": 0.23770809136662796,
      "grad_norm": 0.0941486805677414,
      "learning_rate": 6.368314950360415e-05,
      "loss": 0.0005,
      "step": 614
    },
    {
      "epoch": 0.23809523809523808,
      "grad_norm": 0.0013530774740502238,
      "learning_rate": 6.358096643394387e-05,
      "loss": 0.0,
      "step": 615
    },
    {
      "epoch": 0.23848238482384823,
      "grad_norm": 0.00040758828981779516,
      "learning_rate": 6.34787220877894e-05,
      "loss": 0.0,
      "step": 616
    },
    {
      "epoch": 0.23886953155245838,
      "grad_norm": 0.0036245647352188826,
      "learning_rate": 6.337641692646106e-05,
      "loss": 0.0,
      "step": 617
    },
    {
      "epoch": 0.23925667828106853,
      "grad_norm": 0.002493639010936022,
      "learning_rate": 6.327405141155344e-05,
      "loss": 0.0,
      "step": 618
    },
    {
      "epoch": 0.23964382500967868,
      "grad_norm": 0.01201645378023386,
      "learning_rate": 6.317162600493357e-05,
      "loss": 0.0002,
      "step": 619
    },
    {
      "epoch": 0.2400309717382888,
      "grad_norm": 0.0034420141018927097,
      "learning_rate": 6.306914116873863e-05,
      "loss": 0.0001,
      "step": 620
    },
    {
      "epoch": 0.24041811846689895,
      "grad_norm": 0.08557692170143127,
      "learning_rate": 6.2966597365374e-05,
      "loss": 0.0005,
      "step": 621
    },
    {
      "epoch": 0.2408052651955091,
      "grad_norm": 0.0024753545876592398,
      "learning_rate": 6.286399505751102e-05,
      "loss": 0.0001,
      "step": 622
    },
    {
      "epoch": 0.24119241192411925,
      "grad_norm": 0.19629161059856415,
      "learning_rate": 6.276133470808509e-05,
      "loss": 0.0077,
      "step": 623
    },
    {
      "epoch": 0.2415795586527294,
      "grad_norm": 0.0015175496228039265,
      "learning_rate": 6.265861678029347e-05,
      "loss": 0.0,
      "step": 624
    },
    {
      "epoch": 0.24196670538133952,
      "grad_norm": 0.0008467718726024032,
      "learning_rate": 6.255584173759319e-05,
      "loss": 0.0,
      "step": 625
    },
    {
      "epoch": 0.24235385210994967,
      "grad_norm": 0.0015436846297234297,
      "learning_rate": 6.2453010043699e-05,
      "loss": 0.0,
      "step": 626
    },
    {
      "epoch": 0.24274099883855982,
      "grad_norm": 0.005769604351371527,
      "learning_rate": 6.235012216258121e-05,
      "loss": 0.0001,
      "step": 627
    },
    {
      "epoch": 0.24312814556716997,
      "grad_norm": 0.009738071821630001,
      "learning_rate": 6.224717855846374e-05,
      "loss": 0.0002,
      "step": 628
    },
    {
      "epoch": 0.2435152922957801,
      "grad_norm": 0.002359709469601512,
      "learning_rate": 6.214417969582181e-05,
      "loss": 0.0001,
      "step": 629
    },
    {
      "epoch": 0.24390243902439024,
      "grad_norm": 0.001280121854506433,
      "learning_rate": 6.204112603938006e-05,
      "loss": 0.0,
      "step": 630
    },
    {
      "epoch": 0.2442895857530004,
      "grad_norm": 0.002594890771433711,
      "learning_rate": 6.193801805411028e-05,
      "loss": 0.0,
      "step": 631
    },
    {
      "epoch": 0.24467673248161054,
      "grad_norm": 0.0023838337510824203,
      "learning_rate": 6.183485620522947e-05,
      "loss": 0.0,
      "step": 632
    },
    {
      "epoch": 0.2450638792102207,
      "grad_norm": 0.011900877580046654,
      "learning_rate": 6.173164095819758e-05,
      "loss": 0.0001,
      "step": 633
    },
    {
      "epoch": 0.2454510259388308,
      "grad_norm": 0.07067535817623138,
      "learning_rate": 6.162837277871553e-05,
      "loss": 0.0003,
      "step": 634
    },
    {
      "epoch": 0.24583817266744096,
      "grad_norm": 0.00056896434398368,
      "learning_rate": 6.152505213272307e-05,
      "loss": 0.0,
      "step": 635
    },
    {
      "epoch": 0.2462253193960511,
      "grad_norm": 0.002547027077525854,
      "learning_rate": 6.142167948639667e-05,
      "loss": 0.0,
      "step": 636
    },
    {
      "epoch": 0.24661246612466126,
      "grad_norm": 0.004246006719768047,
      "learning_rate": 6.131825530614741e-05,
      "loss": 0.0,
      "step": 637
    },
    {
      "epoch": 0.24699961285327138,
      "grad_norm": 0.0010026529198512435,
      "learning_rate": 6.121478005861891e-05,
      "loss": 0.0,
      "step": 638
    },
    {
      "epoch": 0.24738675958188153,
      "grad_norm": 0.0007125648553483188,
      "learning_rate": 6.111125421068518e-05,
      "loss": 0.0,
      "step": 639
    },
    {
      "epoch": 0.24777390631049168,
      "grad_norm": 0.0025278409011662006,
      "learning_rate": 6.100767822944856e-05,
      "loss": 0.0,
      "step": 640
    },
    {
      "epoch": 0.24816105303910183,
      "grad_norm": 0.0011645531048998237,
      "learning_rate": 6.090405258223756e-05,
      "loss": 0.0,
      "step": 641
    },
    {
      "epoch": 0.24854819976771197,
      "grad_norm": 0.018444223329424858,
      "learning_rate": 6.0800377736604806e-05,
      "loss": 0.0002,
      "step": 642
    },
    {
      "epoch": 0.2489353464963221,
      "grad_norm": 0.00047034319140948355,
      "learning_rate": 6.069665416032487e-05,
      "loss": 0.0,
      "step": 643
    },
    {
      "epoch": 0.24932249322493225,
      "grad_norm": 0.0012211957946419716,
      "learning_rate": 6.059288232139225e-05,
      "loss": 0.0,
      "step": 644
    },
    {
      "epoch": 0.2497096399535424,
      "grad_norm": 0.0005616419948637486,
      "learning_rate": 6.0489062688019146e-05,
      "loss": 0.0,
      "step": 645
    },
    {
      "epoch": 0.2500967866821525,
      "grad_norm": 0.0016151750460267067,
      "learning_rate": 6.038519572863346e-05,
      "loss": 0.0,
      "step": 646
    },
    {
      "epoch": 0.2504839334107627,
      "grad_norm": 0.008833995088934898,
      "learning_rate": 6.028128191187654e-05,
      "loss": 0.0002,
      "step": 647
    },
    {
      "epoch": 0.2508710801393728,
      "grad_norm": 0.004646562971174717,
      "learning_rate": 6.017732170660123e-05,
      "loss": 0.0001,
      "step": 648
    },
    {
      "epoch": 0.251258226867983,
      "grad_norm": 0.006479341071099043,
      "learning_rate": 6.007331558186967e-05,
      "loss": 0.0001,
      "step": 649
    },
    {
      "epoch": 0.2516453735965931,
      "grad_norm": 0.020141808316111565,
      "learning_rate": 5.996926400695113e-05,
      "loss": 0.0002,
      "step": 650
    },
    {
      "epoch": 0.25203252032520324,
      "grad_norm": 0.03065257892012596,
      "learning_rate": 5.9865167451320005e-05,
      "loss": 0.003,
      "step": 651
    },
    {
      "epoch": 0.2524196670538134,
      "grad_norm": 0.009445116855204105,
      "learning_rate": 5.9761026384653616e-05,
      "loss": 0.0002,
      "step": 652
    },
    {
      "epoch": 0.25280681378242353,
      "grad_norm": 0.03650001436471939,
      "learning_rate": 5.9656841276830133e-05,
      "loss": 0.0008,
      "step": 653
    },
    {
      "epoch": 0.25319396051103366,
      "grad_norm": 0.0038452832959592342,
      "learning_rate": 5.9552612597926415e-05,
      "loss": 0.0001,
      "step": 654
    },
    {
      "epoch": 0.25358110723964383,
      "grad_norm": 0.001224176143296063,
      "learning_rate": 5.9448340818215884e-05,
      "loss": 0.0,
      "step": 655
    },
    {
      "epoch": 0.25396825396825395,
      "grad_norm": 0.002449088031426072,
      "learning_rate": 5.934402640816652e-05,
      "loss": 0.0001,
      "step": 656
    },
    {
      "epoch": 0.25435540069686413,
      "grad_norm": 0.014086930081248283,
      "learning_rate": 5.923966983843856e-05,
      "loss": 0.0001,
      "step": 657
    },
    {
      "epoch": 0.25474254742547425,
      "grad_norm": 0.00690666725859046,
      "learning_rate": 5.913527157988252e-05,
      "loss": 0.0001,
      "step": 658
    },
    {
      "epoch": 0.2551296941540844,
      "grad_norm": 0.11766976863145828,
      "learning_rate": 5.903083210353696e-05,
      "loss": 0.0008,
      "step": 659
    },
    {
      "epoch": 0.25551684088269455,
      "grad_norm": 0.0009361839038319886,
      "learning_rate": 5.892635188062646e-05,
      "loss": 0.0,
      "step": 660
    },
    {
      "epoch": 0.2559039876113047,
      "grad_norm": 0.0004829127574339509,
      "learning_rate": 5.882183138255945e-05,
      "loss": 0.0,
      "step": 661
    },
    {
      "epoch": 0.25629113433991485,
      "grad_norm": 0.02990470826625824,
      "learning_rate": 5.8717271080926004e-05,
      "loss": 0.0003,
      "step": 662
    },
    {
      "epoch": 0.25667828106852497,
      "grad_norm": 0.0015614639269188046,
      "learning_rate": 5.8612671447495884e-05,
      "loss": 0.0,
      "step": 663
    },
    {
      "epoch": 0.2570654277971351,
      "grad_norm": 0.0005513576907105744,
      "learning_rate": 5.850803295421625e-05,
      "loss": 0.0,
      "step": 664
    },
    {
      "epoch": 0.25745257452574527,
      "grad_norm": 0.017185453325510025,
      "learning_rate": 5.840335607320964e-05,
      "loss": 0.0001,
      "step": 665
    },
    {
      "epoch": 0.2578397212543554,
      "grad_norm": 0.010216176509857178,
      "learning_rate": 5.8298641276771746e-05,
      "loss": 0.0001,
      "step": 666
    },
    {
      "epoch": 0.25822686798296557,
      "grad_norm": 0.000682777666952461,
      "learning_rate": 5.819388903736937e-05,
      "loss": 0.0,
      "step": 667
    },
    {
      "epoch": 0.2586140147115757,
      "grad_norm": 0.0019596980419009924,
      "learning_rate": 5.808909982763825e-05,
      "loss": 0.0,
      "step": 668
    },
    {
      "epoch": 0.2590011614401858,
      "grad_norm": 0.0015184786170721054,
      "learning_rate": 5.79842741203809e-05,
      "loss": 0.0,
      "step": 669
    },
    {
      "epoch": 0.259388308168796,
      "grad_norm": 0.0013025728985667229,
      "learning_rate": 5.787941238856456e-05,
      "loss": 0.0,
      "step": 670
    },
    {
      "epoch": 0.2597754548974061,
      "grad_norm": 0.0020054916385561228,
      "learning_rate": 5.777451510531895e-05,
      "loss": 0.0,
      "step": 671
    },
    {
      "epoch": 0.2601626016260163,
      "grad_norm": 0.00536265317350626,
      "learning_rate": 5.7669582743934284e-05,
      "loss": 0.0001,
      "step": 672
    },
    {
      "epoch": 0.2605497483546264,
      "grad_norm": 0.00033802189864218235,
      "learning_rate": 5.756461577785892e-05,
      "loss": 0.0,
      "step": 673
    },
    {
      "epoch": 0.26093689508323653,
      "grad_norm": 0.00032143425778485835,
      "learning_rate": 5.7459614680697495e-05,
      "loss": 0.0,
      "step": 674
    },
    {
      "epoch": 0.2613240418118467,
      "grad_norm": 0.03394626826047897,
      "learning_rate": 5.735457992620851e-05,
      "loss": 0.0002,
      "step": 675
    },
    {
      "epoch": 0.26171118854045683,
      "grad_norm": 0.0029896190389990807,
      "learning_rate": 5.7249511988302415e-05,
      "loss": 0.0,
      "step": 676
    },
    {
      "epoch": 0.26209833526906695,
      "grad_norm": 0.0015679013449698687,
      "learning_rate": 5.714441134103936e-05,
      "loss": 0.0,
      "step": 677
    },
    {
      "epoch": 0.26248548199767713,
      "grad_norm": 0.00024444403243251145,
      "learning_rate": 5.7039278458627054e-05,
      "loss": 0.0,
      "step": 678
    },
    {
      "epoch": 0.26287262872628725,
      "grad_norm": 0.0015600993065163493,
      "learning_rate": 5.6934113815418665e-05,
      "loss": 0.0,
      "step": 679
    },
    {
      "epoch": 0.2632597754548974,
      "grad_norm": 0.0019227303564548492,
      "learning_rate": 5.682891788591066e-05,
      "loss": 0.0,
      "step": 680
    },
    {
      "epoch": 0.26364692218350755,
      "grad_norm": 0.00032327178632840514,
      "learning_rate": 5.67236911447407e-05,
      "loss": 0.0,
      "step": 681
    },
    {
      "epoch": 0.26403406891211767,
      "grad_norm": 0.001297106733545661,
      "learning_rate": 5.661843406668541e-05,
      "loss": 0.0,
      "step": 682
    },
    {
      "epoch": 0.26442121564072785,
      "grad_norm": 0.00023812698782421649,
      "learning_rate": 5.651314712665833e-05,
      "loss": 0.0,
      "step": 683
    },
    {
      "epoch": 0.26480836236933797,
      "grad_norm": 0.012247363105416298,
      "learning_rate": 5.640783079970774e-05,
      "loss": 0.0001,
      "step": 684
    },
    {
      "epoch": 0.26519550909794815,
      "grad_norm": 0.0006516319117508829,
      "learning_rate": 5.6302485561014475e-05,
      "loss": 0.0,
      "step": 685
    },
    {
      "epoch": 0.26558265582655827,
      "grad_norm": 0.00017600289720576257,
      "learning_rate": 5.6197111885889867e-05,
      "loss": 0.0,
      "step": 686
    },
    {
      "epoch": 0.2659698025551684,
      "grad_norm": 0.00019459190662018955,
      "learning_rate": 5.609171024977348e-05,
      "loss": 0.0,
      "step": 687
    },
    {
      "epoch": 0.26635694928377857,
      "grad_norm": 0.00027738159405998886,
      "learning_rate": 5.598628112823114e-05,
      "loss": 0.0,
      "step": 688
    },
    {
      "epoch": 0.2667440960123887,
      "grad_norm": 0.000665661646053195,
      "learning_rate": 5.588082499695261e-05,
      "loss": 0.0,
      "step": 689
    },
    {
      "epoch": 0.26713124274099886,
      "grad_norm": 0.0010483519872650504,
      "learning_rate": 5.577534233174952e-05,
      "loss": 0.0,
      "step": 690
    },
    {
      "epoch": 0.267518389469609,
      "grad_norm": 0.003276185365393758,
      "learning_rate": 5.5669833608553244e-05,
      "loss": 0.0,
      "step": 691
    },
    {
      "epoch": 0.2679055361982191,
      "grad_norm": 0.0013291811337694526,
      "learning_rate": 5.5564299303412734e-05,
      "loss": 0.0,
      "step": 692
    },
    {
      "epoch": 0.2682926829268293,
      "grad_norm": 0.00019930455891881138,
      "learning_rate": 5.545873989249235e-05,
      "loss": 0.0,
      "step": 693
    },
    {
      "epoch": 0.2686798296554394,
      "grad_norm": 0.0002764319651760161,
      "learning_rate": 5.535315585206972e-05,
      "loss": 0.0,
      "step": 694
    },
    {
      "epoch": 0.2690669763840495,
      "grad_norm": 0.0007317042327485979,
      "learning_rate": 5.52475476585336e-05,
      "loss": 0.0,
      "step": 695
    },
    {
      "epoch": 0.2694541231126597,
      "grad_norm": 2.0606796741485596,
      "learning_rate": 5.514191578838177e-05,
      "loss": 0.0594,
      "step": 696
    },
    {
      "epoch": 0.2698412698412698,
      "grad_norm": 0.029833437874913216,
      "learning_rate": 5.503626071821877e-05,
      "loss": 0.0002,
      "step": 697
    },
    {
      "epoch": 0.27022841656988,
      "grad_norm": 0.0008992796647362411,
      "learning_rate": 5.493058292475387e-05,
      "loss": 0.0,
      "step": 698
    },
    {
      "epoch": 0.2706155632984901,
      "grad_norm": 0.0011732947314158082,
      "learning_rate": 5.48248828847988e-05,
      "loss": 0.0,
      "step": 699
    },
    {
      "epoch": 0.27100271002710025,
      "grad_norm": 0.593447744846344,
      "learning_rate": 5.4719161075265765e-05,
      "loss": 0.0047,
      "step": 700
    },
    {
      "epoch": 0.2713898567557104,
      "grad_norm": 0.05254895240068436,
      "learning_rate": 5.4613417973165106e-05,
      "loss": 0.0064,
      "step": 701
    },
    {
      "epoch": 0.27177700348432055,
      "grad_norm": 0.0053640869446098804,
      "learning_rate": 5.4507654055603275e-05,
      "loss": 0.0002,
      "step": 702
    },
    {
      "epoch": 0.2721641502129307,
      "grad_norm": 0.013144250027835369,
      "learning_rate": 5.440186979978061e-05,
      "loss": 0.0002,
      "step": 703
    },
    {
      "epoch": 0.27255129694154084,
      "grad_norm": 0.005304619669914246,
      "learning_rate": 5.429606568298926e-05,
      "loss": 0.0001,
      "step": 704
    },
    {
      "epoch": 0.27293844367015097,
      "grad_norm": 0.006235075183212757,
      "learning_rate": 5.4190242182610976e-05,
      "loss": 0.0001,
      "step": 705
    },
    {
      "epoch": 0.27332559039876114,
      "grad_norm": 0.007272032089531422,
      "learning_rate": 5.4084399776114915e-05,
      "loss": 0.0002,
      "step": 706
    },
    {
      "epoch": 0.27371273712737126,
      "grad_norm": 0.008517012931406498,
      "learning_rate": 5.397853894105559e-05,
      "loss": 0.0001,
      "step": 707
    },
    {
      "epoch": 0.27409988385598144,
      "grad_norm": 0.10108551383018494,
      "learning_rate": 5.387266015507065e-05,
      "loss": 0.006,
      "step": 708
    },
    {
      "epoch": 0.27448703058459156,
      "grad_norm": 0.017542164772748947,
      "learning_rate": 5.376676389587875e-05,
      "loss": 0.0002,
      "step": 709
    },
    {
      "epoch": 0.2748741773132017,
      "grad_norm": 0.0033126119524240494,
      "learning_rate": 5.366085064127734e-05,
      "loss": 0.0001,
      "step": 710
    },
    {
      "epoch": 0.27526132404181186,
      "grad_norm": 0.004340835381299257,
      "learning_rate": 5.355492086914059e-05,
      "loss": 0.0,
      "step": 711
    },
    {
      "epoch": 0.275648470770422,
      "grad_norm": 0.0020051731262356043,
      "learning_rate": 5.344897505741719e-05,
      "loss": 0.0,
      "step": 712
    },
    {
      "epoch": 0.27603561749903216,
      "grad_norm": 0.0017423669341951609,
      "learning_rate": 5.3343013684128206e-05,
      "loss": 0.0,
      "step": 713
    },
    {
      "epoch": 0.2764227642276423,
      "grad_norm": 0.004165020305663347,
      "learning_rate": 5.323703722736489e-05,
      "loss": 0.0001,
      "step": 714
    },
    {
      "epoch": 0.2768099109562524,
      "grad_norm": 0.004314817953854799,
      "learning_rate": 5.3131046165286556e-05,
      "loss": 0.0001,
      "step": 715
    },
    {
      "epoch": 0.2771970576848626,
      "grad_norm": 0.009306171908974648,
      "learning_rate": 5.302504097611847e-05,
      "loss": 0.0001,
      "step": 716
    },
    {
      "epoch": 0.2775842044134727,
      "grad_norm": 0.002385742263868451,
      "learning_rate": 5.2919022138149555e-05,
      "loss": 0.0,
      "step": 717
    },
    {
      "epoch": 0.2779713511420828,
      "grad_norm": 0.04833089932799339,
      "learning_rate": 5.2812990129730363e-05,
      "loss": 0.0003,
      "step": 718
    },
    {
      "epoch": 0.278358497870693,
      "grad_norm": 0.001545402454212308,
      "learning_rate": 5.270694542927088e-05,
      "loss": 0.0,
      "step": 719
    },
    {
      "epoch": 0.2787456445993031,
      "grad_norm": 0.0023753156419843435,
      "learning_rate": 5.260088851523833e-05,
      "loss": 0.0,
      "step": 720
    },
    {
      "epoch": 0.2791327913279133,
      "grad_norm": 0.13964203000068665,
      "learning_rate": 5.2494819866155065e-05,
      "loss": 0.0037,
      "step": 721
    },
    {
      "epoch": 0.2795199380565234,
      "grad_norm": 0.0010260421549901366,
      "learning_rate": 5.2388739960596364e-05,
      "loss": 0.0,
      "step": 722
    },
    {
      "epoch": 0.27990708478513354,
      "grad_norm": 0.0009256897028535604,
      "learning_rate": 5.22826492771883e-05,
      "loss": 0.0,
      "step": 723
    },
    {
      "epoch": 0.2802942315137437,
      "grad_norm": 0.004000549204647541,
      "learning_rate": 5.217654829460561e-05,
      "loss": 0.0001,
      "step": 724
    },
    {
      "epoch": 0.28068137824235384,
      "grad_norm": 0.0011285142973065376,
      "learning_rate": 5.207043749156945e-05,
      "loss": 0.0,
      "step": 725
    },
    {
      "epoch": 0.281068524970964,
      "grad_norm": 0.10327895730733871,
      "learning_rate": 5.19643173468453e-05,
      "loss": 0.0009,
      "step": 726
    },
    {
      "epoch": 0.28145567169957414,
      "grad_norm": 0.000834814622066915,
      "learning_rate": 5.185818833924081e-05,
      "loss": 0.0,
      "step": 727
    },
    {
      "epoch": 0.28184281842818426,
      "grad_norm": 0.006214594002813101,
      "learning_rate": 5.1752050947603614e-05,
      "loss": 0.0,
      "step": 728
    },
    {
      "epoch": 0.28222996515679444,
      "grad_norm": 0.0009311099420301616,
      "learning_rate": 5.164590565081914e-05,
      "loss": 0.0,
      "step": 729
    },
    {
      "epoch": 0.28261711188540456,
      "grad_norm": 0.1762949526309967,
      "learning_rate": 5.153975292780853e-05,
      "loss": 0.0019,
      "step": 730
    },
    {
      "epoch": 0.28300425861401474,
      "grad_norm": 0.025049149990081787,
      "learning_rate": 5.143359325752638e-05,
      "loss": 0.0001,
      "step": 731
    },
    {
      "epoch": 0.28339140534262486,
      "grad_norm": 0.002675428753718734,
      "learning_rate": 5.13274271189587e-05,
      "loss": 0.0,
      "step": 732
    },
    {
      "epoch": 0.283778552071235,
      "grad_norm": 0.001739859115332365,
      "learning_rate": 5.122125499112063e-05,
      "loss": 0.0,
      "step": 733
    },
    {
      "epoch": 0.28416569879984516,
      "grad_norm": 0.13101710379123688,
      "learning_rate": 5.1115077353054356e-05,
      "loss": 0.0008,
      "step": 734
    },
    {
      "epoch": 0.2845528455284553,
      "grad_norm": 0.008799294009804726,
      "learning_rate": 5.10088946838269e-05,
      "loss": 0.0001,
      "step": 735
    },
    {
      "epoch": 0.28493999225706546,
      "grad_norm": 0.00784998293966055,
      "learning_rate": 5.090270746252802e-05,
      "loss": 0.0001,
      "step": 736
    },
    {
      "epoch": 0.2853271389856756,
      "grad_norm": 0.012120920233428478,
      "learning_rate": 5.079651616826802e-05,
      "loss": 0.0002,
      "step": 737
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.033871788531541824,
      "learning_rate": 5.069032128017551e-05,
      "loss": 0.0002,
      "step": 738
    },
    {
      "epoch": 0.2861014324428959,
      "grad_norm": 0.05213552340865135,
      "learning_rate": 5.05841232773954e-05,
      "loss": 0.0006,
      "step": 739
    },
    {
      "epoch": 0.286488579171506,
      "grad_norm": 0.0012123180786147714,
      "learning_rate": 5.0477922639086596e-05,
      "loss": 0.0,
      "step": 740
    },
    {
      "epoch": 0.2868757259001161,
      "grad_norm": 0.01266507152467966,
      "learning_rate": 5.037171984441994e-05,
      "loss": 0.0001,
      "step": 741
    },
    {
      "epoch": 0.2872628726287263,
      "grad_norm": 0.0523063987493515,
      "learning_rate": 5.0265515372575956e-05,
      "loss": 0.0005,
      "step": 742
    },
    {
      "epoch": 0.2876500193573364,
      "grad_norm": 0.001839546370320022,
      "learning_rate": 5.015930970274277e-05,
      "loss": 0.0,
      "step": 743
    },
    {
      "epoch": 0.2880371660859466,
      "grad_norm": 0.0016663463320583105,
      "learning_rate": 5.00531033141139e-05,
      "loss": 0.0,
      "step": 744
    },
    {
      "epoch": 0.2884243128145567,
      "grad_norm": 0.0012716924538835883,
      "learning_rate": 4.99468966858861e-05,
      "loss": 0.0,
      "step": 745
    },
    {
      "epoch": 0.28881145954316684,
      "grad_norm": 0.0038500288501381874,
      "learning_rate": 4.984069029725723e-05,
      "loss": 0.0,
      "step": 746
    },
    {
      "epoch": 0.289198606271777,
      "grad_norm": 0.3896702229976654,
      "learning_rate": 4.973448462742405e-05,
      "loss": 0.0008,
      "step": 747
    },
    {
      "epoch": 0.28958575300038714,
      "grad_norm": 0.0010479100747033954,
      "learning_rate": 4.962828015558006e-05,
      "loss": 0.0,
      "step": 748
    },
    {
      "epoch": 0.2899728997289973,
      "grad_norm": 0.011884471401572227,
      "learning_rate": 4.952207736091341e-05,
      "loss": 0.0001,
      "step": 749
    },
    {
      "epoch": 0.29036004645760743,
      "grad_norm": 0.2021235078573227,
      "learning_rate": 4.941587672260461e-05,
      "loss": 0.0033,
      "step": 750
    },
    {
      "epoch": 0.29074719318621756,
      "grad_norm": 0.045500051230192184,
      "learning_rate": 4.930967871982451e-05,
      "loss": 0.0055,
      "step": 751
    },
    {
      "epoch": 0.29113433991482773,
      "grad_norm": 0.0288743507117033,
      "learning_rate": 4.9203483831732006e-05,
      "loss": 0.0007,
      "step": 752
    },
    {
      "epoch": 0.29152148664343785,
      "grad_norm": 0.020009569823741913,
      "learning_rate": 4.909729253747197e-05,
      "loss": 0.0002,
      "step": 753
    },
    {
      "epoch": 0.29190863337204803,
      "grad_norm": 0.006925898138433695,
      "learning_rate": 4.899110531617311e-05,
      "loss": 0.0001,
      "step": 754
    },
    {
      "epoch": 0.29229578010065815,
      "grad_norm": 0.03738551586866379,
      "learning_rate": 4.8884922646945656e-05,
      "loss": 0.0006,
      "step": 755
    },
    {
      "epoch": 0.2926829268292683,
      "grad_norm": 0.0020933786872774363,
      "learning_rate": 4.877874500887938e-05,
      "loss": 0.0001,
      "step": 756
    },
    {
      "epoch": 0.29307007355787845,
      "grad_norm": 0.002445946214720607,
      "learning_rate": 4.867257288104131e-05,
      "loss": 0.0,
      "step": 757
    },
    {
      "epoch": 0.2934572202864886,
      "grad_norm": 0.01913662999868393,
      "learning_rate": 4.856640674247363e-05,
      "loss": 0.0002,
      "step": 758
    },
    {
      "epoch": 0.2938443670150987,
      "grad_norm": 0.0019830456003546715,
      "learning_rate": 4.8460247072191496e-05,
      "loss": 0.0,
      "step": 759
    },
    {
      "epoch": 0.2942315137437089,
      "grad_norm": 0.0007783794426359236,
      "learning_rate": 4.8354094349180885e-05,
      "loss": 0.0,
      "step": 760
    },
    {
      "epoch": 0.294618660472319,
      "grad_norm": 0.0003296427894383669,
      "learning_rate": 4.82479490523964e-05,
      "loss": 0.0,
      "step": 761
    },
    {
      "epoch": 0.29500580720092917,
      "grad_norm": 0.0009307400323450565,
      "learning_rate": 4.8141811660759185e-05,
      "loss": 0.0,
      "step": 762
    },
    {
      "epoch": 0.2953929539295393,
      "grad_norm": 0.0011843329994007945,
      "learning_rate": 4.8035682653154704e-05,
      "loss": 0.0,
      "step": 763
    },
    {
      "epoch": 0.2957801006581494,
      "grad_norm": 0.001043383264914155,
      "learning_rate": 4.7929562508430556e-05,
      "loss": 0.0,
      "step": 764
    },
    {
      "epoch": 0.2961672473867596,
      "grad_norm": 0.00028426575590856373,
      "learning_rate": 4.782345170539441e-05,
      "loss": 0.0,
      "step": 765
    },
    {
      "epoch": 0.2965543941153697,
      "grad_norm": 0.17862246930599213,
      "learning_rate": 4.7717350722811705e-05,
      "loss": 0.0013,
      "step": 766
    },
    {
      "epoch": 0.2969415408439799,
      "grad_norm": 0.0018550584791228175,
      "learning_rate": 4.761126003940366e-05,
      "loss": 0.0,
      "step": 767
    },
    {
      "epoch": 0.29732868757259,
      "grad_norm": 0.0009288343135267496,
      "learning_rate": 4.7505180133844954e-05,
      "loss": 0.0,
      "step": 768
    },
    {
      "epoch": 0.29771583430120013,
      "grad_norm": 0.00047417895984835923,
      "learning_rate": 4.7399111484761674e-05,
      "loss": 0.0,
      "step": 769
    },
    {
      "epoch": 0.2981029810298103,
      "grad_norm": 0.0017211599042639136,
      "learning_rate": 4.729305457072913e-05,
      "loss": 0.0,
      "step": 770
    },
    {
      "epoch": 0.29849012775842043,
      "grad_norm": 0.0004953122115693986,
      "learning_rate": 4.718700987026964e-05,
      "loss": 0.0,
      "step": 771
    },
    {
      "epoch": 0.2988772744870306,
      "grad_norm": 0.0006206472753547132,
      "learning_rate": 4.708097786185046e-05,
      "loss": 0.0,
      "step": 772
    },
    {
      "epoch": 0.29926442121564073,
      "grad_norm": 0.00035984007990919054,
      "learning_rate": 4.697495902388154e-05,
      "loss": 0.0,
      "step": 773
    },
    {
      "epoch": 0.29965156794425085,
      "grad_norm": 0.0028558659832924604,
      "learning_rate": 4.6868953834713456e-05,
      "loss": 0.0,
      "step": 774
    },
    {
      "epoch": 0.30003871467286103,
      "grad_norm": 0.0013832409167662263,
      "learning_rate": 4.6762962772635125e-05,
      "loss": 0.0,
      "step": 775
    },
    {
      "epoch": 0.30042586140147115,
      "grad_norm": 0.0032838445622473955,
      "learning_rate": 4.665698631587182e-05,
      "loss": 0.0,
      "step": 776
    },
    {
      "epoch": 0.3008130081300813,
      "grad_norm": 0.0006759579991921782,
      "learning_rate": 4.655102494258281e-05,
      "loss": 0.0,
      "step": 777
    },
    {
      "epoch": 0.30120015485869145,
      "grad_norm": 0.0005392953753471375,
      "learning_rate": 4.644507913085942e-05,
      "loss": 0.0,
      "step": 778
    },
    {
      "epoch": 0.30158730158730157,
      "grad_norm": 0.005940568167716265,
      "learning_rate": 4.633914935872268e-05,
      "loss": 0.0,
      "step": 779
    },
    {
      "epoch": 0.30197444831591175,
      "grad_norm": 0.005116292275488377,
      "learning_rate": 4.623323610412126e-05,
      "loss": 0.0,
      "step": 780
    },
    {
      "epoch": 0.30236159504452187,
      "grad_norm": 0.0008851262973621488,
      "learning_rate": 4.612733984492936e-05,
      "loss": 0.0,
      "step": 781
    },
    {
      "epoch": 0.302748741773132,
      "grad_norm": 0.000602389860432595,
      "learning_rate": 4.602146105894442e-05,
      "loss": 0.0,
      "step": 782
    },
    {
      "epoch": 0.30313588850174217,
      "grad_norm": 0.0011187748750671744,
      "learning_rate": 4.591560022388511e-05,
      "loss": 0.0,
      "step": 783
    },
    {
      "epoch": 0.3035230352303523,
      "grad_norm": 0.0007747854106128216,
      "learning_rate": 4.580975781738905e-05,
      "loss": 0.0,
      "step": 784
    },
    {
      "epoch": 0.30391018195896247,
      "grad_norm": 0.021920830011367798,
      "learning_rate": 4.5703934317010735e-05,
      "loss": 0.0003,
      "step": 785
    },
    {
      "epoch": 0.3042973286875726,
      "grad_norm": 0.0004085338150616735,
      "learning_rate": 4.559813020021939e-05,
      "loss": 0.0,
      "step": 786
    },
    {
      "epoch": 0.3046844754161827,
      "grad_norm": 0.48235246539115906,
      "learning_rate": 4.549234594439674e-05,
      "loss": 0.0031,
      "step": 787
    },
    {
      "epoch": 0.3050716221447929,
      "grad_norm": 0.00026683625765144825,
      "learning_rate": 4.5386582026834906e-05,
      "loss": 0.0,
      "step": 788
    },
    {
      "epoch": 0.305458768873403,
      "grad_norm": 0.007353804539889097,
      "learning_rate": 4.528083892473424e-05,
      "loss": 0.0001,
      "step": 789
    },
    {
      "epoch": 0.3058459156020132,
      "grad_norm": 0.0011645752238109708,
      "learning_rate": 4.517511711520121e-05,
      "loss": 0.0,
      "step": 790
    },
    {
      "epoch": 0.3062330623306233,
      "grad_norm": 0.2283429056406021,
      "learning_rate": 4.506941707524615e-05,
      "loss": 0.0007,
      "step": 791
    },
    {
      "epoch": 0.30662020905923343,
      "grad_norm": 0.0008667742949910462,
      "learning_rate": 4.496373928178125e-05,
      "loss": 0.0,
      "step": 792
    },
    {
      "epoch": 0.3070073557878436,
      "grad_norm": 0.0012296324130147696,
      "learning_rate": 4.485808421161823e-05,
      "loss": 0.0,
      "step": 793
    },
    {
      "epoch": 0.3073945025164537,
      "grad_norm": 0.0036023801658302546,
      "learning_rate": 4.4752452341466395e-05,
      "loss": 0.0,
      "step": 794
    },
    {
      "epoch": 0.3077816492450639,
      "grad_norm": 0.001286810147576034,
      "learning_rate": 4.4646844147930294e-05,
      "loss": 0.0,
      "step": 795
    },
    {
      "epoch": 0.308168795973674,
      "grad_norm": 0.0008486775914207101,
      "learning_rate": 4.454126010750766e-05,
      "loss": 0.0,
      "step": 796
    },
    {
      "epoch": 0.30855594270228415,
      "grad_norm": 0.0011814788449555635,
      "learning_rate": 4.443570069658727e-05,
      "loss": 0.0,
      "step": 797
    },
    {
      "epoch": 0.3089430894308943,
      "grad_norm": 0.011497690342366695,
      "learning_rate": 4.433016639144676e-05,
      "loss": 0.0002,
      "step": 798
    },
    {
      "epoch": 0.30933023615950445,
      "grad_norm": 0.0019194751512259245,
      "learning_rate": 4.4224657668250495e-05,
      "loss": 0.0,
      "step": 799
    },
    {
      "epoch": 0.3097173828881146,
      "grad_norm": 1.5372258424758911,
      "learning_rate": 4.411917500304741e-05,
      "loss": 0.0304,
      "step": 800
    },
    {
      "epoch": 0.31010452961672474,
      "grad_norm": 0.04230941832065582,
      "learning_rate": 4.4013718871768855e-05,
      "loss": 0.0087,
      "step": 801
    },
    {
      "epoch": 0.31049167634533487,
      "grad_norm": 0.0006967206718400121,
      "learning_rate": 4.390828975022652e-05,
      "loss": 0.0,
      "step": 802
    },
    {
      "epoch": 0.31087882307394504,
      "grad_norm": 0.011427176184952259,
      "learning_rate": 4.380288811411015e-05,
      "loss": 0.0001,
      "step": 803
    },
    {
      "epoch": 0.31126596980255516,
      "grad_norm": 0.007503742817789316,
      "learning_rate": 4.3697514438985536e-05,
      "loss": 0.0,
      "step": 804
    },
    {
      "epoch": 0.3116531165311653,
      "grad_norm": 0.005879178177565336,
      "learning_rate": 4.359216920029227e-05,
      "loss": 0.0001,
      "step": 805
    },
    {
      "epoch": 0.31204026325977546,
      "grad_norm": 0.006521875504404306,
      "learning_rate": 4.3486852873341676e-05,
      "loss": 0.0001,
      "step": 806
    },
    {
      "epoch": 0.3124274099883856,
      "grad_norm": 0.0018767932197079062,
      "learning_rate": 4.33815659333146e-05,
      "loss": 0.0,
      "step": 807
    },
    {
      "epoch": 0.31281455671699576,
      "grad_norm": 0.006302808877080679,
      "learning_rate": 4.327630885525932e-05,
      "loss": 0.0001,
      "step": 808
    },
    {
      "epoch": 0.3132017034456059,
      "grad_norm": 0.0005288548418320715,
      "learning_rate": 4.3171082114089336e-05,
      "loss": 0.0,
      "step": 809
    },
    {
      "epoch": 0.313588850174216,
      "grad_norm": 0.000991316745057702,
      "learning_rate": 4.306588618458134e-05,
      "loss": 0.0,
      "step": 810
    },
    {
      "epoch": 0.3139759969028262,
      "grad_norm": 0.00023478269577026367,
      "learning_rate": 4.2960721541372964e-05,
      "loss": 0.0,
      "step": 811
    },
    {
      "epoch": 0.3143631436314363,
      "grad_norm": 0.007332727778702974,
      "learning_rate": 4.285558865896065e-05,
      "loss": 0.0,
      "step": 812
    },
    {
      "epoch": 0.3147502903600465,
      "grad_norm": 0.0019596214406192303,
      "learning_rate": 4.275048801169759e-05,
      "loss": 0.0,
      "step": 813
    },
    {
      "epoch": 0.3151374370886566,
      "grad_norm": 0.0015021011931821704,
      "learning_rate": 4.2645420073791496e-05,
      "loss": 0.0,
      "step": 814
    },
    {
      "epoch": 0.3155245838172667,
      "grad_norm": 0.0009233247837983072,
      "learning_rate": 4.254038531930253e-05,
      "loss": 0.0,
      "step": 815
    },
    {
      "epoch": 0.3159117305458769,
      "grad_norm": 0.00019637109653558582,
      "learning_rate": 4.2435384222141085e-05,
      "loss": 0.0,
      "step": 816
    },
    {
      "epoch": 0.316298877274487,
      "grad_norm": 0.00023914426856208593,
      "learning_rate": 4.233041725606572e-05,
      "loss": 0.0,
      "step": 817
    },
    {
      "epoch": 0.3166860240030972,
      "grad_norm": 0.00042257452150806785,
      "learning_rate": 4.222548489468105e-05,
      "loss": 0.0,
      "step": 818
    },
    {
      "epoch": 0.3170731707317073,
      "grad_norm": 0.0009111497784033418,
      "learning_rate": 4.2120587611435445e-05,
      "loss": 0.0,
      "step": 819
    },
    {
      "epoch": 0.31746031746031744,
      "grad_norm": 0.0006815852248109877,
      "learning_rate": 4.201572587961911e-05,
      "loss": 0.0,
      "step": 820
    },
    {
      "epoch": 0.3178474641889276,
      "grad_norm": 0.0004083358508069068,
      "learning_rate": 4.1910900172361764e-05,
      "loss": 0.0,
      "step": 821
    },
    {
      "epoch": 0.31823461091753774,
      "grad_norm": 0.001000471180304885,
      "learning_rate": 4.1806110962630644e-05,
      "loss": 0.0,
      "step": 822
    },
    {
      "epoch": 0.31862175764614786,
      "grad_norm": 0.0013421968324109912,
      "learning_rate": 4.1701358723228266e-05,
      "loss": 0.0,
      "step": 823
    },
    {
      "epoch": 0.31900890437475804,
      "grad_norm": 0.0023856195621192455,
      "learning_rate": 4.159664392679039e-05,
      "loss": 0.0,
      "step": 824
    },
    {
      "epoch": 0.31939605110336816,
      "grad_norm": 0.000275084690656513,
      "learning_rate": 4.1491967045783755e-05,
      "loss": 0.0,
      "step": 825
    },
    {
      "epoch": 0.31978319783197834,
      "grad_norm": 0.000643594772554934,
      "learning_rate": 4.138732855250412e-05,
      "loss": 0.0,
      "step": 826
    },
    {
      "epoch": 0.32017034456058846,
      "grad_norm": 0.00018649610865395516,
      "learning_rate": 4.128272891907401e-05,
      "loss": 0.0,
      "step": 827
    },
    {
      "epoch": 0.3205574912891986,
      "grad_norm": 0.004007772076874971,
      "learning_rate": 4.117816861744057e-05,
      "loss": 0.0,
      "step": 828
    },
    {
      "epoch": 0.32094463801780876,
      "grad_norm": 0.000278328952845186,
      "learning_rate": 4.107364811937355e-05,
      "loss": 0.0,
      "step": 829
    },
    {
      "epoch": 0.3213317847464189,
      "grad_norm": 0.0021861272398382425,
      "learning_rate": 4.096916789646305e-05,
      "loss": 0.0,
      "step": 830
    },
    {
      "epoch": 0.32171893147502906,
      "grad_norm": 0.00044286204501986504,
      "learning_rate": 4.0864728420117505e-05,
      "loss": 0.0,
      "step": 831
    },
    {
      "epoch": 0.3221060782036392,
      "grad_norm": 0.00034147806582041085,
      "learning_rate": 4.0760330161561464e-05,
      "loss": 0.0,
      "step": 832
    },
    {
      "epoch": 0.3224932249322493,
      "grad_norm": 0.001119753229431808,
      "learning_rate": 4.065597359183348e-05,
      "loss": 0.0,
      "step": 833
    },
    {
      "epoch": 0.3228803716608595,
      "grad_norm": 0.0005893710185773671,
      "learning_rate": 4.055165918178412e-05,
      "loss": 0.0,
      "step": 834
    },
    {
      "epoch": 0.3232675183894696,
      "grad_norm": 0.013795977458357811,
      "learning_rate": 4.04473874020736e-05,
      "loss": 0.0001,
      "step": 835
    },
    {
      "epoch": 0.3236546651180798,
      "grad_norm": 0.000739909999538213,
      "learning_rate": 4.034315872316988e-05,
      "loss": 0.0,
      "step": 836
    },
    {
      "epoch": 0.3240418118466899,
      "grad_norm": 0.0003141966590192169,
      "learning_rate": 4.0238973615346395e-05,
      "loss": 0.0,
      "step": 837
    },
    {
      "epoch": 0.3244289585753,
      "grad_norm": 0.0013074502348899841,
      "learning_rate": 4.0134832548680006e-05,
      "loss": 0.0,
      "step": 838
    },
    {
      "epoch": 0.3248161053039102,
      "grad_norm": 0.0003308141021989286,
      "learning_rate": 4.003073599304889e-05,
      "loss": 0.0,
      "step": 839
    },
    {
      "epoch": 0.3252032520325203,
      "grad_norm": 0.0002786907716654241,
      "learning_rate": 3.9926684418130355e-05,
      "loss": 0.0,
      "step": 840
    },
    {
      "epoch": 0.3255903987611305,
      "grad_norm": 0.008617997169494629,
      "learning_rate": 3.982267829339877e-05,
      "loss": 0.0,
      "step": 841
    },
    {
      "epoch": 0.3259775454897406,
      "grad_norm": 0.004956826567649841,
      "learning_rate": 3.971871808812348e-05,
      "loss": 0.0001,
      "step": 842
    },
    {
      "epoch": 0.32636469221835074,
      "grad_norm": 0.0003836003306787461,
      "learning_rate": 3.961480427136656e-05,
      "loss": 0.0,
      "step": 843
    },
    {
      "epoch": 0.3267518389469609,
      "grad_norm": 0.0010854278225451708,
      "learning_rate": 3.951093731198086e-05,
      "loss": 0.0,
      "step": 844
    },
    {
      "epoch": 0.32713898567557104,
      "grad_norm": 0.021701844409108162,
      "learning_rate": 3.9407117678607755e-05,
      "loss": 0.0002,
      "step": 845
    },
    {
      "epoch": 0.32752613240418116,
      "grad_norm": 0.0007447968819178641,
      "learning_rate": 3.930334583967514e-05,
      "loss": 0.0,
      "step": 846
    },
    {
      "epoch": 0.32791327913279134,
      "grad_norm": 0.00031065259827300906,
      "learning_rate": 3.919962226339521e-05,
      "loss": 0.0,
      "step": 847
    },
    {
      "epoch": 0.32830042586140146,
      "grad_norm": 0.003032098524272442,
      "learning_rate": 3.9095947417762465e-05,
      "loss": 0.0,
      "step": 848
    },
    {
      "epoch": 0.32868757259001163,
      "grad_norm": 0.0012839855626225471,
      "learning_rate": 3.899232177055145e-05,
      "loss": 0.0,
      "step": 849
    },
    {
      "epoch": 0.32907471931862176,
      "grad_norm": 0.00046888040378689766,
      "learning_rate": 3.888874578931482e-05,
      "loss": 0.0,
      "step": 850
    },
    {
      "epoch": 0.3294618660472319,
      "grad_norm": 0.03994712233543396,
      "learning_rate": 3.87852199413811e-05,
      "loss": 0.0035,
      "step": 851
    },
    {
      "epoch": 0.32984901277584205,
      "grad_norm": 0.00314146070741117,
      "learning_rate": 3.8681744693852595e-05,
      "loss": 0.0001,
      "step": 852
    },
    {
      "epoch": 0.3302361595044522,
      "grad_norm": 0.013279520906507969,
      "learning_rate": 3.857832051360335e-05,
      "loss": 0.0003,
      "step": 853
    },
    {
      "epoch": 0.33062330623306235,
      "grad_norm": 0.028156403452157974,
      "learning_rate": 3.847494786727694e-05,
      "loss": 0.0005,
      "step": 854
    },
    {
      "epoch": 0.3310104529616725,
      "grad_norm": 0.038008369505405426,
      "learning_rate": 3.8371627221284495e-05,
      "loss": 0.0006,
      "step": 855
    },
    {
      "epoch": 0.3313975996902826,
      "grad_norm": 0.006248411722481251,
      "learning_rate": 3.826835904180244e-05,
      "loss": 0.0001,
      "step": 856
    },
    {
      "epoch": 0.3317847464188928,
      "grad_norm": 0.00033054646337404847,
      "learning_rate": 3.816514379477054e-05,
      "loss": 0.0,
      "step": 857
    },
    {
      "epoch": 0.3321718931475029,
      "grad_norm": 0.0008822673698887229,
      "learning_rate": 3.8061981945889726e-05,
      "loss": 0.0,
      "step": 858
    },
    {
      "epoch": 0.33255903987611307,
      "grad_norm": 0.03951540216803551,
      "learning_rate": 3.795887396061995e-05,
      "loss": 0.0004,
      "step": 859
    },
    {
      "epoch": 0.3329461866047232,
      "grad_norm": 0.0011091261403635144,
      "learning_rate": 3.7855820304178204e-05,
      "loss": 0.0,
      "step": 860
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.00019632387557066977,
      "learning_rate": 3.775282144153627e-05,
      "loss": 0.0,
      "step": 861
    },
    {
      "epoch": 0.3337204800619435,
      "grad_norm": 0.015141610987484455,
      "learning_rate": 3.764987783741879e-05,
      "loss": 0.0001,
      "step": 862
    },
    {
      "epoch": 0.3341076267905536,
      "grad_norm": 0.0024289898574352264,
      "learning_rate": 3.754698995630101e-05,
      "loss": 0.0,
      "step": 863
    },
    {
      "epoch": 0.3344947735191638,
      "grad_norm": 0.030151624232530594,
      "learning_rate": 3.7444158262406824e-05,
      "loss": 0.0001,
      "step": 864
    },
    {
      "epoch": 0.3348819202477739,
      "grad_norm": 0.00045968167250975966,
      "learning_rate": 3.734138321970653e-05,
      "loss": 0.0,
      "step": 865
    },
    {
      "epoch": 0.33526906697638403,
      "grad_norm": 0.0013369093649089336,
      "learning_rate": 3.723866529191491e-05,
      "loss": 0.0,
      "step": 866
    },
    {
      "epoch": 0.3356562137049942,
      "grad_norm": 0.00015746480494271964,
      "learning_rate": 3.7136004942489e-05,
      "loss": 0.0,
      "step": 867
    },
    {
      "epoch": 0.33604336043360433,
      "grad_norm": 0.0004616810765583068,
      "learning_rate": 3.703340263462602e-05,
      "loss": 0.0,
      "step": 868
    },
    {
      "epoch": 0.33643050716221445,
      "grad_norm": 0.000538064050488174,
      "learning_rate": 3.693085883126137e-05,
      "loss": 0.0,
      "step": 869
    },
    {
      "epoch": 0.33681765389082463,
      "grad_norm": 0.0004215772496536374,
      "learning_rate": 3.6828373995066436e-05,
      "loss": 0.0,
      "step": 870
    },
    {
      "epoch": 0.33720480061943475,
      "grad_norm": 0.00042134601972065866,
      "learning_rate": 3.672594858844658e-05,
      "loss": 0.0,
      "step": 871
    },
    {
      "epoch": 0.33759194734804493,
      "grad_norm": 0.004986919928342104,
      "learning_rate": 3.6623583073538966e-05,
      "loss": 0.0001,
      "step": 872
    },
    {
      "epoch": 0.33797909407665505,
      "grad_norm": 0.0002799661597236991,
      "learning_rate": 3.65212779122106e-05,
      "loss": 0.0,
      "step": 873
    },
    {
      "epoch": 0.3383662408052652,
      "grad_norm": 0.0004925346001982689,
      "learning_rate": 3.641903356605614e-05,
      "loss": 0.0,
      "step": 874
    },
    {
      "epoch": 0.33875338753387535,
      "grad_norm": 0.001601546653546393,
      "learning_rate": 3.631685049639586e-05,
      "loss": 0.0,
      "step": 875
    },
    {
      "epoch": 0.33914053426248547,
      "grad_norm": 0.00030690181301906705,
      "learning_rate": 3.621472916427354e-05,
      "loss": 0.0,
      "step": 876
    },
    {
      "epoch": 0.33952768099109565,
      "grad_norm": 0.00021768384613096714,
      "learning_rate": 3.611267003045443e-05,
      "loss": 0.0,
      "step": 877
    },
    {
      "epoch": 0.33991482771970577,
      "grad_norm": 0.020688867196440697,
      "learning_rate": 3.601067355542312e-05,
      "loss": 0.0002,
      "step": 878
    },
    {
      "epoch": 0.3403019744483159,
      "grad_norm": 0.0002073254290735349,
      "learning_rate": 3.59087401993815e-05,
      "loss": 0.0,
      "step": 879
    },
    {
      "epoch": 0.34068912117692607,
      "grad_norm": 0.0022969204001128674,
      "learning_rate": 3.580687042224667e-05,
      "loss": 0.0,
      "step": 880
    },
    {
      "epoch": 0.3410762679055362,
      "grad_norm": 0.000548949814401567,
      "learning_rate": 3.5705064683648856e-05,
      "loss": 0.0,
      "step": 881
    },
    {
      "epoch": 0.34146341463414637,
      "grad_norm": 0.0054088798351585865,
      "learning_rate": 3.560332344292938e-05,
      "loss": 0.0,
      "step": 882
    },
    {
      "epoch": 0.3418505613627565,
      "grad_norm": 0.00040069175884127617,
      "learning_rate": 3.550164715913853e-05,
      "loss": 0.0,
      "step": 883
    },
    {
      "epoch": 0.3422377080913666,
      "grad_norm": 0.0004740517179016024,
      "learning_rate": 3.540003629103349e-05,
      "loss": 0.0,
      "step": 884
    },
    {
      "epoch": 0.3426248548199768,
      "grad_norm": 0.0020350434351712465,
      "learning_rate": 3.529849129707633e-05,
      "loss": 0.0,
      "step": 885
    },
    {
      "epoch": 0.3430120015485869,
      "grad_norm": 0.0003776460944209248,
      "learning_rate": 3.519701263543187e-05,
      "loss": 0.0,
      "step": 886
    },
    {
      "epoch": 0.34339914827719703,
      "grad_norm": 0.001164233428426087,
      "learning_rate": 3.5095600763965676e-05,
      "loss": 0.0,
      "step": 887
    },
    {
      "epoch": 0.3437862950058072,
      "grad_norm": 0.0004077065095771104,
      "learning_rate": 3.499425614024191e-05,
      "loss": 0.0,
      "step": 888
    },
    {
      "epoch": 0.34417344173441733,
      "grad_norm": 0.002573886886239052,
      "learning_rate": 3.489297922152136e-05,
      "loss": 0.0,
      "step": 889
    },
    {
      "epoch": 0.3445605884630275,
      "grad_norm": 0.02411019243299961,
      "learning_rate": 3.479177046475935e-05,
      "loss": 0.0002,
      "step": 890
    },
    {
      "epoch": 0.34494773519163763,
      "grad_norm": 0.001541433739475906,
      "learning_rate": 3.4690630326603604e-05,
      "loss": 0.0,
      "step": 891
    },
    {
      "epoch": 0.34533488192024775,
      "grad_norm": 0.05096811428666115,
      "learning_rate": 3.458955926339228e-05,
      "loss": 0.0003,
      "step": 892
    },
    {
      "epoch": 0.3457220286488579,
      "grad_norm": 0.0001714549434836954,
      "learning_rate": 3.448855773115185e-05,
      "loss": 0.0,
      "step": 893
    },
    {
      "epoch": 0.34610917537746805,
      "grad_norm": 0.0007654880173504353,
      "learning_rate": 3.438762618559511e-05,
      "loss": 0.0,
      "step": 894
    },
    {
      "epoch": 0.3464963221060782,
      "grad_norm": 0.34326171875,
      "learning_rate": 3.428676508211902e-05,
      "loss": 0.0018,
      "step": 895
    },
    {
      "epoch": 0.34688346883468835,
      "grad_norm": 0.003775347489863634,
      "learning_rate": 3.418597487580277e-05,
      "loss": 0.0001,
      "step": 896
    },
    {
      "epoch": 0.34727061556329847,
      "grad_norm": 0.0011271066032350063,
      "learning_rate": 3.408525602140561e-05,
      "loss": 0.0,
      "step": 897
    },
    {
      "epoch": 0.34765776229190865,
      "grad_norm": 0.00141782162245363,
      "learning_rate": 3.3984608973364904e-05,
      "loss": 0.0,
      "step": 898
    },
    {
      "epoch": 0.34804490902051877,
      "grad_norm": 0.006298588123172522,
      "learning_rate": 3.388403418579401e-05,
      "loss": 0.0001,
      "step": 899
    },
    {
      "epoch": 0.34843205574912894,
      "grad_norm": 0.057084690779447556,
      "learning_rate": 3.378353211248024e-05,
      "loss": 0.0006,
      "step": 900
    },
    {
      "epoch": 0.34881920247773907,
      "grad_norm": 0.03301633149385452,
      "learning_rate": 3.368310320688284e-05,
      "loss": 0.0017,
      "step": 901
    },
    {
      "epoch": 0.3492063492063492,
      "grad_norm": 0.03648944944143295,
      "learning_rate": 3.358274792213091e-05,
      "loss": 0.001,
      "step": 902
    },
    {
      "epoch": 0.34959349593495936,
      "grad_norm": 0.045417048037052155,
      "learning_rate": 3.34824667110214e-05,
      "loss": 0.0002,
      "step": 903
    },
    {
      "epoch": 0.3499806426635695,
      "grad_norm": 0.0014944113790988922,
      "learning_rate": 3.338226002601703e-05,
      "loss": 0.0,
      "step": 904
    },
    {
      "epoch": 0.35036778939217966,
      "grad_norm": 0.0017919113161042333,
      "learning_rate": 3.328212831924424e-05,
      "loss": 0.0,
      "step": 905
    },
    {
      "epoch": 0.3507549361207898,
      "grad_norm": 0.0006082098698243499,
      "learning_rate": 3.3182072042491244e-05,
      "loss": 0.0,
      "step": 906
    },
    {
      "epoch": 0.3511420828493999,
      "grad_norm": 0.010628577321767807,
      "learning_rate": 3.308209164720584e-05,
      "loss": 0.0001,
      "step": 907
    },
    {
      "epoch": 0.3515292295780101,
      "grad_norm": 0.08203943073749542,
      "learning_rate": 3.2982187584493515e-05,
      "loss": 0.0019,
      "step": 908
    },
    {
      "epoch": 0.3519163763066202,
      "grad_norm": 0.0009694083128124475,
      "learning_rate": 3.2882360305115294e-05,
      "loss": 0.0,
      "step": 909
    },
    {
      "epoch": 0.3523035230352303,
      "grad_norm": 0.00442282622680068,
      "learning_rate": 3.2782610259485813e-05,
      "loss": 0.0001,
      "step": 910
    },
    {
      "epoch": 0.3526906697638405,
      "grad_norm": 0.08702477067708969,
      "learning_rate": 3.268293789767118e-05,
      "loss": 0.0014,
      "step": 911
    },
    {
      "epoch": 0.3530778164924506,
      "grad_norm": 0.02207796275615692,
      "learning_rate": 3.2583343669387046e-05,
      "loss": 0.0001,
      "step": 912
    },
    {
      "epoch": 0.3534649632210608,
      "grad_norm": 0.0006780868279747665,
      "learning_rate": 3.248382802399649e-05,
      "loss": 0.0,
      "step": 913
    },
    {
      "epoch": 0.3538521099496709,
      "grad_norm": 0.02841506339609623,
      "learning_rate": 3.238439141050807e-05,
      "loss": 0.0003,
      "step": 914
    },
    {
      "epoch": 0.35423925667828104,
      "grad_norm": 0.00042213572305627167,
      "learning_rate": 3.228503427757374e-05,
      "loss": 0.0,
      "step": 915
    },
    {
      "epoch": 0.3546264034068912,
      "grad_norm": 0.0002454633649904281,
      "learning_rate": 3.218575707348681e-05,
      "loss": 0.0,
      "step": 916
    },
    {
      "epoch": 0.35501355013550134,
      "grad_norm": 0.0027822977863252163,
      "learning_rate": 3.208656024618002e-05,
      "loss": 0.0,
      "step": 917
    },
    {
      "epoch": 0.3554006968641115,
      "grad_norm": 0.020471172407269478,
      "learning_rate": 3.198744424322338e-05,
      "loss": 0.0004,
      "step": 918
    },
    {
      "epoch": 0.35578784359272164,
      "grad_norm": 0.0007699733250774443,
      "learning_rate": 3.188840951182229e-05,
      "loss": 0.0,
      "step": 919
    },
    {
      "epoch": 0.35617499032133176,
      "grad_norm": 0.00015266439004335552,
      "learning_rate": 3.178945649881543e-05,
      "loss": 0.0,
      "step": 920
    },
    {
      "epoch": 0.35656213704994194,
      "grad_norm": 0.00025495782028883696,
      "learning_rate": 3.169058565067274e-05,
      "loss": 0.0,
      "step": 921
    },
    {
      "epoch": 0.35694928377855206,
      "grad_norm": 8.566460019210353e-05,
      "learning_rate": 3.159179741349354e-05,
      "loss": 0.0,
      "step": 922
    },
    {
      "epoch": 0.35733643050716224,
      "grad_norm": 0.00040574869490228593,
      "learning_rate": 3.149309223300428e-05,
      "loss": 0.0,
      "step": 923
    },
    {
      "epoch": 0.35772357723577236,
      "grad_norm": 0.00045708048855885863,
      "learning_rate": 3.1394470554556764e-05,
      "loss": 0.0,
      "step": 924
    },
    {
      "epoch": 0.3581107239643825,
      "grad_norm": 0.0041862414218485355,
      "learning_rate": 3.1295932823125985e-05,
      "loss": 0.0,
      "step": 925
    },
    {
      "epoch": 0.35849787069299266,
      "grad_norm": 0.000689912005327642,
      "learning_rate": 3.1197479483308214e-05,
      "loss": 0.0,
      "step": 926
    },
    {
      "epoch": 0.3588850174216028,
      "grad_norm": 0.003948573488742113,
      "learning_rate": 3.109911097931891e-05,
      "loss": 0.0,
      "step": 927
    },
    {
      "epoch": 0.35927216415021296,
      "grad_norm": 0.00029136036755517125,
      "learning_rate": 3.1000827754990766e-05,
      "loss": 0.0,
      "step": 928
    },
    {
      "epoch": 0.3596593108788231,
      "grad_norm": 0.0020344445947557688,
      "learning_rate": 3.090263025377173e-05,
      "loss": 0.0,
      "step": 929
    },
    {
      "epoch": 0.3600464576074332,
      "grad_norm": 0.0002035750512732193,
      "learning_rate": 3.080451891872295e-05,
      "loss": 0.0,
      "step": 930
    },
    {
      "epoch": 0.3604336043360434,
      "grad_norm": 0.0010056099854409695,
      "learning_rate": 3.070649419251681e-05,
      "loss": 0.0,
      "step": 931
    },
    {
      "epoch": 0.3608207510646535,
      "grad_norm": 0.000317671918310225,
      "learning_rate": 3.06085565174349e-05,
      "loss": 0.0,
      "step": 932
    },
    {
      "epoch": 0.3612078977932636,
      "grad_norm": 0.0009133410057984293,
      "learning_rate": 3.0510706335366035e-05,
      "loss": 0.0,
      "step": 933
    },
    {
      "epoch": 0.3615950445218738,
      "grad_norm": 0.06602532416582108,
      "learning_rate": 3.041294408780431e-05,
      "loss": 0.0003,
      "step": 934
    },
    {
      "epoch": 0.3619821912504839,
      "grad_norm": 0.00024213208234868944,
      "learning_rate": 3.0315270215847013e-05,
      "loss": 0.0,
      "step": 935
    },
    {
      "epoch": 0.3623693379790941,
      "grad_norm": 0.0014147679321467876,
      "learning_rate": 3.0217685160192734e-05,
      "loss": 0.0,
      "step": 936
    },
    {
      "epoch": 0.3627564847077042,
      "grad_norm": 0.0003020058502443135,
      "learning_rate": 3.0120189361139273e-05,
      "loss": 0.0,
      "step": 937
    },
    {
      "epoch": 0.36314363143631434,
      "grad_norm": 0.003151725744828582,
      "learning_rate": 3.002278325858177e-05,
      "loss": 0.0,
      "step": 938
    },
    {
      "epoch": 0.3635307781649245,
      "grad_norm": 0.002396926749497652,
      "learning_rate": 2.9925467292010644e-05,
      "loss": 0.0,
      "step": 939
    },
    {
      "epoch": 0.36391792489353464,
      "grad_norm": 0.0011632463429123163,
      "learning_rate": 2.982824190050958e-05,
      "loss": 0.0,
      "step": 940
    },
    {
      "epoch": 0.3643050716221448,
      "grad_norm": 0.2228887379169464,
      "learning_rate": 2.973110752275366e-05,
      "loss": 0.001,
      "step": 941
    },
    {
      "epoch": 0.36469221835075494,
      "grad_norm": 0.0011429457226768136,
      "learning_rate": 2.963406459700726e-05,
      "loss": 0.0,
      "step": 942
    },
    {
      "epoch": 0.36507936507936506,
      "grad_norm": 0.00012259121285751462,
      "learning_rate": 2.9537113561122178e-05,
      "loss": 0.0,
      "step": 943
    },
    {
      "epoch": 0.36546651180797524,
      "grad_norm": 0.000644800195004791,
      "learning_rate": 2.944025485253557e-05,
      "loss": 0.0,
      "step": 944
    },
    {
      "epoch": 0.36585365853658536,
      "grad_norm": 0.00017125460726674646,
      "learning_rate": 2.9343488908268034e-05,
      "loss": 0.0,
      "step": 945
    },
    {
      "epoch": 0.36624080526519553,
      "grad_norm": 0.00020638681598939002,
      "learning_rate": 2.9246816164921655e-05,
      "loss": 0.0,
      "step": 946
    },
    {
      "epoch": 0.36662795199380566,
      "grad_norm": 0.0005838845972903073,
      "learning_rate": 2.915023705867794e-05,
      "loss": 0.0,
      "step": 947
    },
    {
      "epoch": 0.3670150987224158,
      "grad_norm": 0.000993402791209519,
      "learning_rate": 2.905375202529594e-05,
      "loss": 0.0,
      "step": 948
    },
    {
      "epoch": 0.36740224545102595,
      "grad_norm": 0.00029620109125971794,
      "learning_rate": 2.895736150011026e-05,
      "loss": 0.0,
      "step": 949
    },
    {
      "epoch": 0.3677893921796361,
      "grad_norm": 2.658259153366089,
      "learning_rate": 2.8861065918029085e-05,
      "loss": 0.0567,
      "step": 950
    },
    {
      "epoch": 0.3681765389082462,
      "grad_norm": 0.02834871970117092,
      "learning_rate": 2.8764865713532234e-05,
      "loss": 0.0021,
      "step": 951
    },
    {
      "epoch": 0.3685636856368564,
      "grad_norm": 0.03441345691680908,
      "learning_rate": 2.8668761320669156e-05,
      "loss": 0.0008,
      "step": 952
    },
    {
      "epoch": 0.3689508323654665,
      "grad_norm": 0.07143665105104446,
      "learning_rate": 2.8572753173057003e-05,
      "loss": 0.0004,
      "step": 953
    },
    {
      "epoch": 0.3693379790940767,
      "grad_norm": 0.06644494086503983,
      "learning_rate": 2.8476841703878772e-05,
      "loss": 0.0007,
      "step": 954
    },
    {
      "epoch": 0.3697251258226868,
      "grad_norm": 0.005595757160335779,
      "learning_rate": 2.8381027345881102e-05,
      "loss": 0.0001,
      "step": 955
    },
    {
      "epoch": 0.3701122725512969,
      "grad_norm": 0.13141833245754242,
      "learning_rate": 2.8285310531372573e-05,
      "loss": 0.0014,
      "step": 956
    },
    {
      "epoch": 0.3704994192799071,
      "grad_norm": 0.007661540526896715,
      "learning_rate": 2.8189691692221627e-05,
      "loss": 0.0,
      "step": 957
    },
    {
      "epoch": 0.3708865660085172,
      "grad_norm": 0.0015809281030669808,
      "learning_rate": 2.809417125985467e-05,
      "loss": 0.0001,
      "step": 958
    },
    {
      "epoch": 0.3712737127371274,
      "grad_norm": 0.007457795087248087,
      "learning_rate": 2.7998749665254033e-05,
      "loss": 0.0001,
      "step": 959
    },
    {
      "epoch": 0.3716608594657375,
      "grad_norm": 0.0005789248389191926,
      "learning_rate": 2.7903427338956177e-05,
      "loss": 0.0,
      "step": 960
    },
    {
      "epoch": 0.37204800619434764,
      "grad_norm": 0.00037478163721971214,
      "learning_rate": 2.7808204711049624e-05,
      "loss": 0.0,
      "step": 961
    },
    {
      "epoch": 0.3724351529229578,
      "grad_norm": 0.003515935968607664,
      "learning_rate": 2.771308221117309e-05,
      "loss": 0.0,
      "step": 962
    },
    {
      "epoch": 0.37282229965156793,
      "grad_norm": 0.0019339972641319036,
      "learning_rate": 2.7618060268513503e-05,
      "loss": 0.0,
      "step": 963
    },
    {
      "epoch": 0.3732094463801781,
      "grad_norm": 0.00024095486151054502,
      "learning_rate": 2.7523139311804074e-05,
      "loss": 0.0,
      "step": 964
    },
    {
      "epoch": 0.37359659310878823,
      "grad_norm": 0.0007112721796147525,
      "learning_rate": 2.742831976932242e-05,
      "loss": 0.0,
      "step": 965
    },
    {
      "epoch": 0.37398373983739835,
      "grad_norm": 0.0007284593302756548,
      "learning_rate": 2.7333602068888496e-05,
      "loss": 0.0,
      "step": 966
    },
    {
      "epoch": 0.37437088656600853,
      "grad_norm": 0.006084626540541649,
      "learning_rate": 2.7238986637862833e-05,
      "loss": 0.0,
      "step": 967
    },
    {
      "epoch": 0.37475803329461865,
      "grad_norm": 0.000507039949297905,
      "learning_rate": 2.7144473903144496e-05,
      "loss": 0.0,
      "step": 968
    },
    {
      "epoch": 0.37514518002322883,
      "grad_norm": 0.0019885883666574955,
      "learning_rate": 2.7050064291169187e-05,
      "loss": 0.0,
      "step": 969
    },
    {
      "epoch": 0.37553232675183895,
      "grad_norm": 0.04670470952987671,
      "learning_rate": 2.6955758227907335e-05,
      "loss": 0.0002,
      "step": 970
    },
    {
      "epoch": 0.3759194734804491,
      "grad_norm": 0.00021539766748901457,
      "learning_rate": 2.6861556138862155e-05,
      "loss": 0.0,
      "step": 971
    },
    {
      "epoch": 0.37630662020905925,
      "grad_norm": 0.0003146221279166639,
      "learning_rate": 2.676745844906776e-05,
      "loss": 0.0,
      "step": 972
    },
    {
      "epoch": 0.37669376693766937,
      "grad_norm": 0.0013595115160569549,
      "learning_rate": 2.6673465583087176e-05,
      "loss": 0.0,
      "step": 973
    },
    {
      "epoch": 0.3770809136662795,
      "grad_norm": 0.004096793942153454,
      "learning_rate": 2.65795779650105e-05,
      "loss": 0.0,
      "step": 974
    },
    {
      "epoch": 0.37746806039488967,
      "grad_norm": 0.0003931306710001081,
      "learning_rate": 2.648579601845295e-05,
      "loss": 0.0,
      "step": 975
    },
    {
      "epoch": 0.3778552071234998,
      "grad_norm": 0.00021391961490735412,
      "learning_rate": 2.6392120166552992e-05,
      "loss": 0.0,
      "step": 976
    },
    {
      "epoch": 0.37824235385210997,
      "grad_norm": 0.00013036343443673104,
      "learning_rate": 2.629855083197031e-05,
      "loss": 0.0,
      "step": 977
    },
    {
      "epoch": 0.3786295005807201,
      "grad_norm": 0.00223020208068192,
      "learning_rate": 2.6205088436884095e-05,
      "loss": 0.0,
      "step": 978
    },
    {
      "epoch": 0.3790166473093302,
      "grad_norm": 0.0004039799969177693,
      "learning_rate": 2.6111733402991e-05,
      "loss": 0.0,
      "step": 979
    },
    {
      "epoch": 0.3794037940379404,
      "grad_norm": 0.004602463450282812,
      "learning_rate": 2.6018486151503213e-05,
      "loss": 0.0001,
      "step": 980
    },
    {
      "epoch": 0.3797909407665505,
      "grad_norm": 0.00020700599998235703,
      "learning_rate": 2.592534710314668e-05,
      "loss": 0.0,
      "step": 981
    },
    {
      "epoch": 0.3801780874951607,
      "grad_norm": 0.0005082901916466653,
      "learning_rate": 2.5832316678159118e-05,
      "loss": 0.0,
      "step": 982
    },
    {
      "epoch": 0.3805652342237708,
      "grad_norm": 0.00023574243823532015,
      "learning_rate": 2.573939529628816e-05,
      "loss": 0.0,
      "step": 983
    },
    {
      "epoch": 0.38095238095238093,
      "grad_norm": 0.00044681079452857375,
      "learning_rate": 2.564658337678939e-05,
      "loss": 0.0,
      "step": 984
    },
    {
      "epoch": 0.3813395276809911,
      "grad_norm": 0.00034449860686436296,
      "learning_rate": 2.555388133842455e-05,
      "loss": 0.0,
      "step": 985
    },
    {
      "epoch": 0.38172667440960123,
      "grad_norm": 0.00018269941210746765,
      "learning_rate": 2.5461289599459646e-05,
      "loss": 0.0,
      "step": 986
    },
    {
      "epoch": 0.3821138211382114,
      "grad_norm": 0.00023150882043410093,
      "learning_rate": 2.5368808577662913e-05,
      "loss": 0.0,
      "step": 987
    },
    {
      "epoch": 0.38250096786682153,
      "grad_norm": 0.00035011590807698667,
      "learning_rate": 2.5276438690303112e-05,
      "loss": 0.0,
      "step": 988
    },
    {
      "epoch": 0.38288811459543165,
      "grad_norm": 0.0003966555814258754,
      "learning_rate": 2.518418035414756e-05,
      "loss": 0.0,
      "step": 989
    },
    {
      "epoch": 0.3832752613240418,
      "grad_norm": 0.002989868400618434,
      "learning_rate": 2.5092033985460265e-05,
      "loss": 0.0,
      "step": 990
    },
    {
      "epoch": 0.38366240805265195,
      "grad_norm": 0.0004271006619092077,
      "learning_rate": 2.500000000000001e-05,
      "loss": 0.0,
      "step": 991
    },
    {
      "epoch": 0.3840495547812621,
      "grad_norm": 0.0008473137859255075,
      "learning_rate": 2.490807881301855e-05,
      "loss": 0.0,
      "step": 992
    },
    {
      "epoch": 0.38443670150987225,
      "grad_norm": 1.8476226329803467,
      "learning_rate": 2.481627083925869e-05,
      "loss": 0.0797,
      "step": 993
    },
    {
      "epoch": 0.38482384823848237,
      "grad_norm": 0.00014391196600627154,
      "learning_rate": 2.472457649295241e-05,
      "loss": 0.0,
      "step": 994
    },
    {
      "epoch": 0.38521099496709255,
      "grad_norm": 0.0013244638685137033,
      "learning_rate": 2.4632996187819034e-05,
      "loss": 0.0,
      "step": 995
    },
    {
      "epoch": 0.38559814169570267,
      "grad_norm": 0.006371075287461281,
      "learning_rate": 2.4541530337063325e-05,
      "loss": 0.0001,
      "step": 996
    },
    {
      "epoch": 0.3859852884243128,
      "grad_norm": 0.0009385327575728297,
      "learning_rate": 2.445017935337365e-05,
      "loss": 0.0,
      "step": 997
    },
    {
      "epoch": 0.38637243515292297,
      "grad_norm": 0.01769573800265789,
      "learning_rate": 2.435894364892005e-05,
      "loss": 0.0001,
      "step": 998
    },
    {
      "epoch": 0.3867595818815331,
      "grad_norm": 0.003815864212810993,
      "learning_rate": 2.42678236353525e-05,
      "loss": 0.0001,
      "step": 999
    },
    {
      "epoch": 0.38714672861014326,
      "grad_norm": 0.020971521735191345,
      "learning_rate": 2.4176819723798948e-05,
      "loss": 0.0001,
      "step": 1000
    },
    {
      "epoch": 0.3875338753387534,
      "grad_norm": 0.0321972593665123,
      "learning_rate": 2.408593232486351e-05,
      "loss": 0.003,
      "step": 1001
    },
    {
      "epoch": 0.3879210220673635,
      "grad_norm": 0.0008187716011889279,
      "learning_rate": 2.3995161848624597e-05,
      "loss": 0.0,
      "step": 1002
    },
    {
      "epoch": 0.3883081687959737,
      "grad_norm": 0.020097631961107254,
      "learning_rate": 2.390450870463307e-05,
      "loss": 0.0004,
      "step": 1003
    },
    {
      "epoch": 0.3886953155245838,
      "grad_norm": 0.0003399544220883399,
      "learning_rate": 2.381397330191043e-05,
      "loss": 0.0,
      "step": 1004
    },
    {
      "epoch": 0.389082462253194,
      "grad_norm": 0.048977453261613846,
      "learning_rate": 2.372355604894686e-05,
      "loss": 0.0025,
      "step": 1005
    },
    {
      "epoch": 0.3894696089818041,
      "grad_norm": 0.0018951277015730739,
      "learning_rate": 2.3633257353699524e-05,
      "loss": 0.0,
      "step": 1006
    },
    {
      "epoch": 0.3898567557104142,
      "grad_norm": 0.0007534907781518996,
      "learning_rate": 2.354307762359064e-05,
      "loss": 0.0,
      "step": 1007
    },
    {
      "epoch": 0.3902439024390244,
      "grad_norm": 0.00365415564738214,
      "learning_rate": 2.3453017265505673e-05,
      "loss": 0.0001,
      "step": 1008
    },
    {
      "epoch": 0.3906310491676345,
      "grad_norm": 0.07262806594371796,
      "learning_rate": 2.3363076685791435e-05,
      "loss": 0.0021,
      "step": 1009
    },
    {
      "epoch": 0.3910181958962447,
      "grad_norm": 0.0027157345321029425,
      "learning_rate": 2.3273256290254402e-05,
      "loss": 0.0,
      "step": 1010
    },
    {
      "epoch": 0.3914053426248548,
      "grad_norm": 0.002876731101423502,
      "learning_rate": 2.3183556484158736e-05,
      "loss": 0.0,
      "step": 1011
    },
    {
      "epoch": 0.39179248935346495,
      "grad_norm": 0.0003100191242992878,
      "learning_rate": 2.309397767222446e-05,
      "loss": 0.0,
      "step": 1012
    },
    {
      "epoch": 0.3921796360820751,
      "grad_norm": 0.0007354211411438882,
      "learning_rate": 2.300452025862574e-05,
      "loss": 0.0,
      "step": 1013
    },
    {
      "epoch": 0.39256678281068524,
      "grad_norm": 0.16168968379497528,
      "learning_rate": 2.291518464698899e-05,
      "loss": 0.0013,
      "step": 1014
    },
    {
      "epoch": 0.39295392953929537,
      "grad_norm": 0.0008131096255965531,
      "learning_rate": 2.282597124039107e-05,
      "loss": 0.0,
      "step": 1015
    },
    {
      "epoch": 0.39334107626790554,
      "grad_norm": 0.0007273323717527092,
      "learning_rate": 2.27368804413574e-05,
      "loss": 0.0,
      "step": 1016
    },
    {
      "epoch": 0.39372822299651566,
      "grad_norm": 0.0019192282343283296,
      "learning_rate": 2.2647912651860252e-05,
      "loss": 0.0,
      "step": 1017
    },
    {
      "epoch": 0.39411536972512584,
      "grad_norm": 0.0010810550302267075,
      "learning_rate": 2.255906827331693e-05,
      "loss": 0.0,
      "step": 1018
    },
    {
      "epoch": 0.39450251645373596,
      "grad_norm": 0.0015484696486964822,
      "learning_rate": 2.2470347706587813e-05,
      "loss": 0.0,
      "step": 1019
    },
    {
      "epoch": 0.3948896631823461,
      "grad_norm": 0.0008098200778476894,
      "learning_rate": 2.238175135197471e-05,
      "loss": 0.0,
      "step": 1020
    },
    {
      "epoch": 0.39527680991095626,
      "grad_norm": 0.0020570175256580114,
      "learning_rate": 2.2293279609219003e-05,
      "loss": 0.0,
      "step": 1021
    },
    {
      "epoch": 0.3956639566395664,
      "grad_norm": 0.0012723479885607958,
      "learning_rate": 2.220493287749978e-05,
      "loss": 0.0,
      "step": 1022
    },
    {
      "epoch": 0.39605110336817656,
      "grad_norm": 0.0003519760794006288,
      "learning_rate": 2.2116711555432136e-05,
      "loss": 0.0,
      "step": 1023
    },
    {
      "epoch": 0.3964382500967867,
      "grad_norm": 0.00034122299985028803,
      "learning_rate": 2.2028616041065304e-05,
      "loss": 0.0,
      "step": 1024
    },
    {
      "epoch": 0.3968253968253968,
      "grad_norm": 0.026362977921962738,
      "learning_rate": 2.194064673188089e-05,
      "loss": 0.0002,
      "step": 1025
    },
    {
      "epoch": 0.397212543554007,
      "grad_norm": 0.00036965496838092804,
      "learning_rate": 2.185280402479105e-05,
      "loss": 0.0,
      "step": 1026
    },
    {
      "epoch": 0.3975996902826171,
      "grad_norm": 0.11109102517366409,
      "learning_rate": 2.1765088316136755e-05,
      "loss": 0.001,
      "step": 1027
    },
    {
      "epoch": 0.3979868370112273,
      "grad_norm": 0.042729608714580536,
      "learning_rate": 2.1677500001685945e-05,
      "loss": 0.0004,
      "step": 1028
    },
    {
      "epoch": 0.3983739837398374,
      "grad_norm": 0.00787183828651905,
      "learning_rate": 2.159003947663174e-05,
      "loss": 0.0001,
      "step": 1029
    },
    {
      "epoch": 0.3987611304684475,
      "grad_norm": 0.003600550815463066,
      "learning_rate": 2.1502707135590723e-05,
      "loss": 0.0,
      "step": 1030
    },
    {
      "epoch": 0.3991482771970577,
      "grad_norm": 0.007428227458149195,
      "learning_rate": 2.1415503372601096e-05,
      "loss": 0.0001,
      "step": 1031
    },
    {
      "epoch": 0.3995354239256678,
      "grad_norm": 0.0009940440068021417,
      "learning_rate": 2.1328428581120964e-05,
      "loss": 0.0,
      "step": 1032
    },
    {
      "epoch": 0.399922570654278,
      "grad_norm": 0.00677646417170763,
      "learning_rate": 2.124148315402642e-05,
      "loss": 0.0001,
      "step": 1033
    },
    {
      "epoch": 0.4003097173828881,
      "grad_norm": 0.0008721842896193266,
      "learning_rate": 2.1154667483609998e-05,
      "loss": 0.0,
      "step": 1034
    },
    {
      "epoch": 0.40069686411149824,
      "grad_norm": 0.00037974066799506545,
      "learning_rate": 2.106798196157872e-05,
      "loss": 0.0,
      "step": 1035
    },
    {
      "epoch": 0.4010840108401084,
      "grad_norm": 0.0008370023570023477,
      "learning_rate": 2.098142697905236e-05,
      "loss": 0.0,
      "step": 1036
    },
    {
      "epoch": 0.40147115756871854,
      "grad_norm": 0.0004361419996712357,
      "learning_rate": 2.0895002926561736e-05,
      "loss": 0.0,
      "step": 1037
    },
    {
      "epoch": 0.40185830429732866,
      "grad_norm": 0.0016112083103507757,
      "learning_rate": 2.080871019404692e-05,
      "loss": 0.0,
      "step": 1038
    },
    {
      "epoch": 0.40224545102593884,
      "grad_norm": 0.0007014021975919604,
      "learning_rate": 2.0722549170855494e-05,
      "loss": 0.0,
      "step": 1039
    },
    {
      "epoch": 0.40263259775454896,
      "grad_norm": 0.00015222003275994211,
      "learning_rate": 2.063652024574071e-05,
      "loss": 0.0,
      "step": 1040
    },
    {
      "epoch": 0.40301974448315914,
      "grad_norm": 0.00020387036784086376,
      "learning_rate": 2.0550623806859844e-05,
      "loss": 0.0,
      "step": 1041
    },
    {
      "epoch": 0.40340689121176926,
      "grad_norm": 0.004351683426648378,
      "learning_rate": 2.0464860241772455e-05,
      "loss": 0.0,
      "step": 1042
    },
    {
      "epoch": 0.4037940379403794,
      "grad_norm": 0.00455818697810173,
      "learning_rate": 2.0379229937438477e-05,
      "loss": 0.0001,
      "step": 1043
    },
    {
      "epoch": 0.40418118466898956,
      "grad_norm": 0.000623556668870151,
      "learning_rate": 2.0293733280216648e-05,
      "loss": 0.0,
      "step": 1044
    },
    {
      "epoch": 0.4045683313975997,
      "grad_norm": 0.0013906165258958936,
      "learning_rate": 2.020837065586269e-05,
      "loss": 0.0,
      "step": 1045
    },
    {
      "epoch": 0.40495547812620986,
      "grad_norm": 0.003972301259636879,
      "learning_rate": 2.0123142449527584e-05,
      "loss": 0.0001,
      "step": 1046
    },
    {
      "epoch": 0.40534262485482,
      "grad_norm": 0.00038339104503393173,
      "learning_rate": 2.0038049045755775e-05,
      "loss": 0.0,
      "step": 1047
    },
    {
      "epoch": 0.4057297715834301,
      "grad_norm": 0.00043085686047561467,
      "learning_rate": 1.995309082848354e-05,
      "loss": 0.0,
      "step": 1048
    },
    {
      "epoch": 0.4061169183120403,
      "grad_norm": 0.0012404817389324307,
      "learning_rate": 1.9868268181037185e-05,
      "loss": 0.0,
      "step": 1049
    },
    {
      "epoch": 0.4065040650406504,
      "grad_norm": 0.0011212611570954323,
      "learning_rate": 1.978358148613134e-05,
      "loss": 0.0,
      "step": 1050
    },
    {
      "epoch": 0.4068912117692606,
      "grad_norm": 0.039180099964141846,
      "learning_rate": 1.9699031125867213e-05,
      "loss": 0.0023,
      "step": 1051
    },
    {
      "epoch": 0.4072783584978707,
      "grad_norm": 0.02465960942208767,
      "learning_rate": 1.9614617481730883e-05,
      "loss": 0.0004,
      "step": 1052
    },
    {
      "epoch": 0.4076655052264808,
      "grad_norm": 0.013811935670673847,
      "learning_rate": 1.9530340934591608e-05,
      "loss": 0.0003,
      "step": 1053
    },
    {
      "epoch": 0.408052651955091,
      "grad_norm": 0.008351047523319721,
      "learning_rate": 1.9446201864699987e-05,
      "loss": 0.0002,
      "step": 1054
    },
    {
      "epoch": 0.4084397986837011,
      "grad_norm": 0.0017275417922064662,
      "learning_rate": 1.9362200651686407e-05,
      "loss": 0.0001,
      "step": 1055
    },
    {
      "epoch": 0.4088269454123113,
      "grad_norm": 0.031212275847792625,
      "learning_rate": 1.9278337674559237e-05,
      "loss": 0.0005,
      "step": 1056
    },
    {
      "epoch": 0.4092140921409214,
      "grad_norm": 0.001833677408285439,
      "learning_rate": 1.919461331170312e-05,
      "loss": 0.0001,
      "step": 1057
    },
    {
      "epoch": 0.40960123886953154,
      "grad_norm": 0.00048409271403215826,
      "learning_rate": 1.9111027940877284e-05,
      "loss": 0.0,
      "step": 1058
    },
    {
      "epoch": 0.4099883855981417,
      "grad_norm": 0.004958099219948053,
      "learning_rate": 1.902758193921385e-05,
      "loss": 0.0001,
      "step": 1059
    },
    {
      "epoch": 0.41037553232675184,
      "grad_norm": 0.0003586974926292896,
      "learning_rate": 1.89442756832161e-05,
      "loss": 0.0,
      "step": 1060
    },
    {
      "epoch": 0.41076267905536196,
      "grad_norm": 0.004265904892235994,
      "learning_rate": 1.8861109548756766e-05,
      "loss": 0.0,
      "step": 1061
    },
    {
      "epoch": 0.41114982578397213,
      "grad_norm": 0.0007765475893393159,
      "learning_rate": 1.877808391107641e-05,
      "loss": 0.0,
      "step": 1062
    },
    {
      "epoch": 0.41153697251258226,
      "grad_norm": 0.003014490008354187,
      "learning_rate": 1.8695199144781645e-05,
      "loss": 0.0001,
      "step": 1063
    },
    {
      "epoch": 0.41192411924119243,
      "grad_norm": 0.0007445579976774752,
      "learning_rate": 1.861245562384351e-05,
      "loss": 0.0,
      "step": 1064
    },
    {
      "epoch": 0.41231126596980255,
      "grad_norm": 0.0004886957467533648,
      "learning_rate": 1.8529853721595697e-05,
      "loss": 0.0,
      "step": 1065
    },
    {
      "epoch": 0.4126984126984127,
      "grad_norm": 0.0007616803050041199,
      "learning_rate": 1.844739381073301e-05,
      "loss": 0.0,
      "step": 1066
    },
    {
      "epoch": 0.41308555942702285,
      "grad_norm": 0.0005557533586397767,
      "learning_rate": 1.8365076263309545e-05,
      "loss": 0.0,
      "step": 1067
    },
    {
      "epoch": 0.413472706155633,
      "grad_norm": 0.00033164405613206327,
      "learning_rate": 1.8282901450737033e-05,
      "loss": 0.0,
      "step": 1068
    },
    {
      "epoch": 0.41385985288424315,
      "grad_norm": 0.000697499024681747,
      "learning_rate": 1.8200869743783245e-05,
      "loss": 0.0,
      "step": 1069
    },
    {
      "epoch": 0.4142469996128533,
      "grad_norm": 0.008421828970313072,
      "learning_rate": 1.8118981512570253e-05,
      "loss": 0.0001,
      "step": 1070
    },
    {
      "epoch": 0.4146341463414634,
      "grad_norm": 0.0010353690013289452,
      "learning_rate": 1.80372371265728e-05,
      "loss": 0.0,
      "step": 1071
    },
    {
      "epoch": 0.41502129307007357,
      "grad_norm": 0.0004474584711715579,
      "learning_rate": 1.795563695461653e-05,
      "loss": 0.0,
      "step": 1072
    },
    {
      "epoch": 0.4154084397986837,
      "grad_norm": 0.0019867317751049995,
      "learning_rate": 1.7874181364876464e-05,
      "loss": 0.0,
      "step": 1073
    },
    {
      "epoch": 0.41579558652729387,
      "grad_norm": 0.001235696254298091,
      "learning_rate": 1.7792870724875314e-05,
      "loss": 0.0,
      "step": 1074
    },
    {
      "epoch": 0.416182733255904,
      "grad_norm": 0.0007534808246418834,
      "learning_rate": 1.77117054014817e-05,
      "loss": 0.0,
      "step": 1075
    },
    {
      "epoch": 0.4165698799845141,
      "grad_norm": 0.0005520595004782081,
      "learning_rate": 1.7630685760908622e-05,
      "loss": 0.0,
      "step": 1076
    },
    {
      "epoch": 0.4169570267131243,
      "grad_norm": 0.0005588960484601557,
      "learning_rate": 1.7549812168711777e-05,
      "loss": 0.0,
      "step": 1077
    },
    {
      "epoch": 0.4173441734417344,
      "grad_norm": 0.0007228697068057954,
      "learning_rate": 1.746908498978791e-05,
      "loss": 0.0,
      "step": 1078
    },
    {
      "epoch": 0.41773132017034453,
      "grad_norm": 0.00042682827915996313,
      "learning_rate": 1.73885045883731e-05,
      "loss": 0.0,
      "step": 1079
    },
    {
      "epoch": 0.4181184668989547,
      "grad_norm": 0.0011390242725610733,
      "learning_rate": 1.7308071328041243e-05,
      "loss": 0.0,
      "step": 1080
    },
    {
      "epoch": 0.41850561362756483,
      "grad_norm": 0.00039792185998521745,
      "learning_rate": 1.7227785571702303e-05,
      "loss": 0.0,
      "step": 1081
    },
    {
      "epoch": 0.418892760356175,
      "grad_norm": 0.0003800159611273557,
      "learning_rate": 1.7147647681600738e-05,
      "loss": 0.0,
      "step": 1082
    },
    {
      "epoch": 0.41927990708478513,
      "grad_norm": 0.0011553197400644422,
      "learning_rate": 1.7067658019313826e-05,
      "loss": 0.0,
      "step": 1083
    },
    {
      "epoch": 0.41966705381339525,
      "grad_norm": 0.0001602371339686215,
      "learning_rate": 1.6987816945750057e-05,
      "loss": 0.0,
      "step": 1084
    },
    {
      "epoch": 0.42005420054200543,
      "grad_norm": 0.0007585774874314666,
      "learning_rate": 1.690812482114752e-05,
      "loss": 0.0,
      "step": 1085
    },
    {
      "epoch": 0.42044134727061555,
      "grad_norm": 0.0005152882658876479,
      "learning_rate": 1.6828582005072187e-05,
      "loss": 0.0,
      "step": 1086
    },
    {
      "epoch": 0.42082849399922573,
      "grad_norm": 0.00029434141470119357,
      "learning_rate": 1.6749188856416425e-05,
      "loss": 0.0,
      "step": 1087
    },
    {
      "epoch": 0.42121564072783585,
      "grad_norm": 0.0014791539870202541,
      "learning_rate": 1.666994573339729e-05,
      "loss": 0.0,
      "step": 1088
    },
    {
      "epoch": 0.42160278745644597,
      "grad_norm": 0.0017479541711509228,
      "learning_rate": 1.6590852993554935e-05,
      "loss": 0.0,
      "step": 1089
    },
    {
      "epoch": 0.42198993418505615,
      "grad_norm": 0.0161809790879488,
      "learning_rate": 1.651191099375099e-05,
      "loss": 0.0002,
      "step": 1090
    },
    {
      "epoch": 0.42237708091366627,
      "grad_norm": 0.0017758719623088837,
      "learning_rate": 1.6433120090166943e-05,
      "loss": 0.0,
      "step": 1091
    },
    {
      "epoch": 0.42276422764227645,
      "grad_norm": 0.0008196650887839496,
      "learning_rate": 1.6354480638302588e-05,
      "loss": 0.0,
      "step": 1092
    },
    {
      "epoch": 0.42315137437088657,
      "grad_norm": 0.0007899802876636386,
      "learning_rate": 1.6275992992974308e-05,
      "loss": 0.0,
      "step": 1093
    },
    {
      "epoch": 0.4235385210994967,
      "grad_norm": 0.0006489930092357099,
      "learning_rate": 1.6197657508313597e-05,
      "loss": 0.0,
      "step": 1094
    },
    {
      "epoch": 0.42392566782810687,
      "grad_norm": 0.0010093534365296364,
      "learning_rate": 1.6119474537765395e-05,
      "loss": 0.0,
      "step": 1095
    },
    {
      "epoch": 0.424312814556717,
      "grad_norm": 0.0457945354282856,
      "learning_rate": 1.604144443408653e-05,
      "loss": 0.0005,
      "step": 1096
    },
    {
      "epoch": 0.42469996128532717,
      "grad_norm": 0.009583662264049053,
      "learning_rate": 1.5963567549344028e-05,
      "loss": 0.0001,
      "step": 1097
    },
    {
      "epoch": 0.4250871080139373,
      "grad_norm": 0.0023053644690662622,
      "learning_rate": 1.58858442349137e-05,
      "loss": 0.0,
      "step": 1098
    },
    {
      "epoch": 0.4254742547425474,
      "grad_norm": 0.0012813667999580503,
      "learning_rate": 1.580827484147842e-05,
      "loss": 0.0,
      "step": 1099
    },
    {
      "epoch": 0.4258614014711576,
      "grad_norm": 0.001665553660131991,
      "learning_rate": 1.5730859719026536e-05,
      "loss": 0.0,
      "step": 1100
    },
    {
      "epoch": 0.4262485481997677,
      "grad_norm": 0.03397725149989128,
      "learning_rate": 1.5653599216850378e-05,
      "loss": 0.0026,
      "step": 1101
    },
    {
      "epoch": 0.42663569492837783,
      "grad_norm": 0.0028618781361728907,
      "learning_rate": 1.557649368354464e-05,
      "loss": 0.0001,
      "step": 1102
    },
    {
      "epoch": 0.427022841656988,
      "grad_norm": 0.0466277040541172,
      "learning_rate": 1.5499543467004813e-05,
      "loss": 0.0011,
      "step": 1103
    },
    {
      "epoch": 0.4274099883855981,
      "grad_norm": 0.0036304357927292585,
      "learning_rate": 1.5422748914425557e-05,
      "loss": 0.0001,
      "step": 1104
    },
    {
      "epoch": 0.4277971351142083,
      "grad_norm": 0.0021317729260772467,
      "learning_rate": 1.5346110372299216e-05,
      "loss": 0.0,
      "step": 1105
    },
    {
      "epoch": 0.4281842818428184,
      "grad_norm": 0.0051021394319832325,
      "learning_rate": 1.526962818641428e-05,
      "loss": 0.0001,
      "step": 1106
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.0004642395942937583,
      "learning_rate": 1.5193302701853673e-05,
      "loss": 0.0,
      "step": 1107
    },
    {
      "epoch": 0.4289585753000387,
      "grad_norm": 0.0023515617940574884,
      "learning_rate": 1.5117134262993348e-05,
      "loss": 0.0,
      "step": 1108
    },
    {
      "epoch": 0.42934572202864885,
      "grad_norm": 0.004486995283514261,
      "learning_rate": 1.5041123213500674e-05,
      "loss": 0.0001,
      "step": 1109
    },
    {
      "epoch": 0.429732868757259,
      "grad_norm": 0.03434832766652107,
      "learning_rate": 1.4965269896332885e-05,
      "loss": 0.0004,
      "step": 1110
    },
    {
      "epoch": 0.43012001548586914,
      "grad_norm": 0.06803593039512634,
      "learning_rate": 1.488957465373551e-05,
      "loss": 0.0015,
      "step": 1111
    },
    {
      "epoch": 0.43050716221447927,
      "grad_norm": 0.0014034640043973923,
      "learning_rate": 1.4814037827240895e-05,
      "loss": 0.0,
      "step": 1112
    },
    {
      "epoch": 0.43089430894308944,
      "grad_norm": 0.00042223307536914945,
      "learning_rate": 1.47386597576666e-05,
      "loss": 0.0,
      "step": 1113
    },
    {
      "epoch": 0.43128145567169957,
      "grad_norm": 0.0007316286792047322,
      "learning_rate": 1.466344078511389e-05,
      "loss": 0.0,
      "step": 1114
    },
    {
      "epoch": 0.43166860240030974,
      "grad_norm": 0.0007396416622214019,
      "learning_rate": 1.4588381248966187e-05,
      "loss": 0.0,
      "step": 1115
    },
    {
      "epoch": 0.43205574912891986,
      "grad_norm": 0.0020533555652946234,
      "learning_rate": 1.4513481487887577e-05,
      "loss": 0.0,
      "step": 1116
    },
    {
      "epoch": 0.43244289585753,
      "grad_norm": 0.0023500213865190744,
      "learning_rate": 1.4438741839821185e-05,
      "loss": 0.0,
      "step": 1117
    },
    {
      "epoch": 0.43283004258614016,
      "grad_norm": 0.0006268412689678371,
      "learning_rate": 1.4364162641987777e-05,
      "loss": 0.0,
      "step": 1118
    },
    {
      "epoch": 0.4332171893147503,
      "grad_norm": 0.01405635941773653,
      "learning_rate": 1.4289744230884144e-05,
      "loss": 0.0001,
      "step": 1119
    },
    {
      "epoch": 0.43360433604336046,
      "grad_norm": 0.0008690317044965923,
      "learning_rate": 1.4215486942281658e-05,
      "loss": 0.0,
      "step": 1120
    },
    {
      "epoch": 0.4339914827719706,
      "grad_norm": 0.0021271114237606525,
      "learning_rate": 1.4141391111224633e-05,
      "loss": 0.0,
      "step": 1121
    },
    {
      "epoch": 0.4343786295005807,
      "grad_norm": 0.3007533848285675,
      "learning_rate": 1.4067457072029e-05,
      "loss": 0.0013,
      "step": 1122
    },
    {
      "epoch": 0.4347657762291909,
      "grad_norm": 0.00041973349289037287,
      "learning_rate": 1.3993685158280644e-05,
      "loss": 0.0,
      "step": 1123
    },
    {
      "epoch": 0.435152922957801,
      "grad_norm": 0.0007569995941594243,
      "learning_rate": 1.3920075702833918e-05,
      "loss": 0.0,
      "step": 1124
    },
    {
      "epoch": 0.4355400696864111,
      "grad_norm": 0.0018684781389310956,
      "learning_rate": 1.384662903781022e-05,
      "loss": 0.0,
      "step": 1125
    },
    {
      "epoch": 0.4359272164150213,
      "grad_norm": 0.00015323165280278772,
      "learning_rate": 1.3773345494596435e-05,
      "loss": 0.0,
      "step": 1126
    },
    {
      "epoch": 0.4363143631436314,
      "grad_norm": 0.0013038904871791601,
      "learning_rate": 1.3700225403843469e-05,
      "loss": 0.0,
      "step": 1127
    },
    {
      "epoch": 0.4367015098722416,
      "grad_norm": 0.0009059642907232046,
      "learning_rate": 1.3627269095464695e-05,
      "loss": 0.0,
      "step": 1128
    },
    {
      "epoch": 0.4370886566008517,
      "grad_norm": 0.0026266034692525864,
      "learning_rate": 1.3554476898634528e-05,
      "loss": 0.0,
      "step": 1129
    },
    {
      "epoch": 0.43747580332946184,
      "grad_norm": 0.0018995004938915372,
      "learning_rate": 1.3481849141786979e-05,
      "loss": 0.0,
      "step": 1130
    },
    {
      "epoch": 0.437862950058072,
      "grad_norm": 0.000616900681052357,
      "learning_rate": 1.3409386152614023e-05,
      "loss": 0.0,
      "step": 1131
    },
    {
      "epoch": 0.43825009678668214,
      "grad_norm": 0.11412354558706284,
      "learning_rate": 1.3337088258064278e-05,
      "loss": 0.0009,
      "step": 1132
    },
    {
      "epoch": 0.4386372435152923,
      "grad_norm": 0.004002030473202467,
      "learning_rate": 1.3264955784341438e-05,
      "loss": 0.0001,
      "step": 1133
    },
    {
      "epoch": 0.43902439024390244,
      "grad_norm": 0.09928884357213974,
      "learning_rate": 1.3192989056902854e-05,
      "loss": 0.0006,
      "step": 1134
    },
    {
      "epoch": 0.43941153697251256,
      "grad_norm": 0.001109429053030908,
      "learning_rate": 1.312118840045799e-05,
      "loss": 0.0,
      "step": 1135
    },
    {
      "epoch": 0.43979868370112274,
      "grad_norm": 0.0007349811494350433,
      "learning_rate": 1.3049554138967051e-05,
      "loss": 0.0,
      "step": 1136
    },
    {
      "epoch": 0.44018583042973286,
      "grad_norm": 0.0004371061804704368,
      "learning_rate": 1.2978086595639471e-05,
      "loss": 0.0,
      "step": 1137
    },
    {
      "epoch": 0.44057297715834304,
      "grad_norm": 0.0006483121542260051,
      "learning_rate": 1.290678609293246e-05,
      "loss": 0.0,
      "step": 1138
    },
    {
      "epoch": 0.44096012388695316,
      "grad_norm": 0.0007824698113836348,
      "learning_rate": 1.2835652952549537e-05,
      "loss": 0.0,
      "step": 1139
    },
    {
      "epoch": 0.4413472706155633,
      "grad_norm": 0.006598141510039568,
      "learning_rate": 1.276468749543911e-05,
      "loss": 0.0,
      "step": 1140
    },
    {
      "epoch": 0.44173441734417346,
      "grad_norm": 0.0006559481262229383,
      "learning_rate": 1.2693890041793015e-05,
      "loss": 0.0,
      "step": 1141
    },
    {
      "epoch": 0.4421215640727836,
      "grad_norm": 0.0008395421900786459,
      "learning_rate": 1.2623260911045032e-05,
      "loss": 0.0,
      "step": 1142
    },
    {
      "epoch": 0.4425087108013937,
      "grad_norm": 0.005088416859507561,
      "learning_rate": 1.2552800421869504e-05,
      "loss": 0.0,
      "step": 1143
    },
    {
      "epoch": 0.4428958575300039,
      "grad_norm": 0.0006087544606998563,
      "learning_rate": 1.2482508892179884e-05,
      "loss": 0.0,
      "step": 1144
    },
    {
      "epoch": 0.443283004258614,
      "grad_norm": 0.0017820927314460278,
      "learning_rate": 1.2412386639127272e-05,
      "loss": 0.0,
      "step": 1145
    },
    {
      "epoch": 0.4436701509872242,
      "grad_norm": 1.249230146408081,
      "learning_rate": 1.2342433979099e-05,
      "loss": 0.009,
      "step": 1146
    },
    {
      "epoch": 0.4440572977158343,
      "grad_norm": 0.007895207032561302,
      "learning_rate": 1.2272651227717225e-05,
      "loss": 0.0002,
      "step": 1147
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 0.006302273366600275,
      "learning_rate": 1.2203038699837482e-05,
      "loss": 0.0001,
      "step": 1148
    },
    {
      "epoch": 0.4448315911730546,
      "grad_norm": 0.003651441540569067,
      "learning_rate": 1.2133596709547234e-05,
      "loss": 0.0001,
      "step": 1149
    },
    {
      "epoch": 0.4452187379016647,
      "grad_norm": 0.06293000280857086,
      "learning_rate": 1.206432557016453e-05,
      "loss": 0.0002,
      "step": 1150
    },
    {
      "epoch": 0.4456058846302749,
      "grad_norm": 0.04901011288166046,
      "learning_rate": 1.1995225594236537e-05,
      "loss": 0.006,
      "step": 1151
    },
    {
      "epoch": 0.445993031358885,
      "grad_norm": 0.05200514569878578,
      "learning_rate": 1.1926297093538153e-05,
      "loss": 0.0013,
      "step": 1152
    },
    {
      "epoch": 0.44638017808749514,
      "grad_norm": 0.0014073471538722515,
      "learning_rate": 1.1857540379070541e-05,
      "loss": 0.0,
      "step": 1153
    },
    {
      "epoch": 0.4467673248161053,
      "grad_norm": 0.03230893984436989,
      "learning_rate": 1.178895576105985e-05,
      "loss": 0.0005,
      "step": 1154
    },
    {
      "epoch": 0.44715447154471544,
      "grad_norm": 0.016670379787683487,
      "learning_rate": 1.1720543548955709e-05,
      "loss": 0.0002,
      "step": 1155
    },
    {
      "epoch": 0.4475416182733256,
      "grad_norm": 0.006980563048273325,
      "learning_rate": 1.1652304051429824e-05,
      "loss": 0.0001,
      "step": 1156
    },
    {
      "epoch": 0.44792876500193574,
      "grad_norm": 0.09809542447328568,
      "learning_rate": 1.1584237576374673e-05,
      "loss": 0.0017,
      "step": 1157
    },
    {
      "epoch": 0.44831591173054586,
      "grad_norm": 0.11370933800935745,
      "learning_rate": 1.151634443090205e-05,
      "loss": 0.0044,
      "step": 1158
    },
    {
      "epoch": 0.44870305845915603,
      "grad_norm": 0.004476567264646292,
      "learning_rate": 1.1448624921341699e-05,
      "loss": 0.0001,
      "step": 1159
    },
    {
      "epoch": 0.44909020518776616,
      "grad_norm": 0.0005020067328587174,
      "learning_rate": 1.1381079353239915e-05,
      "loss": 0.0,
      "step": 1160
    },
    {
      "epoch": 0.44947735191637633,
      "grad_norm": 0.034463606774806976,
      "learning_rate": 1.1313708031358183e-05,
      "loss": 0.0002,
      "step": 1161
    },
    {
      "epoch": 0.44986449864498645,
      "grad_norm": 0.0015444280579686165,
      "learning_rate": 1.1246511259671843e-05,
      "loss": 0.0,
      "step": 1162
    },
    {
      "epoch": 0.4502516453735966,
      "grad_norm": 0.007986946031451225,
      "learning_rate": 1.1179489341368615e-05,
      "loss": 0.0,
      "step": 1163
    },
    {
      "epoch": 0.45063879210220675,
      "grad_norm": 0.00016696866077836603,
      "learning_rate": 1.1112642578847316e-05,
      "loss": 0.0,
      "step": 1164
    },
    {
      "epoch": 0.4510259388308169,
      "grad_norm": 0.0015510886441916227,
      "learning_rate": 1.1045971273716477e-05,
      "loss": 0.0,
      "step": 1165
    },
    {
      "epoch": 0.451413085559427,
      "grad_norm": 0.031885236501693726,
      "learning_rate": 1.0979475726792982e-05,
      "loss": 0.0002,
      "step": 1166
    },
    {
      "epoch": 0.4518002322880372,
      "grad_norm": 0.00022552194423042238,
      "learning_rate": 1.0913156238100669e-05,
      "loss": 0.0,
      "step": 1167
    },
    {
      "epoch": 0.4521873790166473,
      "grad_norm": 0.0008924312423914671,
      "learning_rate": 1.0847013106869041e-05,
      "loss": 0.0,
      "step": 1168
    },
    {
      "epoch": 0.45257452574525747,
      "grad_norm": 0.0007522033411078155,
      "learning_rate": 1.0781046631531888e-05,
      "loss": 0.0,
      "step": 1169
    },
    {
      "epoch": 0.4529616724738676,
      "grad_norm": 0.0009381092386320233,
      "learning_rate": 1.0715257109725928e-05,
      "loss": 0.0,
      "step": 1170
    },
    {
      "epoch": 0.4533488192024777,
      "grad_norm": 0.00644589914008975,
      "learning_rate": 1.0649644838289491e-05,
      "loss": 0.0,
      "step": 1171
    },
    {
      "epoch": 0.4537359659310879,
      "grad_norm": 0.0002498172107152641,
      "learning_rate": 1.058421011326114e-05,
      "loss": 0.0,
      "step": 1172
    },
    {
      "epoch": 0.454123112659698,
      "grad_norm": 0.0011802815133705735,
      "learning_rate": 1.0518953229878393e-05,
      "loss": 0.0001,
      "step": 1173
    },
    {
      "epoch": 0.4545102593883082,
      "grad_norm": 0.002405043924227357,
      "learning_rate": 1.0453874482576309e-05,
      "loss": 0.0,
      "step": 1174
    },
    {
      "epoch": 0.4548974061169183,
      "grad_norm": 0.002922722604125738,
      "learning_rate": 1.0388974164986249e-05,
      "loss": 0.0001,
      "step": 1175
    },
    {
      "epoch": 0.45528455284552843,
      "grad_norm": 0.03949505090713501,
      "learning_rate": 1.0324252569934495e-05,
      "loss": 0.0002,
      "step": 1176
    },
    {
      "epoch": 0.4556716995741386,
      "grad_norm": 0.000592520518694073,
      "learning_rate": 1.0259709989440952e-05,
      "loss": 0.0,
      "step": 1177
    },
    {
      "epoch": 0.45605884630274873,
      "grad_norm": 0.00023816191242076457,
      "learning_rate": 1.0195346714717813e-05,
      "loss": 0.0,
      "step": 1178
    },
    {
      "epoch": 0.4564459930313589,
      "grad_norm": 0.0009969983948394656,
      "learning_rate": 1.013116303616825e-05,
      "loss": 0.0,
      "step": 1179
    },
    {
      "epoch": 0.45683313975996903,
      "grad_norm": 0.0006109256646595895,
      "learning_rate": 1.0067159243385138e-05,
      "loss": 0.0,
      "step": 1180
    },
    {
      "epoch": 0.45722028648857915,
      "grad_norm": 0.0004375118878670037,
      "learning_rate": 1.0003335625149669e-05,
      "loss": 0.0,
      "step": 1181
    },
    {
      "epoch": 0.45760743321718933,
      "grad_norm": 0.012121845036745071,
      "learning_rate": 9.939692469430129e-06,
      "loss": 0.0001,
      "step": 1182
    },
    {
      "epoch": 0.45799457994579945,
      "grad_norm": 0.000974025868345052,
      "learning_rate": 9.876230063380571e-06,
      "loss": 0.0,
      "step": 1183
    },
    {
      "epoch": 0.45838172667440963,
      "grad_norm": 0.0005102133145555854,
      "learning_rate": 9.812948693339519e-06,
      "loss": 0.0,
      "step": 1184
    },
    {
      "epoch": 0.45876887340301975,
      "grad_norm": 0.000832306977827102,
      "learning_rate": 9.74984864482863e-06,
      "loss": 0.0,
      "step": 1185
    },
    {
      "epoch": 0.45915602013162987,
      "grad_norm": 0.0005011268076486886,
      "learning_rate": 9.68693020255152e-06,
      "loss": 0.0,
      "step": 1186
    },
    {
      "epoch": 0.45954316686024005,
      "grad_norm": 0.0014525051228702068,
      "learning_rate": 9.62419365039237e-06,
      "loss": 0.0,
      "step": 1187
    },
    {
      "epoch": 0.45993031358885017,
      "grad_norm": 0.007520552258938551,
      "learning_rate": 9.561639271414662e-06,
      "loss": 0.0001,
      "step": 1188
    },
    {
      "epoch": 0.4603174603174603,
      "grad_norm": 0.002319543855264783,
      "learning_rate": 9.499267347859959e-06,
      "loss": 0.0,
      "step": 1189
    },
    {
      "epoch": 0.46070460704607047,
      "grad_norm": 0.0008605946786701679,
      "learning_rate": 9.43707816114659e-06,
      "loss": 0.0,
      "step": 1190
    },
    {
      "epoch": 0.4610917537746806,
      "grad_norm": 0.011598880402743816,
      "learning_rate": 9.375071991868383e-06,
      "loss": 0.0001,
      "step": 1191
    },
    {
      "epoch": 0.46147890050329077,
      "grad_norm": 0.0019046214874833822,
      "learning_rate": 9.313249119793388e-06,
      "loss": 0.0,
      "step": 1192
    },
    {
      "epoch": 0.4618660472319009,
      "grad_norm": 0.0007018601754680276,
      "learning_rate": 9.25160982386264e-06,
      "loss": 0.0,
      "step": 1193
    },
    {
      "epoch": 0.462253193960511,
      "grad_norm": 0.004615155514329672,
      "learning_rate": 9.190154382188921e-06,
      "loss": 0.0,
      "step": 1194
    },
    {
      "epoch": 0.4626403406891212,
      "grad_norm": 0.012603587470948696,
      "learning_rate": 9.12888307205541e-06,
      "loss": 0.0001,
      "step": 1195
    },
    {
      "epoch": 0.4630274874177313,
      "grad_norm": 0.0009012959199026227,
      "learning_rate": 9.067796169914549e-06,
      "loss": 0.0,
      "step": 1196
    },
    {
      "epoch": 0.4634146341463415,
      "grad_norm": 0.005865184590220451,
      "learning_rate": 9.006893951386713e-06,
      "loss": 0.0001,
      "step": 1197
    },
    {
      "epoch": 0.4638017808749516,
      "grad_norm": 0.00023360588238574564,
      "learning_rate": 8.946176691259028e-06,
      "loss": 0.0,
      "step": 1198
    },
    {
      "epoch": 0.46418892760356173,
      "grad_norm": 0.1505637913942337,
      "learning_rate": 8.885644663484049e-06,
      "loss": 0.0005,
      "step": 1199
    },
    {
      "epoch": 0.4645760743321719,
      "grad_norm": 2.420736312866211,
      "learning_rate": 8.825298141178611e-06,
      "loss": 0.0156,
      "step": 1200
    },
    {
      "epoch": 0.46496322106078203,
      "grad_norm": 0.026697484776377678,
      "learning_rate": 8.765137396622557e-06,
      "loss": 0.0039,
      "step": 1201
    },
    {
      "epoch": 0.4653503677893922,
      "grad_norm": 0.03793786093592644,
      "learning_rate": 8.705162701257502e-06,
      "loss": 0.001,
      "step": 1202
    },
    {
      "epoch": 0.4657375145180023,
      "grad_norm": 0.020723620429635048,
      "learning_rate": 8.645374325685624e-06,
      "loss": 0.0003,
      "step": 1203
    },
    {
      "epoch": 0.46612466124661245,
      "grad_norm": 0.0008785614045336843,
      "learning_rate": 8.585772539668435e-06,
      "loss": 0.0,
      "step": 1204
    },
    {
      "epoch": 0.4665118079752226,
      "grad_norm": 0.002943212864920497,
      "learning_rate": 8.526357612125574e-06,
      "loss": 0.0,
      "step": 1205
    },
    {
      "epoch": 0.46689895470383275,
      "grad_norm": 0.057028789073228836,
      "learning_rate": 8.467129811133561e-06,
      "loss": 0.0003,
      "step": 1206
    },
    {
      "epoch": 0.46728610143244287,
      "grad_norm": 0.0032581747509539127,
      "learning_rate": 8.40808940392464e-06,
      "loss": 0.0001,
      "step": 1207
    },
    {
      "epoch": 0.46767324816105305,
      "grad_norm": 0.0932520404458046,
      "learning_rate": 8.349236656885545e-06,
      "loss": 0.0021,
      "step": 1208
    },
    {
      "epoch": 0.46806039488966317,
      "grad_norm": 0.006326075177639723,
      "learning_rate": 8.29057183555626e-06,
      "loss": 0.0002,
      "step": 1209
    },
    {
      "epoch": 0.46844754161827334,
      "grad_norm": 0.0002322104701306671,
      "learning_rate": 8.23209520462892e-06,
      "loss": 0.0,
      "step": 1210
    },
    {
      "epoch": 0.46883468834688347,
      "grad_norm": 0.0006544016650877893,
      "learning_rate": 8.173807027946528e-06,
      "loss": 0.0,
      "step": 1211
    },
    {
      "epoch": 0.4692218350754936,
      "grad_norm": 0.001205835840664804,
      "learning_rate": 8.115707568501768e-06,
      "loss": 0.0,
      "step": 1212
    },
    {
      "epoch": 0.46960898180410376,
      "grad_norm": 0.0018138798186555505,
      "learning_rate": 8.05779708843587e-06,
      "loss": 0.0,
      "step": 1213
    },
    {
      "epoch": 0.4699961285327139,
      "grad_norm": 0.00928629282861948,
      "learning_rate": 8.000075849037408e-06,
      "loss": 0.0002,
      "step": 1214
    },
    {
      "epoch": 0.47038327526132406,
      "grad_norm": 0.0003023847530130297,
      "learning_rate": 7.942544110741102e-06,
      "loss": 0.0,
      "step": 1215
    },
    {
      "epoch": 0.4707704219899342,
      "grad_norm": 0.006090863142162561,
      "learning_rate": 7.88520213312664e-06,
      "loss": 0.0,
      "step": 1216
    },
    {
      "epoch": 0.4711575687185443,
      "grad_norm": 0.0023785592056810856,
      "learning_rate": 7.828050174917528e-06,
      "loss": 0.0,
      "step": 1217
    },
    {
      "epoch": 0.4715447154471545,
      "grad_norm": 0.00038500342634506524,
      "learning_rate": 7.771088493979967e-06,
      "loss": 0.0,
      "step": 1218
    },
    {
      "epoch": 0.4719318621757646,
      "grad_norm": 0.08768702298402786,
      "learning_rate": 7.714317347321559e-06,
      "loss": 0.0005,
      "step": 1219
    },
    {
      "epoch": 0.4723190089043748,
      "grad_norm": 0.01025913842022419,
      "learning_rate": 7.657736991090264e-06,
      "loss": 0.0002,
      "step": 1220
    },
    {
      "epoch": 0.4727061556329849,
      "grad_norm": 0.0001984934351639822,
      "learning_rate": 7.601347680573223e-06,
      "loss": 0.0,
      "step": 1221
    },
    {
      "epoch": 0.473093302361595,
      "grad_norm": 0.000818421074654907,
      "learning_rate": 7.545149670195572e-06,
      "loss": 0.0,
      "step": 1222
    },
    {
      "epoch": 0.4734804490902052,
      "grad_norm": 0.021213488653302193,
      "learning_rate": 7.489143213519301e-06,
      "loss": 0.0001,
      "step": 1223
    },
    {
      "epoch": 0.4738675958188153,
      "grad_norm": 0.033915307372808456,
      "learning_rate": 7.433328563242142e-06,
      "loss": 0.0002,
      "step": 1224
    },
    {
      "epoch": 0.4742547425474255,
      "grad_norm": 0.00025099285994656384,
      "learning_rate": 7.377705971196397e-06,
      "loss": 0.0,
      "step": 1225
    },
    {
      "epoch": 0.4746418892760356,
      "grad_norm": 0.00011749350233003497,
      "learning_rate": 7.322275688347818e-06,
      "loss": 0.0,
      "step": 1226
    },
    {
      "epoch": 0.47502903600464574,
      "grad_norm": 0.00033711831201799214,
      "learning_rate": 7.267037964794476e-06,
      "loss": 0.0,
      "step": 1227
    },
    {
      "epoch": 0.4754161827332559,
      "grad_norm": 0.001535428804345429,
      "learning_rate": 7.211993049765614e-06,
      "loss": 0.0,
      "step": 1228
    },
    {
      "epoch": 0.47580332946186604,
      "grad_norm": 0.0005026145954616368,
      "learning_rate": 7.157141191620548e-06,
      "loss": 0.0,
      "step": 1229
    },
    {
      "epoch": 0.47619047619047616,
      "grad_norm": 0.0004242948198225349,
      "learning_rate": 7.102482637847502e-06,
      "loss": 0.0,
      "step": 1230
    },
    {
      "epoch": 0.47657762291908634,
      "grad_norm": 0.0002841683162841946,
      "learning_rate": 7.048017635062559e-06,
      "loss": 0.0,
      "step": 1231
    },
    {
      "epoch": 0.47696476964769646,
      "grad_norm": 0.0008631069213151932,
      "learning_rate": 6.993746429008497e-06,
      "loss": 0.0,
      "step": 1232
    },
    {
      "epoch": 0.47735191637630664,
      "grad_norm": 0.0019993663299828768,
      "learning_rate": 6.9396692645536946e-06,
      "loss": 0.0,
      "step": 1233
    },
    {
      "epoch": 0.47773906310491676,
      "grad_norm": 0.0013842112384736538,
      "learning_rate": 6.885786385691023e-06,
      "loss": 0.0,
      "step": 1234
    },
    {
      "epoch": 0.4781262098335269,
      "grad_norm": 0.00023990539193619043,
      "learning_rate": 6.8320980355367605e-06,
      "loss": 0.0,
      "step": 1235
    },
    {
      "epoch": 0.47851335656213706,
      "grad_norm": 0.0004480506759136915,
      "learning_rate": 6.778604456329485e-06,
      "loss": 0.0,
      "step": 1236
    },
    {
      "epoch": 0.4789005032907472,
      "grad_norm": 0.003706539049744606,
      "learning_rate": 6.725305889428945e-06,
      "loss": 0.0,
      "step": 1237
    },
    {
      "epoch": 0.47928765001935736,
      "grad_norm": 0.0007369938539341092,
      "learning_rate": 6.672202575315045e-06,
      "loss": 0.0,
      "step": 1238
    },
    {
      "epoch": 0.4796747967479675,
      "grad_norm": 0.00016758940182626247,
      "learning_rate": 6.619294753586708e-06,
      "loss": 0.0,
      "step": 1239
    },
    {
      "epoch": 0.4800619434765776,
      "grad_norm": 0.0006811873754486442,
      "learning_rate": 6.566582662960818e-06,
      "loss": 0.0,
      "step": 1240
    },
    {
      "epoch": 0.4804490902051878,
      "grad_norm": 0.00018206462846137583,
      "learning_rate": 6.514066541271086e-06,
      "loss": 0.0,
      "step": 1241
    },
    {
      "epoch": 0.4808362369337979,
      "grad_norm": 0.0009445471223443747,
      "learning_rate": 6.461746625467086e-06,
      "loss": 0.0,
      "step": 1242
    },
    {
      "epoch": 0.4812233836624081,
      "grad_norm": 0.005226759240031242,
      "learning_rate": 6.409623151613103e-06,
      "loss": 0.0001,
      "step": 1243
    },
    {
      "epoch": 0.4816105303910182,
      "grad_norm": 0.0017884591361507773,
      "learning_rate": 6.3576963548870496e-06,
      "loss": 0.0,
      "step": 1244
    },
    {
      "epoch": 0.4819976771196283,
      "grad_norm": 0.0026653881650418043,
      "learning_rate": 6.305966469579489e-06,
      "loss": 0.0,
      "step": 1245
    },
    {
      "epoch": 0.4823848238482385,
      "grad_norm": 0.00038798339664936066,
      "learning_rate": 6.2544337290925185e-06,
      "loss": 0.0,
      "step": 1246
    },
    {
      "epoch": 0.4827719705768486,
      "grad_norm": 0.0003185675013810396,
      "learning_rate": 6.203098365938731e-06,
      "loss": 0.0,
      "step": 1247
    },
    {
      "epoch": 0.4831591173054588,
      "grad_norm": 0.000892972107976675,
      "learning_rate": 6.1519606117401426e-06,
      "loss": 0.0,
      "step": 1248
    },
    {
      "epoch": 0.4835462640340689,
      "grad_norm": 1.4739562273025513,
      "learning_rate": 6.101020697227189e-06,
      "loss": 0.0245,
      "step": 1249
    },
    {
      "epoch": 0.48393341076267904,
      "grad_norm": 0.004798859357833862,
      "learning_rate": 6.050278852237701e-06,
      "loss": 0.0001,
      "step": 1250
    },
    {
      "epoch": 0.4843205574912892,
      "grad_norm": 0.01590668596327305,
      "learning_rate": 5.999735305715754e-06,
      "loss": 0.001,
      "step": 1251
    },
    {
      "epoch": 0.48470770421989934,
      "grad_norm": 0.0043892329558730125,
      "learning_rate": 5.949390285710776e-06,
      "loss": 0.0001,
      "step": 1252
    },
    {
      "epoch": 0.48509485094850946,
      "grad_norm": 0.06413737684488297,
      "learning_rate": 5.899244019376426e-06,
      "loss": 0.0016,
      "step": 1253
    },
    {
      "epoch": 0.48548199767711964,
      "grad_norm": 0.006050611846148968,
      "learning_rate": 5.849296732969623e-06,
      "loss": 0.0001,
      "step": 1254
    },
    {
      "epoch": 0.48586914440572976,
      "grad_norm": 0.0022481298074126244,
      "learning_rate": 5.799548651849457e-06,
      "loss": 0.0,
      "step": 1255
    },
    {
      "epoch": 0.48625629113433994,
      "grad_norm": 0.004415793344378471,
      "learning_rate": 5.750000000476258e-06,
      "loss": 0.0001,
      "step": 1256
    },
    {
      "epoch": 0.48664343786295006,
      "grad_norm": 0.0013036815216764808,
      "learning_rate": 5.700651002410523e-06,
      "loss": 0.0,
      "step": 1257
    },
    {
      "epoch": 0.4870305845915602,
      "grad_norm": 0.0007666132296435535,
      "learning_rate": 5.651501880311933e-06,
      "loss": 0.0,
      "step": 1258
    },
    {
      "epoch": 0.48741773132017036,
      "grad_norm": 0.012893356382846832,
      "learning_rate": 5.6025528559383254e-06,
      "loss": 0.0001,
      "step": 1259
    },
    {
      "epoch": 0.4878048780487805,
      "grad_norm": 0.06497503072023392,
      "learning_rate": 5.553804150144737e-06,
      "loss": 0.0047,
      "step": 1260
    },
    {
      "epoch": 0.48819202477739065,
      "grad_norm": 0.004026665352284908,
      "learning_rate": 5.505255982882357e-06,
      "loss": 0.0,
      "step": 1261
    },
    {
      "epoch": 0.4885791715060008,
      "grad_norm": 0.001962028443813324,
      "learning_rate": 5.456908573197544e-06,
      "loss": 0.0,
      "step": 1262
    },
    {
      "epoch": 0.4889663182346109,
      "grad_norm": 0.0005447390722110868,
      "learning_rate": 5.408762139230888e-06,
      "loss": 0.0,
      "step": 1263
    },
    {
      "epoch": 0.4893534649632211,
      "grad_norm": 0.001602734555490315,
      "learning_rate": 5.360816898216164e-06,
      "loss": 0.0,
      "step": 1264
    },
    {
      "epoch": 0.4897406116918312,
      "grad_norm": 0.0005945903831161559,
      "learning_rate": 5.3130730664793795e-06,
      "loss": 0.0,
      "step": 1265
    },
    {
      "epoch": 0.4901277584204414,
      "grad_norm": 0.012403713539242744,
      "learning_rate": 5.265530859437801e-06,
      "loss": 0.0001,
      "step": 1266
    },
    {
      "epoch": 0.4905149051490515,
      "grad_norm": 0.03977184742689133,
      "learning_rate": 5.218190491598973e-06,
      "loss": 0.0001,
      "step": 1267
    },
    {
      "epoch": 0.4909020518776616,
      "grad_norm": 0.0005819656071253121,
      "learning_rate": 5.17105217655976e-06,
      "loss": 0.0,
      "step": 1268
    },
    {
      "epoch": 0.4912891986062718,
      "grad_norm": 0.0012119417078793049,
      "learning_rate": 5.124116127005363e-06,
      "loss": 0.0,
      "step": 1269
    },
    {
      "epoch": 0.4916763453348819,
      "grad_norm": 0.0011156294494867325,
      "learning_rate": 5.077382554708382e-06,
      "loss": 0.0,
      "step": 1270
    },
    {
      "epoch": 0.49206349206349204,
      "grad_norm": 0.00014334850129671395,
      "learning_rate": 5.030851670527853e-06,
      "loss": 0.0,
      "step": 1271
    },
    {
      "epoch": 0.4924506387921022,
      "grad_norm": 0.01676209457218647,
      "learning_rate": 4.984523684408304e-06,
      "loss": 0.0002,
      "step": 1272
    },
    {
      "epoch": 0.49283778552071233,
      "grad_norm": 0.0005447831354103982,
      "learning_rate": 4.938398805378763e-06,
      "loss": 0.0,
      "step": 1273
    },
    {
      "epoch": 0.4932249322493225,
      "grad_norm": 0.00024948938516899943,
      "learning_rate": 4.892477241551901e-06,
      "loss": 0.0,
      "step": 1274
    },
    {
      "epoch": 0.49361207897793263,
      "grad_norm": 0.001213126815855503,
      "learning_rate": 4.84675920012303e-06,
      "loss": 0.0,
      "step": 1275
    },
    {
      "epoch": 0.49399922570654275,
      "grad_norm": 0.00047369138337671757,
      "learning_rate": 4.801244887369144e-06,
      "loss": 0.0,
      "step": 1276
    },
    {
      "epoch": 0.49438637243515293,
      "grad_norm": 0.00033764008549042046,
      "learning_rate": 4.755934508648058e-06,
      "loss": 0.0,
      "step": 1277
    },
    {
      "epoch": 0.49477351916376305,
      "grad_norm": 0.0003957216104026884,
      "learning_rate": 4.7108282683974544e-06,
      "loss": 0.0,
      "step": 1278
    },
    {
      "epoch": 0.49516066589237323,
      "grad_norm": 0.0009651805739849806,
      "learning_rate": 4.665926370133949e-06,
      "loss": 0.0,
      "step": 1279
    },
    {
      "epoch": 0.49554781262098335,
      "grad_norm": 0.0003068702353630215,
      "learning_rate": 4.621229016452156e-06,
      "loss": 0.0,
      "step": 1280
    },
    {
      "epoch": 0.4959349593495935,
      "grad_norm": 0.0007897759787738323,
      "learning_rate": 4.576736409023813e-06,
      "loss": 0.0,
      "step": 1281
    },
    {
      "epoch": 0.49632210607820365,
      "grad_norm": 0.0002980028511956334,
      "learning_rate": 4.532448748596885e-06,
      "loss": 0.0,
      "step": 1282
    },
    {
      "epoch": 0.4967092528068138,
      "grad_norm": 0.0010113201569765806,
      "learning_rate": 4.488366234994579e-06,
      "loss": 0.0,
      "step": 1283
    },
    {
      "epoch": 0.49709639953542395,
      "grad_norm": 0.0002822222013492137,
      "learning_rate": 4.444489067114521e-06,
      "loss": 0.0,
      "step": 1284
    },
    {
      "epoch": 0.49748354626403407,
      "grad_norm": 0.0009353689965792,
      "learning_rate": 4.4008174429278185e-06,
      "loss": 0.0,
      "step": 1285
    },
    {
      "epoch": 0.4978706929926442,
      "grad_norm": 0.00048544921446591616,
      "learning_rate": 4.357351559478201e-06,
      "loss": 0.0,
      "step": 1286
    },
    {
      "epoch": 0.49825783972125437,
      "grad_norm": 0.00021501122682821006,
      "learning_rate": 4.31409161288106e-06,
      "loss": 0.0,
      "step": 1287
    },
    {
      "epoch": 0.4986449864498645,
      "grad_norm": 0.0011064648861065507,
      "learning_rate": 4.271037798322658e-06,
      "loss": 0.0,
      "step": 1288
    },
    {
      "epoch": 0.49903213317847467,
      "grad_norm": 0.0019151505548506975,
      "learning_rate": 4.2281903100591824e-06,
      "loss": 0.0,
      "step": 1289
    },
    {
      "epoch": 0.4994192799070848,
      "grad_norm": 0.003015509806573391,
      "learning_rate": 4.185549341415901e-06,
      "loss": 0.0001,
      "step": 1290
    },
    {
      "epoch": 0.4998064266356949,
      "grad_norm": 0.0012218995252624154,
      "learning_rate": 4.143115084786259e-06,
      "loss": 0.0,
      "step": 1291
    },
    {
      "epoch": 0.500193573364305,
      "grad_norm": 0.0007260179845616221,
      "learning_rate": 4.100887731631054e-06,
      "loss": 0.0,
      "step": 1292
    },
    {
      "epoch": 0.5005807200929152,
      "grad_norm": 0.0013869833201169968,
      "learning_rate": 4.058867472477529e-06,
      "loss": 0.0,
      "step": 1293
    },
    {
      "epoch": 0.5009678668215254,
      "grad_norm": 0.0003031007363460958,
      "learning_rate": 4.017054496918521e-06,
      "loss": 0.0,
      "step": 1294
    },
    {
      "epoch": 0.5013550135501355,
      "grad_norm": 0.0006917419377714396,
      "learning_rate": 3.975448993611652e-06,
      "loss": 0.0,
      "step": 1295
    },
    {
      "epoch": 0.5017421602787456,
      "grad_norm": 0.007670098915696144,
      "learning_rate": 3.934051150278417e-06,
      "loss": 0.0001,
      "step": 1296
    },
    {
      "epoch": 0.5021293070073558,
      "grad_norm": 0.0030773854814469814,
      "learning_rate": 3.892861153703342e-06,
      "loss": 0.0,
      "step": 1297
    },
    {
      "epoch": 0.502516453735966,
      "grad_norm": 0.0010149420704692602,
      "learning_rate": 3.851879189733221e-06,
      "loss": 0.0,
      "step": 1298
    },
    {
      "epoch": 0.502903600464576,
      "grad_norm": 0.0006883174646645784,
      "learning_rate": 3.811105443276164e-06,
      "loss": 0.0,
      "step": 1299
    },
    {
      "epoch": 0.5032907471931862,
      "grad_norm": 0.15667524933815002,
      "learning_rate": 3.77054009830084e-06,
      "loss": 0.001,
      "step": 1300
    },
    {
      "epoch": 0.5036778939217964,
      "grad_norm": 0.016902372241020203,
      "learning_rate": 3.7301833378356076e-06,
      "loss": 0.002,
      "step": 1301
    },
    {
      "epoch": 0.5040650406504065,
      "grad_norm": 0.0014611108927056193,
      "learning_rate": 3.690035343967724e-06,
      "loss": 0.0001,
      "step": 1302
    },
    {
      "epoch": 0.5044521873790166,
      "grad_norm": 0.0020591116044670343,
      "learning_rate": 3.6500962978424924e-06,
      "loss": 0.0001,
      "step": 1303
    },
    {
      "epoch": 0.5048393341076268,
      "grad_norm": 0.05074653401970863,
      "learning_rate": 3.6103663796624555e-06,
      "loss": 0.0012,
      "step": 1304
    },
    {
      "epoch": 0.5052264808362369,
      "grad_norm": 0.0005115514504723251,
      "learning_rate": 3.570845768686576e-06,
      "loss": 0.0,
      "step": 1305
    },
    {
      "epoch": 0.5056136275648471,
      "grad_norm": 0.04323451220989227,
      "learning_rate": 3.5315346432294726e-06,
      "loss": 0.0072,
      "step": 1306
    },
    {
      "epoch": 0.5060007742934572,
      "grad_norm": 0.0019445320358499885,
      "learning_rate": 3.4924331806605314e-06,
      "loss": 0.0,
      "step": 1307
    },
    {
      "epoch": 0.5063879210220673,
      "grad_norm": 0.002080347388982773,
      "learning_rate": 3.4535415574031783e-06,
      "loss": 0.0,
      "step": 1308
    },
    {
      "epoch": 0.5067750677506775,
      "grad_norm": 0.0003646564146038145,
      "learning_rate": 3.4148599489340517e-06,
      "loss": 0.0,
      "step": 1309
    },
    {
      "epoch": 0.5071622144792877,
      "grad_norm": 0.0001835545408539474,
      "learning_rate": 3.376388529782215e-06,
      "loss": 0.0,
      "step": 1310
    },
    {
      "epoch": 0.5075493612078978,
      "grad_norm": 0.001469470327720046,
      "learning_rate": 3.3381274735283684e-06,
      "loss": 0.0,
      "step": 1311
    },
    {
      "epoch": 0.5079365079365079,
      "grad_norm": 0.0013578154612332582,
      "learning_rate": 3.3000769528040653e-06,
      "loss": 0.0,
      "step": 1312
    },
    {
      "epoch": 0.5083236546651181,
      "grad_norm": 0.004504669923335314,
      "learning_rate": 3.2622371392909524e-06,
      "loss": 0.0001,
      "step": 1313
    },
    {
      "epoch": 0.5087108013937283,
      "grad_norm": 0.0013383106561377645,
      "learning_rate": 3.2246082037199532e-06,
      "loss": 0.0,
      "step": 1314
    },
    {
      "epoch": 0.5090979481223383,
      "grad_norm": 0.0010525296675041318,
      "learning_rate": 3.187190315870542e-06,
      "loss": 0.0,
      "step": 1315
    },
    {
      "epoch": 0.5094850948509485,
      "grad_norm": 0.00011710682156262919,
      "learning_rate": 3.149983644569948e-06,
      "loss": 0.0,
      "step": 1316
    },
    {
      "epoch": 0.5098722415795587,
      "grad_norm": 0.0007033901638351381,
      "learning_rate": 3.1129883576924203e-06,
      "loss": 0.0,
      "step": 1317
    },
    {
      "epoch": 0.5102593883081687,
      "grad_norm": 0.0018138588638976216,
      "learning_rate": 3.0762046221584294e-06,
      "loss": 0.0,
      "step": 1318
    },
    {
      "epoch": 0.5106465350367789,
      "grad_norm": 0.17213991284370422,
      "learning_rate": 3.0396326039339507e-06,
      "loss": 0.0025,
      "step": 1319
    },
    {
      "epoch": 0.5110336817653891,
      "grad_norm": 0.00046120636397972703,
      "learning_rate": 3.0032724680297107e-06,
      "loss": 0.0,
      "step": 1320
    },
    {
      "epoch": 0.5114208284939993,
      "grad_norm": 0.0004718530108220875,
      "learning_rate": 2.96712437850043e-06,
      "loss": 0.0,
      "step": 1321
    },
    {
      "epoch": 0.5118079752226093,
      "grad_norm": 0.0023787531536072493,
      "learning_rate": 2.9311884984440875e-06,
      "loss": 0.0,
      "step": 1322
    },
    {
      "epoch": 0.5121951219512195,
      "grad_norm": 0.0011449018493294716,
      "learning_rate": 2.8954649900011845e-06,
      "loss": 0.0,
      "step": 1323
    },
    {
      "epoch": 0.5125822686798297,
      "grad_norm": 0.04674423485994339,
      "learning_rate": 2.8599540143540327e-06,
      "loss": 0.0004,
      "step": 1324
    },
    {
      "epoch": 0.5129694154084398,
      "grad_norm": 0.00027449653134681284,
      "learning_rate": 2.8246557317259727e-06,
      "loss": 0.0,
      "step": 1325
    },
    {
      "epoch": 0.5133565621370499,
      "grad_norm": 0.0003032211388926953,
      "learning_rate": 2.7895703013807118e-06,
      "loss": 0.0,
      "step": 1326
    },
    {
      "epoch": 0.5137437088656601,
      "grad_norm": 0.0003420572611503303,
      "learning_rate": 2.7546978816215874e-06,
      "loss": 0.0,
      "step": 1327
    },
    {
      "epoch": 0.5141308555942702,
      "grad_norm": 0.00011091092164861038,
      "learning_rate": 2.7200386297908387e-06,
      "loss": 0.0,
      "step": 1328
    },
    {
      "epoch": 0.5145180023228804,
      "grad_norm": 0.0007387629011645913,
      "learning_rate": 2.6855927022688743e-06,
      "loss": 0.0,
      "step": 1329
    },
    {
      "epoch": 0.5149051490514905,
      "grad_norm": 0.00038926073466427624,
      "learning_rate": 2.651360254473645e-06,
      "loss": 0.0,
      "step": 1330
    },
    {
      "epoch": 0.5152922957801006,
      "grad_norm": 0.0004905264358967543,
      "learning_rate": 2.6173414408598827e-06,
      "loss": 0.0,
      "step": 1331
    },
    {
      "epoch": 0.5156794425087108,
      "grad_norm": 0.0011100409319624305,
      "learning_rate": 2.5835364149183803e-06,
      "loss": 0.0,
      "step": 1332
    },
    {
      "epoch": 0.516066589237321,
      "grad_norm": 0.0009677124326117337,
      "learning_rate": 2.549945329175363e-06,
      "loss": 0.0,
      "step": 1333
    },
    {
      "epoch": 0.5164537359659311,
      "grad_norm": 0.0003739868989214301,
      "learning_rate": 2.516568335191777e-06,
      "loss": 0.0,
      "step": 1334
    },
    {
      "epoch": 0.5168408826945412,
      "grad_norm": 0.00016991533630061895,
      "learning_rate": 2.483405583562587e-06,
      "loss": 0.0,
      "step": 1335
    },
    {
      "epoch": 0.5172280294231514,
      "grad_norm": 0.0003253307659178972,
      "learning_rate": 2.450457223916097e-06,
      "loss": 0.0,
      "step": 1336
    },
    {
      "epoch": 0.5176151761517616,
      "grad_norm": 0.00035142659908160567,
      "learning_rate": 2.417723404913302e-06,
      "loss": 0.0,
      "step": 1337
    },
    {
      "epoch": 0.5180023228803716,
      "grad_norm": 0.00029594748048111796,
      "learning_rate": 2.3852042742472215e-06,
      "loss": 0.0,
      "step": 1338
    },
    {
      "epoch": 0.5183894696089818,
      "grad_norm": 0.000983862904831767,
      "learning_rate": 2.3528999786421756e-06,
      "loss": 0.0,
      "step": 1339
    },
    {
      "epoch": 0.518776616337592,
      "grad_norm": 0.00020164911984466016,
      "learning_rate": 2.3208106638531846e-06,
      "loss": 0.0,
      "step": 1340
    },
    {
      "epoch": 0.519163763066202,
      "grad_norm": 0.0005400851950980723,
      "learning_rate": 2.2889364746652874e-06,
      "loss": 0.0,
      "step": 1341
    },
    {
      "epoch": 0.5195509097948122,
      "grad_norm": 0.00030268036061897874,
      "learning_rate": 2.2572775548928893e-06,
      "loss": 0.0,
      "step": 1342
    },
    {
      "epoch": 0.5199380565234224,
      "grad_norm": 0.0007914683665148914,
      "learning_rate": 2.2258340473790996e-06,
      "loss": 0.0,
      "step": 1343
    },
    {
      "epoch": 0.5203252032520326,
      "grad_norm": 0.0005217034486122429,
      "learning_rate": 2.1946060939951164e-06,
      "loss": 0.0,
      "step": 1344
    },
    {
      "epoch": 0.5207123499806426,
      "grad_norm": 0.00027877753018401563,
      "learning_rate": 2.163593835639566e-06,
      "loss": 0.0,
      "step": 1345
    },
    {
      "epoch": 0.5210994967092528,
      "grad_norm": 0.002984454622492194,
      "learning_rate": 2.132797412237869e-06,
      "loss": 0.0,
      "step": 1346
    },
    {
      "epoch": 0.521486643437863,
      "grad_norm": 0.00027315475745126605,
      "learning_rate": 2.1022169627416153e-06,
      "loss": 0.0,
      "step": 1347
    },
    {
      "epoch": 0.5218737901664731,
      "grad_norm": 0.001330073457211256,
      "learning_rate": 2.0718526251279346e-06,
      "loss": 0.0,
      "step": 1348
    },
    {
      "epoch": 0.5222609368950832,
      "grad_norm": 0.0018169977702200413,
      "learning_rate": 2.0417045363988752e-06,
      "loss": 0.0,
      "step": 1349
    },
    {
      "epoch": 0.5226480836236934,
      "grad_norm": 0.0006412347429431975,
      "learning_rate": 2.0117728325807606e-06,
      "loss": 0.0,
      "step": 1350
    },
    {
      "epoch": 0.5230352303523035,
      "grad_norm": 0.07314898818731308,
      "learning_rate": 1.9820576487236284e-06,
      "loss": 0.0036,
      "step": 1351
    },
    {
      "epoch": 0.5234223770809137,
      "grad_norm": 0.001173462369479239,
      "learning_rate": 1.9525591189005878e-06,
      "loss": 0.0001,
      "step": 1352
    },
    {
      "epoch": 0.5238095238095238,
      "grad_norm": 0.029476692900061607,
      "learning_rate": 1.9232773762071944e-06,
      "loss": 0.0013,
      "step": 1353
    },
    {
      "epoch": 0.5241966705381339,
      "grad_norm": 0.038780104368925095,
      "learning_rate": 1.8942125527609045e-06,
      "loss": 0.0012,
      "step": 1354
    },
    {
      "epoch": 0.5245838172667441,
      "grad_norm": 0.010619153268635273,
      "learning_rate": 1.8653647797004236e-06,
      "loss": 0.0001,
      "step": 1355
    },
    {
      "epoch": 0.5249709639953543,
      "grad_norm": 0.027320994064211845,
      "learning_rate": 1.8367341871851518e-06,
      "loss": 0.0002,
      "step": 1356
    },
    {
      "epoch": 0.5253581107239644,
      "grad_norm": 0.005327960010617971,
      "learning_rate": 1.8083209043945782e-06,
      "loss": 0.0001,
      "step": 1357
    },
    {
      "epoch": 0.5257452574525745,
      "grad_norm": 0.010818984359502792,
      "learning_rate": 1.7801250595277096e-06,
      "loss": 0.0001,
      "step": 1358
    },
    {
      "epoch": 0.5261324041811847,
      "grad_norm": 0.0013082500081509352,
      "learning_rate": 1.7521467798024772e-06,
      "loss": 0.0,
      "step": 1359
    },
    {
      "epoch": 0.5265195509097949,
      "grad_norm": 0.00018383938004262745,
      "learning_rate": 1.724386191455185e-06,
      "loss": 0.0,
      "step": 1360
    },
    {
      "epoch": 0.5269066976384049,
      "grad_norm": 0.06656922399997711,
      "learning_rate": 1.6968434197399075e-06,
      "loss": 0.0007,
      "step": 1361
    },
    {
      "epoch": 0.5272938443670151,
      "grad_norm": 0.004147067666053772,
      "learning_rate": 1.6695185889279597e-06,
      "loss": 0.0,
      "step": 1362
    },
    {
      "epoch": 0.5276809910956253,
      "grad_norm": 0.003467077622190118,
      "learning_rate": 1.6424118223073214e-06,
      "loss": 0.0,
      "step": 1363
    },
    {
      "epoch": 0.5280681378242353,
      "grad_norm": 0.00022673551575280726,
      "learning_rate": 1.6155232421820653e-06,
      "loss": 0.0,
      "step": 1364
    },
    {
      "epoch": 0.5284552845528455,
      "grad_norm": 0.004515313543379307,
      "learning_rate": 1.5888529698718346e-06,
      "loss": 0.0,
      "step": 1365
    },
    {
      "epoch": 0.5288424312814557,
      "grad_norm": 0.0015816775849089026,
      "learning_rate": 1.5624011257112714e-06,
      "loss": 0.0,
      "step": 1366
    },
    {
      "epoch": 0.5292295780100658,
      "grad_norm": 0.0008955047815106809,
      "learning_rate": 1.5361678290494952e-06,
      "loss": 0.0,
      "step": 1367
    },
    {
      "epoch": 0.5296167247386759,
      "grad_norm": 0.00045474289800040424,
      "learning_rate": 1.5101531982495308e-06,
      "loss": 0.0,
      "step": 1368
    },
    {
      "epoch": 0.5300038714672861,
      "grad_norm": 0.014475700445473194,
      "learning_rate": 1.4843573506878094e-06,
      "loss": 0.0003,
      "step": 1369
    },
    {
      "epoch": 0.5303910181958963,
      "grad_norm": 0.00026755264843814075,
      "learning_rate": 1.4587804027536456e-06,
      "loss": 0.0,
      "step": 1370
    },
    {
      "epoch": 0.5307781649245064,
      "grad_norm": 0.0009568330715410411,
      "learning_rate": 1.4334224698486554e-06,
      "loss": 0.0,
      "step": 1371
    },
    {
      "epoch": 0.5311653116531165,
      "grad_norm": 0.005675049964338541,
      "learning_rate": 1.4082836663862898e-06,
      "loss": 0.0,
      "step": 1372
    },
    {
      "epoch": 0.5315524583817267,
      "grad_norm": 0.0010455515002831817,
      "learning_rate": 1.3833641057913017e-06,
      "loss": 0.0,
      "step": 1373
    },
    {
      "epoch": 0.5319396051103368,
      "grad_norm": 0.0003837356634903699,
      "learning_rate": 1.3586639004992407e-06,
      "loss": 0.0,
      "step": 1374
    },
    {
      "epoch": 0.532326751838947,
      "grad_norm": 0.007896430790424347,
      "learning_rate": 1.3341831619559208e-06,
      "loss": 0.0001,
      "step": 1375
    },
    {
      "epoch": 0.5327138985675571,
      "grad_norm": 0.000981520744971931,
      "learning_rate": 1.309922000616942e-06,
      "loss": 0.0,
      "step": 1376
    },
    {
      "epoch": 0.5331010452961672,
      "grad_norm": 0.1035628467798233,
      "learning_rate": 1.2858805259471974e-06,
      "loss": 0.0023,
      "step": 1377
    },
    {
      "epoch": 0.5334881920247774,
      "grad_norm": 0.00016075636085588485,
      "learning_rate": 1.2620588464203553e-06,
      "loss": 0.0,
      "step": 1378
    },
    {
      "epoch": 0.5338753387533876,
      "grad_norm": 0.00027309221331961453,
      "learning_rate": 1.2384570695183783e-06,
      "loss": 0.0,
      "step": 1379
    },
    {
      "epoch": 0.5342624854819977,
      "grad_norm": 0.00031268582097254694,
      "learning_rate": 1.21507530173105e-06,
      "loss": 0.0,
      "step": 1380
    },
    {
      "epoch": 0.5346496322106078,
      "grad_norm": 0.0011569913476705551,
      "learning_rate": 1.1919136485554983e-06,
      "loss": 0.0,
      "step": 1381
    },
    {
      "epoch": 0.535036778939218,
      "grad_norm": 0.056165896356105804,
      "learning_rate": 1.1689722144956671e-06,
      "loss": 0.0001,
      "step": 1382
    },
    {
      "epoch": 0.5354239256678281,
      "grad_norm": 0.0005352521548047662,
      "learning_rate": 1.1462511030619295e-06,
      "loss": 0.0,
      "step": 1383
    },
    {
      "epoch": 0.5358110723964382,
      "grad_norm": 0.00021589416428469121,
      "learning_rate": 1.1237504167705525e-06,
      "loss": 0.0,
      "step": 1384
    },
    {
      "epoch": 0.5361982191250484,
      "grad_norm": 0.0008537312969565392,
      "learning_rate": 1.1014702571432612e-06,
      "loss": 0.0,
      "step": 1385
    },
    {
      "epoch": 0.5365853658536586,
      "grad_norm": 0.0002650670357979834,
      "learning_rate": 1.0794107247067807e-06,
      "loss": 0.0,
      "step": 1386
    },
    {
      "epoch": 0.5369725125822686,
      "grad_norm": 0.0031000212766230106,
      "learning_rate": 1.0575719189923838e-06,
      "loss": 0.0,
      "step": 1387
    },
    {
      "epoch": 0.5373596593108788,
      "grad_norm": 0.008421260863542557,
      "learning_rate": 1.0359539385354389e-06,
      "loss": 0.0001,
      "step": 1388
    },
    {
      "epoch": 0.537746806039489,
      "grad_norm": 0.0008919943356886506,
      "learning_rate": 1.0145568808749396e-06,
      "loss": 0.0,
      "step": 1389
    },
    {
      "epoch": 0.538133952768099,
      "grad_norm": 0.047346923500299454,
      "learning_rate": 9.933808425531155e-07,
      "loss": 0.0005,
      "step": 1390
    },
    {
      "epoch": 0.5385210994967092,
      "grad_norm": 0.0025044046342372894,
      "learning_rate": 9.724259191149776e-07,
      "loss": 0.0,
      "step": 1391
    },
    {
      "epoch": 0.5389082462253194,
      "grad_norm": 0.0005212011164985597,
      "learning_rate": 9.516922051078458e-07,
      "loss": 0.0,
      "step": 1392
    },
    {
      "epoch": 0.5392953929539296,
      "grad_norm": 0.0008081488776952028,
      "learning_rate": 9.311797940809775e-07,
      "loss": 0.0,
      "step": 1393
    },
    {
      "epoch": 0.5396825396825397,
      "grad_norm": 0.0008383397944271564,
      "learning_rate": 9.108887785851338e-07,
      "loss": 0.0,
      "step": 1394
    },
    {
      "epoch": 0.5400696864111498,
      "grad_norm": 0.00039093257510103285,
      "learning_rate": 8.908192501721424e-07,
      "loss": 0.0,
      "step": 1395
    },
    {
      "epoch": 0.54045683313976,
      "grad_norm": 0.0004751708183903247,
      "learning_rate": 8.709712993944852e-07,
      "loss": 0.0,
      "step": 1396
    },
    {
      "epoch": 0.5408439798683701,
      "grad_norm": 0.00026852250448428094,
      "learning_rate": 8.513450158049108e-07,
      "loss": 0.0,
      "step": 1397
    },
    {
      "epoch": 0.5412311265969802,
      "grad_norm": 0.0006111898110248148,
      "learning_rate": 8.319404879560233e-07,
      "loss": 0.0,
      "step": 1398
    },
    {
      "epoch": 0.5416182733255904,
      "grad_norm": 0.0026882546953856945,
      "learning_rate": 8.127578033998662e-07,
      "loss": 0.0,
      "step": 1399
    },
    {
      "epoch": 0.5420054200542005,
      "grad_norm": 0.0006612004362978041,
      "learning_rate": 7.93797048687539e-07,
      "loss": 0.0,
      "step": 1400
    },
    {
      "epoch": 0.5423925667828107,
      "grad_norm": 0.015117373317480087,
      "learning_rate": 7.750583093688257e-07,
      "loss": 0.0009,
      "step": 1401
    },
    {
      "epoch": 0.5427797135114208,
      "grad_norm": 0.017428463324904442,
      "learning_rate": 7.565416699917671e-07,
      "loss": 0.0006,
      "step": 1402
    },
    {
      "epoch": 0.543166860240031,
      "grad_norm": 0.0017397633055225015,
      "learning_rate": 7.382472141023223e-07,
      "loss": 0.0,
      "step": 1403
    },
    {
      "epoch": 0.5435540069686411,
      "grad_norm": 0.0027253013104200363,
      "learning_rate": 7.20175024243952e-07,
      "loss": 0.0001,
      "step": 1404
    },
    {
      "epoch": 0.5439411536972513,
      "grad_norm": 0.024540890008211136,
      "learning_rate": 7.023251819572918e-07,
      "loss": 0.0003,
      "step": 1405
    },
    {
      "epoch": 0.5443283004258614,
      "grad_norm": 0.04364755377173424,
      "learning_rate": 6.84697767779735e-07,
      "loss": 0.0009,
      "step": 1406
    },
    {
      "epoch": 0.5447154471544715,
      "grad_norm": 0.08018162101507187,
      "learning_rate": 6.672928612451002e-07,
      "loss": 0.0042,
      "step": 1407
    },
    {
      "epoch": 0.5451025938830817,
      "grad_norm": 0.013882956467568874,
      "learning_rate": 6.501105408832697e-07,
      "loss": 0.0001,
      "step": 1408
    },
    {
      "epoch": 0.5454897406116919,
      "grad_norm": 0.0005265059880912304,
      "learning_rate": 6.331508842198297e-07,
      "loss": 0.0,
      "step": 1409
    },
    {
      "epoch": 0.5458768873403019,
      "grad_norm": 0.06079065427184105,
      "learning_rate": 6.164139677757141e-07,
      "loss": 0.0004,
      "step": 1410
    },
    {
      "epoch": 0.5462640340689121,
      "grad_norm": 0.04179280623793602,
      "learning_rate": 5.998998670668721e-07,
      "loss": 0.0002,
      "step": 1411
    },
    {
      "epoch": 0.5466511807975223,
      "grad_norm": 0.015492278151214123,
      "learning_rate": 5.83608656603929e-07,
      "loss": 0.0002,
      "step": 1412
    },
    {
      "epoch": 0.5470383275261324,
      "grad_norm": 0.0033214192371815443,
      "learning_rate": 5.675404098918258e-07,
      "loss": 0.0,
      "step": 1413
    },
    {
      "epoch": 0.5474254742547425,
      "grad_norm": 0.001019191462546587,
      "learning_rate": 5.516951994295139e-07,
      "loss": 0.0,
      "step": 1414
    },
    {
      "epoch": 0.5478126209833527,
      "grad_norm": 0.00021250147256068885,
      "learning_rate": 5.360730967096272e-07,
      "loss": 0.0,
      "step": 1415
    },
    {
      "epoch": 0.5481997677119629,
      "grad_norm": 0.014712055213749409,
      "learning_rate": 5.206741722181386e-07,
      "loss": 0.0002,
      "step": 1416
    },
    {
      "epoch": 0.548586914440573,
      "grad_norm": 0.0003640944487415254,
      "learning_rate": 5.054984954340481e-07,
      "loss": 0.0,
      "step": 1417
    },
    {
      "epoch": 0.5489740611691831,
      "grad_norm": 0.00032771058613434434,
      "learning_rate": 4.905461348291007e-07,
      "loss": 0.0,
      "step": 1418
    },
    {
      "epoch": 0.5493612078977933,
      "grad_norm": 0.013061568140983582,
      "learning_rate": 4.7581715786743643e-07,
      "loss": 0.0,
      "step": 1419
    },
    {
      "epoch": 0.5497483546264034,
      "grad_norm": 0.010962588712573051,
      "learning_rate": 4.6131163100529563e-07,
      "loss": 0.0001,
      "step": 1420
    },
    {
      "epoch": 0.5501355013550135,
      "grad_norm": 0.003091776045039296,
      "learning_rate": 4.470296196907364e-07,
      "loss": 0.0001,
      "step": 1421
    },
    {
      "epoch": 0.5505226480836237,
      "grad_norm": 0.000399423879571259,
      "learning_rate": 4.329711883633236e-07,
      "loss": 0.0,
      "step": 1422
    },
    {
      "epoch": 0.5509097948122338,
      "grad_norm": 0.0104044359177351,
      "learning_rate": 4.191364004538456e-07,
      "loss": 0.0001,
      "step": 1423
    },
    {
      "epoch": 0.551296941540844,
      "grad_norm": 0.000441914948169142,
      "learning_rate": 4.0552531838402575e-07,
      "loss": 0.0,
      "step": 1424
    },
    {
      "epoch": 0.5516840882694541,
      "grad_norm": 0.10242582112550735,
      "learning_rate": 3.921380035662281e-07,
      "loss": 0.0005,
      "step": 1425
    },
    {
      "epoch": 0.5520712349980643,
      "grad_norm": 0.0005255243158899248,
      "learning_rate": 3.7897451640321323e-07,
      "loss": 0.0,
      "step": 1426
    },
    {
      "epoch": 0.5524583817266744,
      "grad_norm": 0.0023901767563074827,
      "learning_rate": 3.660349162878329e-07,
      "loss": 0.0,
      "step": 1427
    },
    {
      "epoch": 0.5528455284552846,
      "grad_norm": 0.0006369251641444862,
      "learning_rate": 3.533192616027692e-07,
      "loss": 0.0,
      "step": 1428
    },
    {
      "epoch": 0.5532326751838947,
      "grad_norm": 0.0010434030555188656,
      "learning_rate": 3.408276097202845e-07,
      "loss": 0.0,
      "step": 1429
    },
    {
      "epoch": 0.5536198219125048,
      "grad_norm": 0.002758053131401539,
      "learning_rate": 3.285600170019609e-07,
      "loss": 0.0,
      "step": 1430
    },
    {
      "epoch": 0.554006968641115,
      "grad_norm": 0.0003661249065771699,
      "learning_rate": 3.1651653879841703e-07,
      "loss": 0.0,
      "step": 1431
    },
    {
      "epoch": 0.5543941153697252,
      "grad_norm": 0.0017078618984669447,
      "learning_rate": 3.046972294491024e-07,
      "loss": 0.0,
      "step": 1432
    },
    {
      "epoch": 0.5547812620983352,
      "grad_norm": 0.0005711820558644831,
      "learning_rate": 2.9310214228202013e-07,
      "loss": 0.0,
      "step": 1433
    },
    {
      "epoch": 0.5551684088269454,
      "grad_norm": 0.037004098296165466,
      "learning_rate": 2.817313296134938e-07,
      "loss": 0.0002,
      "step": 1434
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 0.0002180523006245494,
      "learning_rate": 2.7058484274794515e-07,
      "loss": 0.0,
      "step": 1435
    },
    {
      "epoch": 0.5559427022841656,
      "grad_norm": 0.0006866626208648086,
      "learning_rate": 2.59662731977639e-07,
      "loss": 0.0,
      "step": 1436
    },
    {
      "epoch": 0.5563298490127758,
      "grad_norm": 0.011221646331250668,
      "learning_rate": 2.489650465824778e-07,
      "loss": 0.0,
      "step": 1437
    },
    {
      "epoch": 0.556716995741386,
      "grad_norm": 0.0015853219665586948,
      "learning_rate": 2.3849183482976824e-07,
      "loss": 0.0,
      "step": 1438
    },
    {
      "epoch": 0.5571041424699962,
      "grad_norm": 0.00038003918598406017,
      "learning_rate": 2.2824314397399406e-07,
      "loss": 0.0,
      "step": 1439
    },
    {
      "epoch": 0.5574912891986062,
      "grad_norm": 0.008126421831548214,
      "learning_rate": 2.182190202566381e-07,
      "loss": 0.0001,
      "step": 1440
    },
    {
      "epoch": 0.5578784359272164,
      "grad_norm": 0.00028225293499417603,
      "learning_rate": 2.0841950890593265e-07,
      "loss": 0.0,
      "step": 1441
    },
    {
      "epoch": 0.5582655826558266,
      "grad_norm": 0.0004284578317310661,
      "learning_rate": 1.9884465413667064e-07,
      "loss": 0.0,
      "step": 1442
    },
    {
      "epoch": 0.5586527293844367,
      "grad_norm": 0.0009348619496449828,
      "learning_rate": 1.8949449915002248e-07,
      "loss": 0.0,
      "step": 1443
    },
    {
      "epoch": 0.5590398761130468,
      "grad_norm": 0.010935988277196884,
      "learning_rate": 1.8036908613331405e-07,
      "loss": 0.0002,
      "step": 1444
    },
    {
      "epoch": 0.559427022841657,
      "grad_norm": 0.0023175193928182125,
      "learning_rate": 1.7146845625985454e-07,
      "loss": 0.0,
      "step": 1445
    },
    {
      "epoch": 0.5598141695702671,
      "grad_norm": 0.005808443762362003,
      "learning_rate": 1.627926496887422e-07,
      "loss": 0.0001,
      "step": 1446
    },
    {
      "epoch": 0.5602013162988773,
      "grad_norm": 0.01145158614963293,
      "learning_rate": 1.5434170556469228e-07,
      "loss": 0.0,
      "step": 1447
    },
    {
      "epoch": 0.5605884630274874,
      "grad_norm": 0.0004185729194432497,
      "learning_rate": 1.4611566201785387e-07,
      "loss": 0.0,
      "step": 1448
    },
    {
      "epoch": 0.5609756097560976,
      "grad_norm": 0.0017726552905514836,
      "learning_rate": 1.38114556163621e-07,
      "loss": 0.0,
      "step": 1449
    },
    {
      "epoch": 0.5613627564847077,
      "grad_norm": 0.00791022926568985,
      "learning_rate": 1.3033842410251075e-07,
      "loss": 0.0001,
      "step": 1450
    },
    {
      "epoch": 0.5617499032133179,
      "grad_norm": 0.027528367936611176,
      "learning_rate": 1.227873009199465e-07,
      "loss": 0.002,
      "step": 1451
    },
    {
      "epoch": 0.562137049941928,
      "grad_norm": 0.0025549372658133507,
      "learning_rate": 1.1546122068614717e-07,
      "loss": 0.0001,
      "step": 1452
    },
    {
      "epoch": 0.5625241966705381,
      "grad_norm": 0.002865710761398077,
      "learning_rate": 1.0836021645593276e-07,
      "loss": 0.0001,
      "step": 1453
    },
    {
      "epoch": 0.5629113433991483,
      "grad_norm": 0.002182627562433481,
      "learning_rate": 1.0148432026860777e-07,
      "loss": 0.0001,
      "step": 1454
    },
    {
      "epoch": 0.5632984901277585,
      "grad_norm": 0.002365889959037304,
      "learning_rate": 9.483356314779479e-08,
      "loss": 0.0,
      "step": 1455
    },
    {
      "epoch": 0.5636856368563685,
      "grad_norm": 0.0056472038850188255,
      "learning_rate": 8.840797510130671e-08,
      "loss": 0.0001,
      "step": 1456
    },
    {
      "epoch": 0.5640727835849787,
      "grad_norm": 0.0037379420828074217,
      "learning_rate": 8.220758512100246e-08,
      "loss": 0.0,
      "step": 1457
    },
    {
      "epoch": 0.5644599303135889,
      "grad_norm": 0.003960594069212675,
      "learning_rate": 7.623242118267038e-08,
      "loss": 0.0001,
      "step": 1458
    },
    {
      "epoch": 0.5648470770421989,
      "grad_norm": 0.0035338555462658405,
      "learning_rate": 7.048251024587837e-08,
      "loss": 0.0001,
      "step": 1459
    },
    {
      "epoch": 0.5652342237708091,
      "grad_norm": 0.0007295788382180035,
      "learning_rate": 6.49578782538851e-08,
      "loss": 0.0,
      "step": 1460
    },
    {
      "epoch": 0.5656213704994193,
      "grad_norm": 0.010685405693948269,
      "learning_rate": 5.965855013347899e-08,
      "loss": 0.0001,
      "step": 1461
    },
    {
      "epoch": 0.5660085172280295,
      "grad_norm": 0.0006623066728934646,
      "learning_rate": 5.458454979492267e-08,
      "loss": 0.0,
      "step": 1462
    },
    {
      "epoch": 0.5663956639566395,
      "grad_norm": 0.0034896975848823786,
      "learning_rate": 4.9735900131786525e-08,
      "loss": 0.0,
      "step": 1463
    },
    {
      "epoch": 0.5667828106852497,
      "grad_norm": 0.0008585017058067024,
      "learning_rate": 4.511262302088759e-08,
      "loss": 0.0,
      "step": 1464
    },
    {
      "epoch": 0.5671699574138599,
      "grad_norm": 0.003946602810174227,
      "learning_rate": 4.0714739322167404e-08,
      "loss": 0.0001,
      "step": 1465
    },
    {
      "epoch": 0.56755710414247,
      "grad_norm": 0.010520562529563904,
      "learning_rate": 3.654226887860879e-08,
      "loss": 0.0001,
      "step": 1466
    },
    {
      "epoch": 0.5679442508710801,
      "grad_norm": 0.00258624367415905,
      "learning_rate": 3.259523051615254e-08,
      "loss": 0.0,
      "step": 1467
    },
    {
      "epoch": 0.5683313975996903,
      "grad_norm": 0.002552057383581996,
      "learning_rate": 2.8873642043586445e-08,
      "loss": 0.0,
      "step": 1468
    },
    {
      "epoch": 0.5687185443283004,
      "grad_norm": 0.0009302432299591601,
      "learning_rate": 2.537752025249529e-08,
      "loss": 0.0,
      "step": 1469
    },
    {
      "epoch": 0.5691056910569106,
      "grad_norm": 0.0006743933190591633,
      "learning_rate": 2.2106880917166506e-08,
      "loss": 0.0,
      "step": 1470
    },
    {
      "epoch": 0.5694928377855207,
      "grad_norm": 0.0005312124267220497,
      "learning_rate": 1.9061738794523553e-08,
      "loss": 0.0,
      "step": 1471
    },
    {
      "epoch": 0.5698799845141309,
      "grad_norm": 0.0035967177245765924,
      "learning_rate": 1.6242107624070412e-08,
      "loss": 0.0,
      "step": 1472
    },
    {
      "epoch": 0.570267131242741,
      "grad_norm": 0.0016068631084635854,
      "learning_rate": 1.3648000127808314e-08,
      "loss": 0.0,
      "step": 1473
    },
    {
      "epoch": 0.5706542779713512,
      "grad_norm": 0.0003538385790307075,
      "learning_rate": 1.1279428010185777e-08,
      "loss": 0.0,
      "step": 1474
    },
    {
      "epoch": 0.5710414246999613,
      "grad_norm": 0.00034932824200950563,
      "learning_rate": 9.13640195805976e-09,
      "loss": 0.0,
      "step": 1475
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.00040175963658839464,
      "learning_rate": 7.2189316406345855e-09,
      "loss": 0.0,
      "step": 1476
    },
    {
      "epoch": 0.5718157181571816,
      "grad_norm": 0.023909885436296463,
      "learning_rate": 5.527025709423095e-09,
      "loss": 0.0001,
      "step": 1477
    },
    {
      "epoch": 0.5722028648857918,
      "grad_norm": 0.0002910869079641998,
      "learning_rate": 4.060691798196681e-09,
      "loss": 0.0,
      "step": 1478
    },
    {
      "epoch": 0.5725900116144018,
      "grad_norm": 0.0007408217643387616,
      "learning_rate": 2.8199365229686357e-09,
      "loss": 0.0,
      "step": 1479
    },
    {
      "epoch": 0.572977158343012,
      "grad_norm": 0.18819063901901245,
      "learning_rate": 1.8047654819441928e-09,
      "loss": 0.0007,
      "step": 1480
    },
    {
      "epoch": 0.5733643050716222,
      "grad_norm": 0.0005039856769144535,
      "learning_rate": 1.0151832555205242e-09,
      "loss": 0.0,
      "step": 1481
    },
    {
      "epoch": 0.5737514518002322,
      "grad_norm": 0.00021944046602584422,
      "learning_rate": 4.5119340624233397e-10,
      "loss": 0.0,
      "step": 1482
    },
    {
      "epoch": 0.5741385985288424,
      "grad_norm": 0.0005209447117522359,
      "learning_rate": 1.1279847879630545e-10,
      "loss": 0.0,
      "step": 1483
    },
    {
      "epoch": 0.5745257452574526,
      "grad_norm": 0.00040242273826152086,
      "learning_rate": 0.0,
      "loss": 0.0,
      "step": 1484
    }
  ],
  "logging_steps": 1,
  "max_steps": 1484,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 371,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.2233368783355904e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}